<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/">
    <channel>
        <title>阿尔法的小破站</title>
        <link>https://yjy.hauchet.cn/</link>
        <description>hauchet</description>
        <lastBuildDate>Tue, 15 Sep 2026 03:35:06 GMT</lastBuildDate>
        <docs>https://validator.w3.org/feed/docs/rss2.html</docs>
        <generator>https://github.com/jpmonette/feed</generator>
        <language>en-US</language>
        <copyright>All rights reserved 2026, 阿尔法</copyright>
        <item>
            <title><![CDATA[一族运动曲线究竟扫过了多大的面积？]]></title>
            <link>https://yjy.hauchet.cn/article/area-swept-by-a-family-of-moving-curves</link>
            <guid>https://yjy.hauchet.cn/article/area-swept-by-a-family-of-moving-curves</guid>
            <pubDate>Tue, 15 Sep 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[从“圆弧族扫过的磁场区域”出发，把问题抽象成一族随参数变化的曲线究竟覆盖平面中的哪些点。文章依次引出曲线族并集、包络线、参数曲线面积公式、二维参数映射与 Jacobian，并解释为什么包络线与 J=0 其实来自同一个几何退化机制；最后讨论重复覆盖时的面积重数问题，并回到上一篇磁场模型统一理解。]]></description>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-3dc9f85982ee815cb410fc3271bdfc27"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><div class="notion-callout notion-blue_background_co notion-block-2207c7cf82f94be6b33145aee03fe503"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="〰️">〰️</span></div><div class="notion-callout-text"><div class="notion-text notion-block-551b890e68044bdabc7338c3e4629833">上一篇里，我真正关心的对象其实已经不再是“某一条粒子轨迹”，而是<b>一整族圆弧在连续变化时究竟扫过了多大的区域</b>。当时我们依靠具体几何把那块磁场区域算了出来；但如果把圆弧换成任意曲线，问题应该怎样统一描述？</div><div class="notion-text notion-block-c4ab2a1d665c43aeb7a5053af585cba2">这篇文章就从那个问题继续向前走：<b>一族运动曲线究竟覆盖了哪些点？它的边界在哪里？面积又该怎样计算？</b></div></div></div><div class="notion-table-of-contents notion-gray notion-block-974a4caac17144979c94adca6017de52"><a href="#71ad3a2bac3b4e7c89582a91e9e1ae61" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">一、从“一条曲线”升级到“一族曲线”</span></a><a href="#bca8aa5eabee4b93a9a366bf6a466473" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">二、先看一个最简单的运动：圆沿直线平移</span></a><a href="#721df0f1fa3b4433b6b56f63422e11a7" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">三、包络线为什么满足  与 ？</span></a><a href="#996306d2e84d48d0a73c90142ebbc6da" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">四、知道边界以后，面积当然可以直接算</span></a><a href="#f84030bffab94794b76676bf79057e4b" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">五、把“曲线上的位置”和“曲线的运动”同时参数化</span></a><a href="#304ef7f643874752b28fac4eff5d5537" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">六、Jacobian：一个小矩形究竟被拉成了多大的平行四边形？</span></a><a href="#138d19ff7b324231894bdcdc4c2181e0" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">七、最漂亮的地方：包络线为什么会和  连在一起？</span></a><a href="#fedc0f812ec74b7bab9e5cce75b3c4a1" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">八、扫掠区域的边界，其实不只来自包络线</span></a><a href="#262da65619f1444ca493361a58e83352" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">九、一个危险的问题：同一个地方如果被扫过两遍呢？</span></a><a href="#b87c7e75d9ee4178936f89cb251b16e9" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">十、回到上一篇的磁场圆弧：它现在变成了一个标准参数映射</span></a><a href="#432554c6f6404bdebf2a1b90dda27872" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">十一、从扫面积继续向上：运动曲面扫出的体积</span></a><a href="#3e8705d119604feb9f4a2b5a7008e409" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">十二、最后把整条逻辑链压缩成一句话</span></a></div><hr class="notion-hr notion-block-ef7cdbfa95694ac4b225d8215611fd7d"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-71ad3a2bac3b4e7c89582a91e9e1ae61" data-id="71ad3a2bac3b4e7c89582a91e9e1ae61"><span><div id="71ad3a2bac3b4e7c89582a91e9e1ae61" class="notion-header-anchor"></div><a class="notion-hash-link" href="#71ad3a2bac3b4e7c89582a91e9e1ae61" title="一、从“一条曲线”升级到“一族曲线”"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">一、从“一条曲线”升级到“一族曲线”</span></span></h3><div class="notion-text notion-block-1acd73d2f89c4305bf30a5488ec07850">在高中阶段，我们习惯研究一条曲线：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d74fef55c4c14b638809010c0775bc6b">但如果曲线本身还会随着参数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 改变，就会得到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-405df556217c4d6dab81fbd02d2ad805">对每一个固定的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，这都是平面上的一条曲线 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。真正值得研究的对象，是它们全部放在一起以后得到的集合：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f17101d4b1ae40209d9c97361014a20c">也就是说，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-81a1d37c64414848951cae3b68f7e0dc">这时问题自然分成三层：</div><ol start="1" class="notion-list notion-list-numbered notion-block-dec2b2aa8bdc4119a100e0a3eeea8752" style="list-style-type:decimal"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 到底是什么区域？</li></ol><ol start="2" class="notion-list notion-list-numbered notion-block-83ae73c2ee8e437c8401f4a116a95d4b" style="list-style-type:decimal"><li>它的边界在哪里？</li></ol><ol start="3" class="notion-list notion-list-numbered notion-block-d9795b148f1c46a99d06bfb98c4fc836" style="list-style-type:decimal"><li>它的面积是多少？</li></ol><div class="notion-text notion-block-9ae6a1308e024c72a55e6cd3082cf963">上一篇的圆弧扫掠磁场，其实就是这个问题的一个具体例子。</div><hr class="notion-hr notion-block-4573720cdae24162b042024ff4f7cd40"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-bca8aa5eabee4b93a9a366bf6a466473" data-id="bca8aa5eabee4b93a9a366bf6a466473"><span><div id="bca8aa5eabee4b93a9a366bf6a466473" class="notion-header-anchor"></div><a class="notion-hash-link" href="#bca8aa5eabee4b93a9a366bf6a466473" title="二、先看一个最简单的运动：圆沿直线平移"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">二、先看一个最简单的运动：圆沿直线平移</span></span></h3><div class="notion-text notion-block-30482466bebd4d3b88d41c727d184d61">考虑一族单位圆</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-b3a2e590dfdf4a21920ec10a869a90fc">当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 增加到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时，单位圆沿 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 轴平移。</div><figure class="notion-asset-wrapper notion-asset-wrapper-image notion-block-0966b65e4ba3452db16638b6cd174e79"><div style="position:relative;display:flex;justify-content:center;align-self:center;width:100%;max-width:100%;flex-direction:column"><img style="object-fit:cover" src="https://file.notion.com/f/f/13703af5-ae9d-433a-8dc7-4bca5f353087/73d00cd5-03c5-4657-83ac-93379c9c5b68/circle-sweep.svg?table=block&amp;id=0966b65e-4ba3-452d-b166-38b6cd174e79&amp;spaceId=13703af5-ae9d-433a-8dc7-4bca5f353087&amp;expirationTimestamp=1789466400000&amp;signature=6JKBATnHszPXtce1QRxwF7n4y1kJhFG7JzEQd-aWRBY&amp;t=0966b65e-4ba3-452d-b166-38b6cd174e79" alt="单位圆平移时形成的扫掠区域；上下两条直线是包络线，左右两端来自参数区间的端点圆。" loading="lazy" decoding="async"/><figcaption class="notion-asset-caption">单位圆平移时形成的扫掠区域；上下两条直线是包络线，左右两端来自参数区间的端点圆。</figcaption></div></figure><div class="notion-text notion-block-d19babc5d45d4218b1e1406f1d3697eb">肉眼很容易看出，它扫出来的是一个“胶囊形”区域。</div><div class="notion-text notion-block-a79932bd84f147658b5eb2ba6cbfacb0">但这里有一个值得注意的事实：</div><blockquote class="notion-quote notion-block-8bae54b18f334de5ac2396e232b7dc98"><div><b>扫掠区域的边界，并不只是 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 和 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 时的两条圆。</b></div></blockquote><div class="notion-text notion-block-f067a79a60dc4363ace7c84d831f0d72">左右两端确实来自最初和最终位置，但上下两条直线并不是任何某个固定 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 对应的整条曲线，而是随着圆不断移动，由相邻圆的位置极限共同“挤”出来的。</div><div class="notion-text notion-block-12e864edf8404079a13d85f83b1a7a5c">这就是包络线最自然的出场方式。</div><hr class="notion-hr notion-block-3f4e93fcbc02448dab0423ee64e538c0"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-721df0f1fa3b4433b6b56f63422e11a7" data-id="721df0f1fa3b4433b6b56f63422e11a7"><span><div id="721df0f1fa3b4433b6b56f63422e11a7" class="notion-header-anchor"></div><a class="notion-hash-link" href="#721df0f1fa3b4433b6b56f63422e11a7" title="三、包络线为什么满足  与 ？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">三、包络线为什么满足 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 与 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>？</span></span></h3><div class="notion-text notion-block-dc0d4a19630946b980b2989e2c488941">设曲线族为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-033020e97c8b4efab956d7e14fb7db53">取两条非常接近的曲线</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a53dfba267e54d71abeb0ae602da28ea">如果它们在某点附近相交，那么两式相减：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-29a4b272b8734ce1bee9914fd04042bc">当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 很小时，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-eb13188a987645cc979ad40e1910746f">于是当</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e22590c990704b98aa2b498b065803aa">相邻曲线交点的极限位置会满足</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-b1527e62b4644596b73fdb56177d1002">这就是求包络线时最常见的方程组。</div><div class="notion-callout notion-teal_background_co notion-block-2ec477f6f73346cc84bc9bfe66f3e2dd"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="💡">💡</span></div><div class="notion-callout-text"><div class="notion-text notion-block-b518721e2c824853aaaba0caf8691334">包络线并不是一个需要死记的“高级公式”。从几何上说，它就是<b>相邻曲线交点的极限轨迹</b>。</div></div></div><div class="notion-text notion-block-186f8658e74f4df3920b8e0f7a80fa67">以刚才的平移圆为例：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-9b596c326c6d4eeca7d3346aceb4d60e">于是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e416635788d1493c9493701cee5cc730">令 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 得</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-03c4acb6d48b4452827a8bfb000e2620">代回圆方程：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c4187d88c3e54b6da124c7e70b009590">所以得到两条包络线</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-eaebfbae0b944ec79ff03aba60cb9269">它们正是胶囊形区域的上下边界。</div><hr class="notion-hr notion-block-9e63c465f0254e41ad9e73d3bce2f69a"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-996306d2e84d48d0a73c90142ebbc6da" data-id="996306d2e84d48d0a73c90142ebbc6da"><span><div id="996306d2e84d48d0a73c90142ebbc6da" class="notion-header-anchor"></div><a class="notion-hash-link" href="#996306d2e84d48d0a73c90142ebbc6da" title="四、知道边界以后，面积当然可以直接算"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">四、知道边界以后，面积当然可以直接算</span></span></h3><div class="notion-text notion-block-a11791dcf84c4da4a2fe34646eb440c3">如果扫掠区域的上下边界能写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-576317e848f949d0bda52788aa0200fd">那么最熟悉的方法就是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ffb6f6b94f7a43b7be0f3a61ef73122a">但对很多曲线族来说，边界本身并不适合写成 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。这时参数曲线形式反而更自然。</div><div class="notion-text notion-block-df4a22c7fdc44498b7b7742c3375e9c8">若闭合边界写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-2bcd2adde76548909d2d8118ad2eb7f7">则由 Green 公式可得</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-6df431355a7e4934b5776ee13ad5b657">也就是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8319bd40bf4e4005a8e49270d879dda4">这一步已经比“上减下积分”更一般，但它仍然有一个前提：<b>我们先得知道边界是谁。</b></div><div class="notion-text notion-block-bfca180f1b084c139b0fbd9d4a943791">能不能干脆不追边界，直接研究整块区域是怎样被生成的？</div><div class="notion-text notion-block-17b2f5f6133d4647acbe3fd559d208ed">答案是可以。</div><hr class="notion-hr notion-block-4d83123612e0498187734f2d9602cdb8"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-f84030bffab94794b76676bf79057e4b" data-id="f84030bffab94794b76676bf79057e4b"><span><div id="f84030bffab94794b76676bf79057e4b" class="notion-header-anchor"></div><a class="notion-hash-link" href="#f84030bffab94794b76676bf79057e4b" title="五、把“曲线上的位置”和“曲线的运动”同时参数化"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">五、把“曲线上的位置”和“曲线的运动”同时参数化</span></span></h3><div class="notion-text notion-block-9c865af00ea84e3d96d2c8c6e21d95dd">设一条曲线本身用参数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 表示，而它的运动由另一个参数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 控制。于是可以写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8194547270ce4cbfab7ae58cf2a35b41">这里：</div><ul class="notion-list notion-list-disc notion-block-31f5ef4023844afe9e1d555dda24efd4"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 决定“当前曲线上的哪一个点”；</li></ul><ul class="notion-list notion-list-disc notion-block-05b8990e030f4caf9cbe8d17d7d7ba4f"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 决定“当前是哪一条曲线”。</li></ul><div class="notion-text notion-block-386950a251c742efbed44133e408f52e">对固定的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，让 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 变化，就得到一条曲线。</div><div class="notion-text notion-block-b121c0b9a8ab4b74bda1cdc4c0cdab18">让 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 同时变化，则二维参数域 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 被映射到平面中的扫掠区域：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-17d6f55eb7cf4ae4a9d21a6a42be4590">这一步是整个问题最重要的升维：</div><div class="notion-callout notion-blue_background_co notion-block-fea04938ee2b44169d460c07cd9fe85f"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="🧭">🧭</span></div><div class="notion-callout-text"><div class="notion-text notion-block-f4fdd00815b14e67bbb51bb8360e99ae">一族运动曲线，本质上就是把“曲线位置参数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> + 运动参数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>”组成的二维参数空间，映射到真实平面。</div></div></div><div class="notion-text notion-block-1b2bc20d48b34c7287b0e16e6cae7310">于是“扫过多少面积”就变成了：参数平面里一个很小的矩形 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，映到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 平面以后，会变成多大的小图形？</div><hr class="notion-hr notion-block-d37d95c3538840b38b2915620e27e1c3"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-304ef7f643874752b28fac4eff5d5537" data-id="304ef7f643874752b28fac4eff5d5537"><span><div id="304ef7f643874752b28fac4eff5d5537" class="notion-header-anchor"></div><a class="notion-hash-link" href="#304ef7f643874752b28fac4eff5d5537" title="六、Jacobian：一个小矩形究竟被拉成了多大的平行四边形？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">六、Jacobian：一个小矩形究竟被拉成了多大的平行四边形？</span></span></h3><div class="notion-text notion-block-25d1433077b64b91b9a36ca8acff5090">在参数点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 附近，沿 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 方向移动 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，平面中的位移近似为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-1f45af0cc4f5497e932f8267c77f5a68">沿 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 方向移动 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，位移近似为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-3367d5294f8a41458922d691d038f2bf">这两个向量围成一个小平行四边形，它的面积为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-6e885c02394f49f0a60ff4de05c7ee41">于是定义</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-0478449b7d5845ff96fc960714311a6b">因此局部面积元满足</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8235dc623fc04a7ca9fc57ded44347e4">如果这个参数映射在几乎所有地方都是一一对应的，那么</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c4c5e20dc2be4f3ebd06df7b0449bcd7">这时我们甚至不需要先把外边界完整求出来，就能直接累计每一小块参数区域对应的真实面积。</div><hr class="notion-hr notion-block-744d5ca2a7c845cc99c486c0ae655a1e"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-138d19ff7b324231894bdcdc4c2181e0" data-id="138d19ff7b324231894bdcdc4c2181e0"><span><div id="138d19ff7b324231894bdcdc4c2181e0" class="notion-header-anchor"></div><a class="notion-hash-link" href="#138d19ff7b324231894bdcdc4c2181e0" title="七、最漂亮的地方：包络线为什么会和  连在一起？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">七、最漂亮的地方：包络线为什么会和 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 连在一起？</span></span></h3><div class="notion-text notion-block-64b298bbc9d84560b4eb7c64aec36a2c">注意</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f14e37dd6f5040e69d0d899695fc940d">若</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ed3183c313524a4b8268f606089df173">就意味着两个向量线性相关：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-5cf8caa728c948b7866909ba1d491860">而这两个向量的几何意义非常直接：</div><ul class="notion-list notion-list-disc notion-block-01c5cae559fc408db43c5edc2c05cbc2"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：沿当前曲线本身的切向方向；</li></ul><ul class="notion-list notion-list-disc notion-block-53413768b6984f2eb245daafb65a8c5b"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：随着曲线继续运动时，这个点的瞬时运动方向。</li></ul><div class="notion-text notion-block-ee6f25217b7f437fa05e748bdd532b42">所以 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 表示：</div><blockquote class="notion-quote notion-block-838a2340cd1e42829fe3335373e15cdb"><div><b>曲线上的这个点，运动方向恰好沿着曲线自己的切线。</b></div></blockquote><div class="notion-text notion-block-0c43dab4af124625a123ba752b9efc2f">此时曲线在局部并没有“横向扫出”新的面积，而是在沿自己的切线滑动。</div><div class="notion-text notion-block-fec45bbd52fd445b8b9b262f92c3bf91">这正是为什么这类点常常会落在扫掠区域的包络边界上。</div><div class="notion-text notion-block-58210878b8fb448cafb63c8a07ae0fd3">也就是说，在合适的正则条件下，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-019d6f453dc145dca108e4b3626749a0">它们不是两个偶然碰巧有关的技巧，而是在描述<b>同一种局部退化现象</b>。</div><div class="notion-callout notion-purple_background_co notion-block-966f87ef722843db97903f9c7a485517"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="✨">✨</span></div><div class="notion-callout-text"><div class="notion-text notion-block-1780d8115acb4f16b4a459810478d04e">前半篇从隐式曲线族得到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；后半篇从参数映射得到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。这两套写法背后其实都在捕捉同一个事实：某些参数位置上，曲线的变化不再向横向撑开面积，于是那里可能形成边界。</div></div></div><hr class="notion-hr notion-block-678a798939134934839007d523f1e6b4"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-fedc0f812ec74b7bab9e5cce75b3c4a1" data-id="fedc0f812ec74b7bab9e5cce75b3c4a1"><span><div id="fedc0f812ec74b7bab9e5cce75b3c4a1" class="notion-header-anchor"></div><a class="notion-hash-link" href="#fedc0f812ec74b7bab9e5cce75b3c4a1" title="八、扫掠区域的边界，其实不只来自包络线"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">八、扫掠区域的边界，其实不只来自包络线</span></span></h3><div class="notion-text notion-block-4f6861d0c57d4f0abbf4f2c7cdf2dea6">这里还要补一个很重要的细节。</div><div class="notion-text notion-block-0156a04a2e5045f9bcb41c8040b9083d">参数映射</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-17268a1d0ca94aa3a2717b62737bd197">的像集边界，通常可能来自两部分：</div><ol start="1" class="notion-list notion-list-numbered notion-block-9b1c4550ad274b7b9fb5c55eeb13df26" style="list-style-type:decimal"><li>参数域本身的边界 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；</li></ol><ol start="2" class="notion-list notion-list-numbered notion-block-1ff65471ac554ac6be4aca8fd35e1788" style="list-style-type:decimal"><li>映射退化的临界点，即 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的地方。</li></ol><div class="notion-text notion-block-bce29e7768de4ec7aeb99c9d98e80604">所以潜在边界通常包含在</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-38d8b46fd4954a638406e0cb013f304a">中。</div><div class="notion-text notion-block-9621d59320f94f849a9ecd8e6c38b69f">刚才“平移圆”的例子就非常典型：</div><ul class="notion-list notion-list-disc notion-block-b4ae6f6079114ec292c52e29a4b81dae"><li>左右两个半圆来自 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 与 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，也就是参数域边界；</li></ul><ul class="notion-list notion-list-disc notion-block-b961c2f3cfcf4e6abd41e6b7343a939c"><li>上下两条直线来自包络条件，也就是退化临界集。</li></ul><div class="notion-text notion-block-9d20d8e29740494a97695a7c63386376">这两类边界拼起来，才构成完整的扫掠区域边界。</div><hr class="notion-hr notion-block-60b39e19c740424ca6f525720b3f29a3"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-262da65619f1444ca493361a58e83352" data-id="262da65619f1444ca493361a58e83352"><span><div id="262da65619f1444ca493361a58e83352" class="notion-header-anchor"></div><a class="notion-hash-link" href="#262da65619f1444ca493361a58e83352" title="九、一个危险的问题：同一个地方如果被扫过两遍呢？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">九、一个危险的问题：同一个地方如果被扫过两遍呢？</span></span></h3><div class="notion-text notion-block-7c1469c5ecd3471fa9ebf3bb8316b279">公式</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-272556b497564e328cb74ad1137bdfab">有一个容易被忽略的前提：不同参数点最好不要大量映到同一个真实点。</div><div class="notion-text notion-block-47b6ee78c6684485aa841a1fade8f2a2">否则，同一块区域会被重复计算。</div><div class="notion-text notion-block-28d53f14084948e69b9fac8425fab8a1">设真实平面上的点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 一共被参数域覆盖了 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 次，那么更准确地说，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8f0135802df34a7aa77e2c9d3b39482e">所以：</div><ul class="notion-list notion-list-disc notion-block-539b1e9da3a744c4974cb5cbd6d7303c"><li>若 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 几乎处处成立，那么 Jacobian 积分就是真实覆盖面积；</li></ul><ul class="notion-list notion-list-disc notion-block-751f5cbb8c994961844efba70b288362"><li>若某些地方被重复扫过，那么它算出来的是<b>带重数的累计扫掠面积</b>。</li></ul><div class="notion-text notion-block-4ac96cdd65004ce3906ac0b349b8e653">这两者不是一回事。</div><div class="notion-text notion-block-ab5825553fbb43d597577183b08e8927">例如某条曲线沿同一轨迹完整运动两遍，真实覆盖区域没有变大，但累计扫掠面积却可能被算两次。</div><div class="notion-text notion-block-7d3513a26be8422d8827a111177a2132">这也是为什么真正严谨的问题不只是“局部面积元是多少”，还要问整个参数映射是不是单射、是否发生自重叠。</div><hr class="notion-hr notion-block-9b739d355022404491c80f4d67453ee4"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-b87c7e75d9ee4178936f89cb251b16e9" data-id="b87c7e75d9ee4178936f89cb251b16e9"><span><div id="b87c7e75d9ee4178936f89cb251b16e9" class="notion-header-anchor"></div><a class="notion-hash-link" href="#b87c7e75d9ee4178936f89cb251b16e9" title="十、回到上一篇的磁场圆弧：它现在变成了一个标准参数映射"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">十、回到上一篇的磁场圆弧：它现在变成了一个标准参数映射</span></span></h3><div class="notion-text notion-block-9f2ed2ce0a6e4bb9a2c47c3147cc9d2f">上一篇里，每个粒子的必要圆弧可以写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-1ec20862167e4c878fba854dd9a0c682">其中</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-26affc44f0894e01a2dcb5fdf890be14">而对每个 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-74e05b1724334b4180769a365d628a88">这里：</div><ul class="notion-list notion-list-disc notion-block-c974fc8e6f6f4fa0a1ddbb607a5750b1"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 决定是哪一个入射粒子；</li></ul><ul class="notion-list notion-list-disc notion-block-6124d7ad5831454587f231462c23c890"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 决定该粒子沿圆轨道转到了哪里。</li></ul><div class="notion-text notion-block-958421eb0ecf4a219d276d970e45aca0">所以上一篇的最小磁场区域</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-84834e2b70e44809b859ff900a10b09a">现在就可以统一写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e28df4d04ec64ac0b47d1bd9b667c081">甚至 Jacobian 也非常简单：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-cb2bba52631a4344b0f9843cc1aded33">因此</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ede546893ff842028b21f12ca9830139">于是退化条件是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-4982aa54a0bf4c1d96c59926dfb70c15">这意味着每条圆弧转到最低点附近时，<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 方向和 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 方向的局部张成面积会退化。</div><div class="notion-text notion-block-352667f2270e4ec1ac5681da119c7f0a">从这一个简单结果，就已经能够重新理解上一篇中扫掠区域为什么会出现特殊的内部/外部边界转换。</div><div class="notion-text notion-block-a6772dd1fd894f9ab6ec6db98f2dc41e">但这里还不能粗暴地直接把</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d0da0cf2992c430986cc630b18c11861">当作上一篇最小磁场面积，因为随着束宽增大，圆弧族会发生重叠；这恰好对应上一篇里“超过临界束宽以后为什么要额外处理”的现象。</div><div class="notion-text notion-block-08d8286c3447482ba6d86878b1800030">也就是说，上一篇看起来像一个特殊的高中物理几何题，实际上已经完整包含了这篇文章里的三层结构：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><hr class="notion-hr notion-block-0a8bbfff8a8e44e7abda3d9114911b97"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-432554c6f6404bdebf2a1b90dda27872" data-id="432554c6f6404bdebf2a1b90dda27872"><span><div id="432554c6f6404bdebf2a1b90dda27872" class="notion-header-anchor"></div><a class="notion-hash-link" href="#432554c6f6404bdebf2a1b90dda27872" title="十一、从扫面积继续向上：运动曲面扫出的体积"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">十一、从扫面积继续向上：运动曲面扫出的体积</span></span></h3><div class="notion-text notion-block-3ac7ef0c8d9948589a4f98944e7b8b3e">一旦接受“参数空间映射”这个观点，问题就很容易继续推广。</div><div class="notion-text notion-block-7ac5e493ab744343a2f0776ff5930941">如果一个曲面本身由 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 两个参数表示，又随时间 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 运动，那么</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-4d33b944a42e43a4b493344af36883e1">就把一个三维参数域映射到三维空间。</div><div class="notion-text notion-block-caf223e210064d619077ce1e5747b964">这时可以继续问：</div><blockquote class="notion-quote notion-block-fd01a136ca3f4b0aa47f9b6c28d16b92"><div>一个运动曲面究竟扫出了多大的体积？</div></blockquote><div class="notion-text notion-block-7ba20ecd9aa543d2af15403f6b6fdfb8">本质仍然是同一个问题，只不过 Jacobian 从二维行列式升级成三维体积伸缩因子。</div><div class="notion-text notion-block-7df6a212c4a14ab7ae79dcd24582a244">所以，从上一篇的高中磁场题继续追问下去，我们真正遇到的并不只是某个面积公式，而是一个更一般的几何主题：</div><div class="notion-callout notion-blue_background_co notion-block-8e4cf3a71ffc4bdead66bca35d2f876b"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="🌌">🌌</span></div><div class="notion-callout-text"><div class="notion-text notion-block-a3ec7783dc4d43edbd708b2870ec608e"><b>当一个低维几何对象连续运动时，它在更高维空间中留下的“痕迹”，应该怎样描述、怎样找边界、又怎样测量？</b></div></div></div><hr class="notion-hr notion-block-3d2a0800d41147cead380287a76b656f"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-3e8705d119604feb9f4a2b5a7008e409" data-id="3e8705d119604feb9f4a2b5a7008e409"><span><div id="3e8705d119604feb9f4a2b5a7008e409" class="notion-header-anchor"></div><a class="notion-hash-link" href="#3e8705d119604feb9f4a2b5a7008e409" title="十二、最后把整条逻辑链压缩成一句话"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">十二、最后把整条逻辑链压缩成一句话</span></span></h3><div class="notion-text notion-block-aa8bdf1e64af40b1a1d75e72fa6f9b02">这篇文章真正想做的，并不是分别介绍“包络线”“参数曲线”“Jacobian”三个知识点，而是把它们放回同一个问题中：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-31008f72866845139c9e305a1fee7463">而其中最值得记住的一点是：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-9f9d9fcddc4044c0b4478426fcaf7e38">从“画很多条曲线”到“研究一个二维映射”，问题的层次也就彻底改变了。</div></main></div>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[一道高中磁场题之后：如果磁场可以是任意形状，最小面积是多少？]]></title>
            <link>https://yjy.hauchet.cn/article/minimum-magnetic-field-area-beyond-rectangle</link>
            <guid>https://yjy.hauchet.cn/article/minimum-magnetic-field-area-beyond-rectangle</guid>
            <pubDate>Mon, 14 Sep 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[从一道已经找不到原题的高中带电粒子磁场题出发，用 GeoGebra 中的实际构造重新建模：O 不是所有粒子必须经过的公共点，而是粒子离开磁场后再次到达 x 轴时不能越过的最远极限点。由此把任意形状磁场化为每个粒子“最早允许离场”之前的必要圆弧并集，推导其最小面积，并解释临界束宽之后粒子为什么会被迫多转一段、却仍然满足极限条件。]]></description>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-3db9f85982ee81daae99d09902bd0ca3"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><div class="notion-callout notion-blue_background_co notion-block-778a4dd290154916bd370afc7a36e19f"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="🧲">🧲</span></div><div class="notion-callout-text"><div class="notion-text notion-block-62219d1568904944a4c8fe9e54fc053c"><b>这篇文章的起点，是一道我已经找不到原题的高中磁场题。</b></div><div class="notion-text notion-block-76184cad2fb3401da4c104fa1febd8fe">我还记得它最核心的结构：一列彼此平行的带电粒子进入匀强磁场，在磁场中沿圆弧偏转，离开磁场以后沿切线继续运动；粒子再次到达 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 轴时，<b>不能越过一个给定的最远极限点</b>。原题要求的是矩形磁场；真正让我一直记到现在的问题却是：<b>如果把“矩形”这个人为限制去掉，真正需要的磁场区域应该是什么形状，它的面积又是多少？</b></div><div class="notion-text notion-block-2be424c8417042e6833e9e97e0c32335">原题已经无法找回，所以本文不会假装复述原题原文。下面完全按照我后来在 GeoGebra 中重新搭出的几何构造来研究，而且把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 都保留为参数，不把滑块当前的临时数值当成题设。</div></div></div><div class="notion-table-of-contents notion-gray notion-block-0a59781718804bf895f8f7232e727222"><a href="#3e43ae874e134896a9d74fe987ea7bcd" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">一、先把 GeoGebra 里的真实模型写成数学</span></a><a href="#6c72a8212b2f4063b03d399ee0b54b4f" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">二、单个粒子的切点在哪里？</span></a><a href="#6fcf648120e04da3a1be6ee4e8abd2f0" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">三、真正想要的区域：让圆弧  扫过去</span></a><a href="#9e19101c04cd482ba5f999f9dea56d24" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">四、为什么圆弧并集本身就是最小静态磁场？</span></a><a href="#28791b307f494505a7aca40f4b027add" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">五、这个扫掠区域的边界究竟长什么样？</span></a><a href="#c5baf8089e10453387038f05f563dc83" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">六、面积怎么求？关键是横着切</span></a><a href="#b5d83fd63c6b4cc388e90992ce705fe5" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">七、第一阶段：所有粒子都恰好在极限切点离场</span></a><a href="#b1073e4c0d3d487da2a4ba389cf1a71c" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">八、超过临界束宽以后会发生什么？</span></a><a href="#4c710a72b85446388da7ee36a4e2ebfc" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">九、和“矩形磁场”比较</span></a><a href="#c7763129326e4ee498baefe54d954f26" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">十、用一组便于画图的数值看看量级</span></a><a href="#faec07c83f324dde97ae358dc91ee3d6" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">十一、这道题最后变成了一个更一般的问题</span></a></div><hr class="notion-hr notion-block-9894c6f597ad41249745f53e13680b74"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-3e43ae874e134896a9d74fe987ea7bcd" data-id="3e43ae874e134896a9d74fe987ea7bcd"><span><div id="3e43ae874e134896a9d74fe987ea7bcd" class="notion-header-anchor"></div><a class="notion-hash-link" href="#3e43ae874e134896a9d74fe987ea7bcd" title="一、先把 GeoGebra 里的真实模型写成数学"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">一、先把 GeoGebra 里的真实模型写成数学</span></span></h3><div class="notion-text notion-block-523cd543774c45a7af3ab843afe4a462">取粒子离开磁场后再次到达 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 轴时<b>不能越过的最远极限点</b></div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-87f3024ab807460a9d655018083f78eb">也就是说，粒子最终与 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 轴重新相交于某点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时，只要求</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-121bd4749ce945bcb7670b39a70bbc8e">并不要求所有粒子都必须经过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。</div><div class="notion-text notion-block-abd9e2939a134238822ccecf08315397">粒子从上方向下、彼此平行地穿过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 轴上的一段连续区间入射：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-680eaa06e4934b7a85f5e98a65a67073">其中某一个粒子的入射点记作</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e806a87c2f1949938fe2086fc3e23e1e">磁场使所有粒子做半径相同的圆周运动，半径记作 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。按照 GeoGebra 中的构造，圆心在入射点左侧 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e07504be213b4aaf84338f8d2157e495">因此粒子的圆轨道为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-9cc349e96e714f028392abdfb4b9d707">从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 向这条圆作下方切线，切点记作 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。这条切线代表的是一种<b>极限情形</b>：如果粒子恰好在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 离开磁场，它随后会沿切线经过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；如果它比 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 更晚离开，则最终只会落在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的右侧，同样满足题设。</div><div class="notion-text notion-block-37b39710e8a945afa0c81c229b46f428">为了使 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 位于每个圆外，必须有</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-2197667612054540a20cc8f070fac1a8">所以整束粒子至少要满足</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-86727a58597a4eb7be990a1b66d3ff61">这也是后面所有公式的基本条件。</div><hr class="notion-hr notion-block-8dd8e98bd0744f6f93c5960af5a89201"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-6c72a8212b2f4063b03d399ee0b54b4f" data-id="6c72a8212b2f4063b03d399ee0b54b4f"><span><div id="6c72a8212b2f4063b03d399ee0b54b4f" class="notion-header-anchor"></div><a class="notion-hash-link" href="#6c72a8212b2f4063b03d399ee0b54b4f" title="二、单个粒子的切点在哪里？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">二、单个粒子的切点在哪里？</span></span></h3><div class="notion-text notion-block-c7fbcb2da3404952ba9d87a2b3b61981">为了少写一些重复项，令</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c8f2235030d94d86a3941bfc23ddc3f5">此时</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-acacb9e7b98245e9b406c2d23d922499">由于 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 是圆在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 点的切线，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-dc1e7542bddd4ae0b1264dc94105bd20">所以直角三角形 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 给出</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-13af7f42675c47f98b6ae301d9dcbcb7">进一步可以得到切点坐标</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-96926f15542142d8882f40d6c652b92a">也就是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-cf7eeea7e7b24aeeb91502539928d1f2">这和 GeoGebra 里“从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 向圆作切线，再取切点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>”的构造完全一致。</div><div class="notion-text notion-block-36481f61aedb49738627a30e03c68126">如果以圆心 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 为基准，把粒子从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 向下转过的圆心角记作 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，那么圆弧可以参数化成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ce8a45e2dd054d2a82b8f6a651676954">起点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 对应 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。终点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 对应的角度 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 满足</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-39e86cb1390f4f3b9402cbd9c2b8e763">所以</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c9206f0e0b844a928196d19c44b0902c">这也解释了为什么 GeoGebra 里某些圆弧会明显超过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：那不是画错了，而是这个模型本来就会出现钝角偏转。</div><div class="notion-text notion-block-c9d7994262834ef0ae506e3a91c05a84">更重要的是，<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的物理意义现在也清楚了。若粒子在某个 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 处离开磁场，沿该点切线继续运动，并在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 轴上重新落到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，那么切线的横截距为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-4aeb3ab874614291917b91925cadcd23">在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a79873f9dceb4223b657f560db52d537">于是恰好有</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-b8ca869f36e14ccbb06e343aa8811181">而在</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-b4f09fabd10a49628b3eb0b4d51e621a">时，<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，粒子会越过极限点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；只有</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-094dc8bb2b644c5095caaf1a280a68fc">才满足 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。</div><div class="notion-text notion-block-6945ae89172048e683a165f12a31d4c2">所以 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> <b>不是粒子必须离开的点，而是最早允许离开的点</b>。从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的整段圆弧都是不可缺少的。</div><div class="notion-callout notion-teal_background_co notion-block-3a60dca120f54f8fb4bc2f8f2397f8ea"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="✅">✅</span></div><div class="notion-callout-text"><div class="notion-text notion-block-a3436ba206814769ad47ada0be1af828">这个解释还有一个很重要的自洽性检查：如果误把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 记成“所有粒子都必须经过的公共点”，那么连续粒子束就要求每个粒子都恰好在各自不同的极限切点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 离场，而这些 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 连成的是一条曲线，不可能由一个普通矩形边界同时实现。把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 恢复为“不能越过的最远极限点”后，原题存在矩形磁场解就重新变得合理了。</div></div></div><hr class="notion-hr notion-block-7f2536b42fcb4506a0bcfc307fefbcfa"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-6fcf648120e04da3a1be6ee4e8abd2f0" data-id="6fcf648120e04da3a1be6ee4e8abd2f0"><span><div id="6fcf648120e04da3a1be6ee4e8abd2f0" class="notion-header-anchor"></div><a class="notion-hash-link" href="#6fcf648120e04da3a1be6ee4e8abd2f0" title="三、真正想要的区域：让圆弧  扫过去"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">三、真正想要的区域：让圆弧 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 扫过去</span></span></h3><div class="notion-text notion-block-2eb8803c428f442ca3c375d77c315855">我在 GeoGebra 中真正想看的对象并不是某一条圆弧，而是</div><blockquote class="notion-quote notion-block-a993e25fbf6e422eb824ff4fb27c4269"><div><b>当 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 从 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 连续移动到 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 时，圆弧 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 扫过的全部区域。</b></div></blockquote><div class="notion-text notion-block-bc95611372b94bbb85a129582b2a27d8">用数学语言，这个扫掠集合就是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-786a4ae4db1d42ecac897cba545adc46">代入上面的参数式，可以直接写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ff57acd2b645419aa60090e5bdb77bae">这才是“取消矩形限制”以后最自然的候选磁场区域：每个粒子需要受磁场偏转的那一段轨迹全部包含进去，而没有粒子经过的地方就没有必要铺磁场。</div><figure class="notion-asset-wrapper notion-asset-wrapper-image notion-block-6f08b45c53504941b3e69e59f48f8e83"><div style="position:relative;display:flex;justify-content:center;align-self:center;width:100%;max-width:100%;flex-direction:column"><img style="object-fit:cover" src="https://raw.githubusercontent.com/sincostancscseccot/NotionNext/main/public/images/posts/minimum-magnetic-field/sweep-diagram.svg?spaceId=13703af5-ae9d-433a-8dc7-4bca5f353087&amp;t=6f08b45c-5350-4941-b3e6-9e59f48f8e83" alt="圆弧族 e 随 C 从 A 移到 B 时扫出的磁场区域 Ω；示意图取 a=5、b=6、r=1.8，但正文推导中的 a、b、r 均保留为参数。虚线表示部分粒子离开磁场后通向 O 的切线。" loading="lazy" decoding="async"/><figcaption class="notion-asset-caption">圆弧族 e 随 C 从 A 移到 B 时扫出的磁场区域 Ω；示意图取 a=5、b=6、r=1.8，但正文推导中的 a、b、r 均保留为参数。虚线表示部分粒子离开磁场后通向 O 的切线。</figcaption></div></figure><div class="notion-callout notion-blue_background_co notion-block-b843bd099c5847539c01e51ce2335ec7"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="📐">📐</span></div><div class="notion-callout-text"><div class="notion-text notion-block-6050d0d3e07c45d4b165de4942eb0239">图中蓝色区域不是凭视觉猜出来的“扇形”，而是严格按 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 计算得到的扫掠集合。示意参数采用 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，与 GeoGebra 当前文件一致；它们只负责让图形直观，后面的公式仍对一般 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 推导。</div></div></div><hr class="notion-hr notion-block-c6e9db1091a8406380c60cb81aa85aa2"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-9e19101c04cd482ba5f999f9dea56d24" data-id="9e19101c04cd482ba5f999f9dea56d24"><span><div id="9e19101c04cd482ba5f999f9dea56d24" class="notion-header-anchor"></div><a class="notion-hash-link" href="#9e19101c04cd482ba5f999f9dea56d24" title="四、为什么圆弧并集本身就是最小静态磁场？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">四、为什么圆弧并集本身就是最小静态磁场？</span></span></h3><div class="notion-text notion-block-e8532e5d185f4f7ca0384e2979c5ca36">现在这个问题反而比“所有粒子都必须经过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>”时更自然。</div><div class="notion-text notion-block-a8a4f5187e534792aabc25595f7e63fb">对任意一个粒子而言，如果磁场在它到达 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 之前就结束，那么它会提前离场，并落到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 左侧；所以从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的整条圆弧都是<b>必须存在磁场</b>的地方。于是任何满足题意的静态磁场都必须包含</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c0efcc118f704d2eb35022a7fdbdc207">这说明 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 是必要的。</div><div class="notion-text notion-block-f551b9e7d31e46948e1a89e09a32fb06">反过来，直接把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 作为磁场区域也足够：某个粒子到达自己的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时，如果那里仍被别的必要圆弧覆盖，它只会继续多转一段；而更晚离场只会使最终横截距从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 增大到正值，因此仍满足“不能越过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>”的要求。</div><div class="notion-text notion-block-9f904ee15cd142198a31f519ca7b2f6c">所以</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-5fa95171b1d248369709c7514322b7c6"><b>圆弧族 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 的扫掠区域本身就是真正的最小静态磁场区域。</b></div><div class="notion-text notion-block-6174a0d4ea1c485fb2ffd6127effad94">不过，随着束宽增大，扫掠区域会出现一个很重要的几何转折。</div><div class="notion-text notion-block-3fefd629dbd74ac18854ce54c534d882">令</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-3976aa750dc141a988c213389f07e55f">最右端粒子的切点为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-7e543fb9a3774577857194cb24f06bc2">在同样的高度上，最左端粒子圆轨道的右半支横坐标为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-348df18232d7401e87970e6f41e6023b">当最右端粒子的极限切点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 第一次被前面那些必要圆弧覆盖时，临界状态满足</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-1e98deab497141479603968cabad56b8">整理得到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-71217870940143238eba16017f1a83eb">于是出现一个很自然的临界值</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-fa45fbac7d684ae08bff75cba142bcf8">这个 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 不再是“静态磁场是否可实现”的边界，而是<b>所有粒子是否都能恰好在自己的最早允许离场点 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 离开</b>的分界：</div><ul class="notion-list notion-list-disc notion-block-0f287e1a876e4a2fa3868728001f60fa"><li>当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时，每个粒子都在自己的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 离场，因此都会恰好经过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；</li></ul><ul class="notion-list notion-list-disc notion-block-07a35ec94f4e427a8eea99d46ff8cd51"><li>当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时，一部分粒子的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 已经落入其他必要圆弧扫出的磁场内部，它们会继续转动一段，最终落在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的右侧。</li></ul><div class="notion-callout notion-yellow_background_co notion-block-6913e46815464b12b80cada949806cb7"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="💡">💡</span></div><div class="notion-callout-text"><div class="notion-text notion-block-657b4751ec76477390bc2177261a544f">我在 GeoGebra 里临时用过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。这组数只是滑块的临时取值，却恰好跨过了这个临界点：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-aba82eef8317469c98cc727f3a80a2a3">所以 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 并不是“物理上非法”的参数。它只是已经越过第一个临界点：右侧一部分粒子不能在自己的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 立即离场，而会继续在最小磁场区域中多转一段。</div></div></div><details class="notion-toggle notion-block-ddbb892719f34f7b902fc1e523a22c03"><summary>更正记录：把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 误当成共同汇聚点时的旧推导</summary><div><div class="notion-text notion-block-b42156d861154703819bf2bf69049db3"><b>这一折叠段建立在“粒子必须在 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 离场并经过 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b>”的旧假设上，因此不再作为本文结论。</b> 当时为了检查所谓“离场后是否重新进入磁场”，曾设某条粒子的圆心横坐标为 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，并在旧假设下取切线上任意一个高度 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，令</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-4896e1863bf84344bc897c35d365fdbc">沿切线 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的横坐标可以写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-48da458a398a47f5848533d3daa01ead">而同一高度上，扫掠区域左侧那一小支的最右边界为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-2562613cbce1468d815f966ee7aba648">主区域的最左边界则为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a903534d97824bbfac3d51cc2e5d53df">离开切点以后有 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，于是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-94794df25ee54ca5801e95f7f2a2635c">另一方面，在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 且 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时又有</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-52fe4b03f82944c8936aa7abd786c393">所以离场后的直线正好穿过两块扫掠区域之间的“空隙”，而不是重新进入磁场。</div></div></details><hr class="notion-hr notion-block-5e9a56ad510b493aacdb4ff72382d79d"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-28791b307f494505a7aca40f4b027add" data-id="28791b307f494505a7aca40f4b027add"><span><div id="28791b307f494505a7aca40f4b027add" class="notion-header-anchor"></div><a class="notion-hash-link" href="#28791b307f494505a7aca40f4b027add" title="五、这个扫掠区域的边界究竟长什么样？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">五、这个扫掠区域的边界究竟长什么样？</span></span></h3><div class="notion-text notion-block-032eec1924294e60b60b1368f283af8a">现在把参数</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f2d41b17a5f04cd396c59e63b88d64ff">直接作为横向平移参数。圆弧族可以写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e5189b3709f84161880c18da30d11b1f">其中</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d51619fa5f654fc4b895354890aa76b3">它的 Jacobian 为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d9dec9acbcd6486ca93b942d0e2b2417">因此在</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-9287b832c6664452931fc6ec5d4b0fda">处出现折叠。所有粒子的最低点刚好连成一条水平线段：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-103261883a274d75976b83a563752e4d">这条线不是凭图形猜出来的，而是参数映射本身 Jacobian 变号产生的“折线边界”。</div><div class="notion-text notion-block-217f722b4f4840788b3a00d4b902db7e">另一方面，把所有切点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的参数消掉也能得到一条很漂亮的通用曲线。由</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-43e5b2a0d75d461a91964cf5c5213797">可得</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8186130b4b6e45068410ef716f37bdf5">代回 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，得到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-6f167144d2e14d01b296ac12a3c9200c">也就是说，<b>所有粒子的切点轨迹本身与 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 无关；</b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 只决定我们截取这条通用曲线的哪一段。</b></div><div class="notion-text notion-block-d874bbe19760457b83372dae27a27fb6">在第一阶段 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 下，整个外边界由以下几段拼起来：</div><ul class="notion-list notion-list-disc notion-block-ed4145683a80415d83649052d061e64a"><li>入射线段 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；</li></ul><ul class="notion-list notion-list-disc notion-block-d451181a4e8145ea9fad0f5fb4e3d750"><li>最右端粒子圆弧从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 到最低点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的一段；</li></ul><ul class="notion-list notion-list-disc notion-block-0725ff19228e4711be10552523b0320f"><li>最低点线段 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；</li></ul><ul class="notion-list notion-list-disc notion-block-f5140a0e709e483d89e874cb81fd45b3"><li>最左端粒子圆弧从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的一段；</li></ul><ul class="notion-list notion-list-disc notion-block-d5cbfabcfde34086a4ae9c3a49f68211"><li>切点轨迹从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；</li></ul><ul class="notion-list notion-list-disc notion-block-09071c99e214493aa9aefad0ef83327f"><li>最右端粒子圆弧从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 到两端圆轨道的交点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；</li></ul><ul class="notion-list notion-list-disc notion-block-4622413c26fb4651b9ef67a9fb279791"><li>最左端粒子圆弧从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 回到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。</li></ul><div class="notion-text notion-block-ae10620d53ed452cbb1cbaa6551c5207">两端圆轨道半径相同，圆心距离为 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，所以它们的下方交点为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d99e9a9d5f094298ae2700030bd7ba53">这就是 GeoGebra 里应该看到的真实边界结构：它不是扇形，也不是简单的“某一条圆弧加一条直线”。</div><hr class="notion-hr notion-block-6ba24a8d79234e34a587283cc8f147e7"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-c5baf8089e10453387038f05f563dc83" data-id="c5baf8089e10453387038f05f563dc83"><span><div id="c5baf8089e10453387038f05f563dc83" class="notion-header-anchor"></div><a class="notion-hash-link" href="#c5baf8089e10453387038f05f563dc83" title="六、面积怎么求？关键是横着切"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">六、面积怎么求？关键是横着切</span></span></h3><div class="notion-text notion-block-64117252eed0445c8fc93a0dd76542c3">直接沿复杂外边界做 Green 公式当然可以，但这里有一个更直观的方法：固定一个高度 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，看这个高度上被多少横向区间覆盖。</div><div class="notion-text notion-block-051c97b477394cec8e7deb9d39821e7d">令</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c921226272264b2bb4920e95ca76eeeb">在高度 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 上，每个半径为 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的圆都有两个候选横坐标</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-5e139c17049f4f1586ff336592a7ce24">其中右半支 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 对所有粒子都存在，于是扫出</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e6dc5f7372e740a39b14d74225753257">左半支只有在粒子还没有提前到达切点时才存在。条件等价于</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-b3558906209f49aa847c795ed7649091">因此左半支扫出的区间是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-346915f531754e6695c608a764b52dfb">这两个区间的并集，就是该高度上的真实磁场截面。</div><div class="notion-text notion-block-671ad6a063c940c1b48cb3f31293de85">再定义</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-4710eebd6b1745cbbf9a0e5289c8b4da">以及三个关键量</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-0586e75fd4154db58851dbad9bd0c46c">在第一阶段 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 下，它们满足</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d72aaeae83544e5caf53c60b82fe0139">于是横截面总长度 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 可以完整写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a4e57e499fde4374b052c39d0e6c25de">因为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8f82915c58f94e02b9529bf5da1ff44d">所以</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-3bd37f612cd4446aa9690aff4fc35508">于是面积就是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-43457cf92afc4e6fb86014f2b222be23">这一步把“圆弧族扫过的复杂区域”彻底变成了普通的一元积分。</div><hr class="notion-hr notion-block-ec425939585e4060b9b2714955a05962"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-b5d83fd63c6b4cc388e90992ce705fe5" data-id="b5d83fd63c6b4cc388e90992ce705fe5"><span><div id="b5d83fd63c6b4cc388e90992ce705fe5" class="notion-header-anchor"></div><a class="notion-hash-link" href="#b5d83fd63c6b4cc388e90992ce705fe5" title="七、第一阶段：所有粒子都恰好在极限切点离场"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">七、第一阶段：所有粒子都恰好在极限切点离场</span></span></h3><div class="notion-text notion-block-98e37313b5ac470986404e11cd90d6a7">把上面的四段积分分别算完并整理，可以得到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-6eb66a549aa343438f4783ed667d5892">它的适用条件是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-dc5ab850cac74c39831108cedf1d89fd">这个公式描述的是 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的第一阶段。此时每个粒子的极限切点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 都还位于扫掠区域的外边界上，所以粒子恰好在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 离场，并全部经过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。这里的上界不是“静态磁场能否实现”的限制，而只是下一种几何行为开始出现的位置。</div><hr class="notion-hr notion-block-51bff291bea046e493cf6ecbb2ccf8f3"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-b1073e4c0d3d487da2a4ba389cf1a71c" data-id="b1073e4c0d3d487da2a4ba389cf1a71c"><span><div id="b1073e4c0d3d487da2a4ba389cf1a71c" class="notion-header-anchor"></div><a class="notion-hash-link" href="#b1073e4c0d3d487da2a4ba389cf1a71c" title="八、超过临界束宽以后会发生什么？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">八、超过临界束宽以后会发生什么？</span></span></h3><div class="notion-text notion-block-cece403d35e947f18affa2c0ae704041">如果</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-43b8a5d356c5486db5602dddfcc5a393">这时圆弧族当然仍然继续扫，而且这个扫掠区域<b>依旧是真正可实现的最小静态磁场</b>。变化的只是部分粒子的实际离场位置：它们到达自己的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 后发现那里仍在磁场内部，于是被迫继续转动。</div><div class="notion-text notion-block-611a36653c7b4b0babb141c47aea670e">有意思的是，一旦超过临界点，左侧那一块复杂结构就不再继续长大；之后每把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 向右增加一点，只会多出一条高度为 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的横向带状区域。因此</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-b693e0ed9f724e218b222185fb0dd785">设某个粒子的入射参数为 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，并在它进入左半圆后记</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c744d487c0cf4139a186c92adf00cd96">它自己的“最早允许离场点”对应</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f6e734460acf44f28604d57b366ae992">但同一位置还可能被更左侧粒子的右半圆弧覆盖。覆盖能够把它继续推到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-da6d7a2b12eb4f1f845d462a69982b6a">所以粒子真正离开最小磁场时有</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-79d972082802406985d32c79227c2e6b">于是实际落到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 轴上的位置为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-cc58cad66dc548d785adf43816d8c640">这给出了两个清楚的几何阶段变化：</div><ul class="notion-list notion-list-disc notion-block-ddda4082977c49658efd518a510a45fc"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：粒子恰好在自己的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 离场，所以经过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；</li></ul><ul class="notion-list notion-list-disc notion-block-b245859d4a09468fa58addfbd1953e87"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：粒子被其他圆弧继续“托住”一段，离场后落在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 右侧；</li></ul><ul class="notion-list notion-list-disc notion-block-dbabe7fe29d742cf9b951572c769bd32"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：覆盖一直持续到半圆终点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，粒子转过整整半圈后在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 处离开。</li></ul><div class="notion-text notion-block-d8606c9f600744b697e48e213b741569">因此 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的真正意义是：<b>从这里开始，“O 是共同经过点”这一特殊状态结束；但 O 始终保持“最远极限点”的身份。</b></div><hr class="notion-hr notion-block-14de00bf0a1848fab488a63e878bcf2e"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-4c710a72b85446388da7ee36a4e2ebfc" data-id="4c710a72b85446388da7ee36a4e2ebfc"><span><div id="4c710a72b85446388da7ee36a4e2ebfc" class="notion-header-anchor"></div><a class="notion-hash-link" href="#4c710a72b85446388da7ee36a4e2ebfc" title="九、和“矩形磁场”比较"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">九、和“矩形磁场”比较</span></span></h3><div class="notion-text notion-block-44759c7904454ec6828aa0ba6f4a4f1b">因为原题已经找不到，我不再声称下面这个矩形就是原题唯一的原始设定；但在当前重构模型里，如果仍要求磁场是与坐标轴平行的矩形，并且要完整包住同一批圆弧，那么它的边界很容易确定。</div><div class="notion-text notion-block-a4d0c30ff7a34999844bb87f9b07ae49">整个扫掠区域的最高点是 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，最低点是 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，所以矩形高度至少为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-4a552ae752274ef28bc4dd45a6eef50d">最右端是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-dd4f6b92a4664ff1964e47ce35205125">最左端则来自最左粒子的切点 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-bbfe9a110e714616b13522f31e3be178">因此最小轴对齐矩形面积为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-809fd5c0235f46fa8912d187df761258">这也把“矩形为什么会浪费很多面积”说得非常直观：它不得不把圆弧之间、切点轨迹旁边和离场空隙中的大片无用区域一起包进去。</div><hr class="notion-hr notion-block-aa51cef055fc46f5a6b2d7b55c7588c9"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-c7763129326e4ee498baefe54d954f26" data-id="c7763129326e4ee498baefe54d954f26"><span><div id="c7763129326e4ee498baefe54d954f26" class="notion-header-anchor"></div><a class="notion-hash-link" href="#c7763129326e4ee498baefe54d954f26" title="十、用一组便于画图的数值看看量级"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">十、用一组便于画图的数值看看量级</span></span></h3><div class="notion-text notion-block-fe1ce881236641b896e12a7d3e290621">为了让 GeoGebra 里的图形结构清楚，而且先处在“所有粒子都恰好经过 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>”的第一阶段，可以暂时取</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c3ebebfe481d4e7e963c0843075d8347">此时</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-bacbaf47a3ed410caf52aab35b413e7e">所以 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，这组示意参数还没有进入“切点被其他圆弧覆盖”的第二阶段。</div><div class="notion-text notion-block-02efc4dccdc94f199b0b3024c311c506">代入面积公式：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-edc79bf55c584901bf18698462a33673">而对应的轴对齐矩形面积为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f79263362c89468bba795e2a63aa778f">两者之比约为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-b6511359ed0b4426b9768fde6e876bee">也就是说，在这一组示意参数下，真正沿轨迹“贴着铺”的磁场区域只需要矩形包围面积的大约三成。</div><div class="notion-text notion-block-75744406ca1245b8931879a64296ab26">把 GeoGebra 的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 拉到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 也依然是物理上自洽的；它只是会进入第二阶段，更适合展示“粒子到达极限切点后仍被磁场继续偏转”的现象。正文配图仍采用 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，只是因为第一阶段的几何边界更容易一眼看清。</div><hr class="notion-hr notion-block-ba615f1dda1a40c28cd57cc215871679"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-faec07c83f324dde97ae358dc91ee3d6" data-id="faec07c83f324dde97ae358dc91ee3d6"><span><div id="faec07c83f324dde97ae358dc91ee3d6" class="notion-header-anchor"></div><a class="notion-hash-link" href="#faec07c83f324dde97ae358dc91ee3d6" title="十一、这道题最后变成了一个更一般的问题"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">十一、这道题最后变成了一个更一般的问题</span></span></h3><div class="notion-text notion-block-58e21ccd362240d3b4e569d2ec325ede">高中时我只是觉得：</div><blockquote class="notion-quote notion-block-8807fabade1a498b97443504258ced7a"><div><b>既然矩形只是题目人为加上的限制，那把它拿掉以后，最优区域到底是什么？</b></div></blockquote><div class="notion-text notion-block-27b9f1f9085f438faa56a5e6044e3573">真正算下去以后，它自然变成了一个一般的“曲线族扫掠面积”问题。</div><div class="notion-text notion-block-78cad700988b4ba990d1fe25ee1d286f">如果一族曲线可以写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ab7ddfbb521348d1ac55e67bd2c2ba7a">那么至少要问三件事：</div><ul class="notion-list notion-list-disc notion-block-53c1470418514cefbf99c466026dd5d6"><li>参数区域的边界映到平面后形成哪些候选边界；</li></ul><ul class="notion-list notion-list-disc notion-block-87d7173a0dfc4603a4035e11ded98438"><li>Jacobian</li></ul><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-34327dfe1ba740179ee66365af564772">在哪里变成零，从而产生折叠、包络或重叠；</div><ul class="notion-list notion-list-disc notion-block-76a670b4f1bd4647afce124e33007817"><li>不同参数对应的曲线是否会覆盖同一片区域，从而使“积分参数面积”不能直接当成几何并集面积。</li></ul><div class="notion-text notion-block-a7a08ea9f92a48a4b6ad0f9cd63d3728">这个磁场问题里，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-35f1ee73a02146a9ab164fa42ced0afd">的 Jacobian 恰好在</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c811c87c83b94aaaa6cbd9697fcb3429">变号，所以最低点线段自然成为扫掠边界；而切点条件又截掉了每条圆的后半部分，最终才形成现在这个看起来有些奇怪、却完全可以严格计算的区域。</div><div class="notion-text notion-block-5ea937831b8f474b838ac6ae66ee76ce">这也回到了我后来一直在想的那个问题：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-40e99ed66b5e44548f3135ce54e43a87">表示一族随参数变化的曲线时，我们究竟怎样计算它们真正划过的面积？</div><div class="notion-text notion-block-b291d5bd3cf245b588654f549a912a76">对我来说，这道已经找不到原题的高中物理题，真正有意思的部分反而是在“标准答案结束以后”才开始的。</div></main></div>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[从一串不等式到一个函数：平均数为什么排成这个顺序？]]></title>
            <link>https://yjy.hauchet.cn/article/power-mean-inequality-chain</link>
            <guid>https://yjy.hauchet.cn/article/power-mean-inequality-chain</guid>
            <pubDate>Sun, 06 Sep 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[我想解决的并不是“怎样再记住一串平均值不等式”，而是“为什么这些式子本来就应该排在一起”。把它们统一写成幂平均 M_p 后，调和、几何、算术、平方平均只不过是同一个函数在 p=-1,0,1,2 处的四个取值。]]></description>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-3d39f85982ee817e8269e29654950c81"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><div class="notion-callout notion-blue_background_co notion-block-77017474eb284b63b7207e13073fc300"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="📐">📐</span></div><div class="notion-callout-text"><div class="notion-text notion-block-10d6ca6aa2ff48d0a734dda0fd0b0e07">我研究这个函数，最初并不是为了得到一个更高级的公式，而是因为高中数学里那一串平均值不等式实在太常用：调和平均、几何平均、算术平均、平方平均。单纯背诵当然能做题，可如果只把它们当成四个彼此独立的式子，就看不出<b>为什么它们偏偏要连成这一串不等号</b>。</div><div class="notion-text notion-block-5e1694a9b0bb4fd196f9e496a32d1920">真正让我满意的解释，是把它们看成<b>同一个函数在不同参数处的取值</b>。</div></div></div><div class="notion-table-of-contents notion-gray notion-block-1cd30a7311c54d4595cb3dfcb0ecb835"><a href="#06438868514c42deaca5926018179707" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">一、先别背不等式，先把四个平均数摆在一起</span></a><a href="#2ba2c83500cc44bb9ccbbdc605bb57df" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">二、把算术平均和平方平均写成同一种样子</span></a><a href="#9fe2c990f6a04f30a7e4a002c1c8a1bc" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">三、最漂亮的缺口恰好在 </span></a><a href="#4d7b4aead1294fb597b5485d1f7250cf" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">四、整串不等式其实只是函数单调性的四个采样点</span></a><a href="#33b4745c8dc542ccad3d4806170ed091" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">五、把参数继续往两边推，整幅图景就出来了</span></a><a href="#a2a9078d12794846a92b45742e54602c" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">六、如果给高中生讲，我会让这个公式自己“被发现”出来</span></a><a href="#06f1803bd6e84960ae65e49f55861d5b" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:24px">第一步：从已经认识的式子出发</span></a><a href="#38427af9baf94aa6b84ae4381edb852b" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:24px">第二步：让学生自己猜参数式</span></a><a href="#3a8553b072f243669ff9986afd4faa86" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:24px">第三步：故意撞上  的“故障”</span></a><a href="#8e97ed60e69848ab9b5f0049998eeb4a" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:24px">第四步：把不等式问题改写成函数问题</span></a><a href="#e6955dfe323b49168b7d372e0078c78c" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">七、严格证明可以分层：高中课堂不必把所有技术一次讲完</span></a><a href="#1667140f602e4b1187d500a3b7e6f099" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">八、我真正想让学生记住的，不是那四个字母</span></a></div><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-06438868514c42deaca5926018179707" data-id="06438868514c42deaca5926018179707"><span><div id="06438868514c42deaca5926018179707" class="notion-header-anchor"></div><a class="notion-hash-link" href="#06438868514c42deaca5926018179707" title="一、先别背不等式，先把四个平均数摆在一起"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">一、先别背不等式，先把四个平均数摆在一起</span></span></h3><div class="notion-text notion-block-e344aa939bda474983510628feaaf8f0">对两个正数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，高中里经常见到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-69a66ef1c1ff4ad9b9d459c265fce0c5">通常的记法是：</div><ul class="notion-list notion-list-disc notion-block-a30a92a998e64f04adfc007d996f289e"><li>调和平均 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></li></ul><ul class="notion-list notion-list-disc notion-block-a1162b917eee45aaa751a51b2f26f672"><li>几何平均 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></li></ul><ul class="notion-list notion-list-disc notion-block-27fb15fd9a67437694e19d8eab8bf553"><li>算术平均 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></li></ul><ul class="notion-list notion-list-disc notion-block-0c39cdc44fa04eb8922c48a9672d1871"><li>平方平均 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></li></ul><div class="notion-text notion-block-abc9585d81a1412a8c679dec7ae52558">于是我们记住</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-570117de8f23451a9ba90bc516441c53">当然，每两个式子之间都可以单独证明。但这仍然没有回答我真正关心的问题：</div><blockquote class="notion-quote notion-block-dd2bc1bfd792487b832088abfbc7339a"><div><b>为什么是这四个式子？为什么它们应该按照这个顺序排成一条链？</b></div></blockquote><div class="notion-text notion-block-e12037e2cdcc49b88223c385561c944f">如果只是逐个证明，得到的是三个局部结论；我想找的是把整条链一次解释清楚的结构。</div><hr class="notion-hr notion-block-1144cb4b48b84a928b1de206e6ea580d"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-2ba2c83500cc44bb9ccbbdc605bb57df" data-id="2ba2c83500cc44bb9ccbbdc605bb57df"><span><div id="2ba2c83500cc44bb9ccbbdc605bb57df" class="notion-header-anchor"></div><a class="notion-hash-link" href="#2ba2c83500cc44bb9ccbbdc605bb57df" title="二、把算术平均和平方平均写成同一种样子"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">二、把算术平均和平方平均写成同一种样子</span></span></h3><div class="notion-text notion-block-8bc3cb17bb1645baa63e1c205454eddb">先看最熟悉的两个：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-65c4c9abbd7d4e568a460cf4ffa0da0f">它们的外形几乎已经在提示一个更一般的式子：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-bb391f9315d14c8e84892c417d549752">这就是二元幂平均。</div><div class="notion-text notion-block-245cd46f6a004b74be4ae422ee5e02f1">现在不妨像做实验一样，把不同的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 代进去。</div><div class="notion-text notion-block-08179aacceb24900b8de0ed511cb29d9">当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-7b7f0e32aad248eaa9d4312cf297d790">当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-0e147ad1479c4b848ecae760f0e45b93">再试一个不那么显眼的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-1345975765354bc5895104785a3cbc16">于是原本看起来风格完全不同的三个平均数，突然排进了同一个参数族：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e7d025cab45b449d9db85b84853e6e3f">这时候只剩一个最有意思的问题：<b>几何平均在哪里？</b></div><hr class="notion-hr notion-block-0b81f94d3ae644d0bd0058463c17fc8c"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-9fe2c990f6a04f30a7e4a002c1c8a1bc" data-id="9fe2c990f6a04f30a7e4a002c1c8a1bc"><span><div id="9fe2c990f6a04f30a7e4a002c1c8a1bc" class="notion-header-anchor"></div><a class="notion-hash-link" href="#9fe2c990f6a04f30a7e4a002c1c8a1bc" title="三、最漂亮的缺口恰好在 "><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">三、最漂亮的缺口恰好在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></span></span></h3><div class="notion-text notion-block-d2a68e23e6cc4df2a64ec801af93d760">直接把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 代入</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d4f23cdb39ac4063951e7d444bf65652">当然不行，因为它会变成 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 型。</div><div class="notion-text notion-block-89f59dd86c274df69cc0518658e81fb0">但这个“公式坏掉的地方”恰恰最漂亮，因为极限正好把几何平均补了回来：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-aaddad454d5242e69de3a643933cdfda">因此我们完全可以定义</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-95fc6ead10f349f69dfa29e8dfc9e976">这样一来，四个平均数终于全部进入同一个函数：</div><table class="notion-simple-table notion-block-29554c3ff1154932a3b5dcef3c98c1dc"><tbody><tr class="notion-simple-table-row notion-simple-table-header-row notion-block-59b3f0085e6d4eeca5765847bc667f16"><td class="" style="width:120px"><div class="notion-simple-table-cell">参数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td><td class="" style="width:120px"><div class="notion-simple-table-cell"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td><td class="" style="width:120px"><div class="notion-simple-table-cell"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td><td class="" style="width:120px"><div class="notion-simple-table-cell"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td><td class="" style="width:120px"><div class="notion-simple-table-cell"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td></tr><tr class="notion-simple-table-row notion-block-88ae0a02051046f091bb9dceb28c86fe"><td class="" style="width:120px"><div class="notion-simple-table-cell">平均数</div></td><td class="" style="width:120px"><div class="notion-simple-table-cell">调和平均</div></td><td class="" style="width:120px"><div class="notion-simple-table-cell">几何平均</div></td><td class="" style="width:120px"><div class="notion-simple-table-cell">算术平均</div></td><td class="" style="width:120px"><div class="notion-simple-table-cell">平方平均</div></td></tr><tr class="notion-simple-table-row notion-block-87e3e772bad24b55a1333baeb69399fc"><td class="" style="width:120px"><div class="notion-simple-table-cell"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td><td class="" style="width:120px"><div class="notion-simple-table-cell"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td><td class="" style="width:120px"><div class="notion-simple-table-cell"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td><td class="" style="width:120px"><div class="notion-simple-table-cell"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td><td class="" style="width:120px"><div class="notion-simple-table-cell"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></div></td></tr></tbody></table><div class="notion-text notion-block-fead5c5b612747f9add591076f593efc">这一步对我来说特别重要。几何平均不再像一个为了补齐不等式链而硬塞进去的特殊式子，而是整个幂平均函数在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 处的<b>连续延拓</b>。</div><div class="notion-text notion-block-937d1ddb25d3403fbdab6c44a288985f">如果想严格算这个极限，可以取对数：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-7b821631273b4429bdcd1eee0083138b">使用洛必达法则，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-03dcfb53254c46eba2467aa95760ef84">因此</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><hr class="notion-hr notion-block-295364aca01e44e78723528d20d52be7"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-4d7b4aead1294fb597b5485d1f7250cf" data-id="4d7b4aead1294fb597b5485d1f7250cf"><span><div id="4d7b4aead1294fb597b5485d1f7250cf" class="notion-header-anchor"></div><a class="notion-hash-link" href="#4d7b4aead1294fb597b5485d1f7250cf" title="四、整串不等式其实只是函数单调性的四个采样点"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">四、整串不等式其实只是函数单调性的四个采样点</span></span></h3><div class="notion-text notion-block-9025a36fe9a34a31ab75160174292abb">真正统一一切的是幂平均的核心性质：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ffe810c7c6584503a78140f58ef65da2">当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时，不等号严格成立。</div><div class="notion-text notion-block-c662ba037e4049b29b60e15922e85b6d">于是把</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-aa7d7a52170245eb9440ab3aa45f6af2">依次代进去，根本不需要再把三段不等式当成三个独立事实：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-93801809f959428f88ecf60c17ca7b2e">也就是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a64a1af2edf54e329a286a35736ae824">所以我现在更愿意把这串关系理解成：</div><blockquote class="notion-quote notion-block-c0032ec16f3e4c23b1dcd69426e4e2c1"><div><b>调和、几何、算术、平方平均不是四个碰巧能连起来的公式，而是同一个单调函数上的四个点。</b></div></blockquote><div class="notion-text notion-block-0ff25e7cf49e40a6b42be46c95abdd2e">这正是我觉得比“背 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>”更有数学美感的地方。</div><hr class="notion-hr notion-block-40579c5d519f4738904c4ef34a34af46"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-33b4745c8dc542ccad3d4806170ed091" data-id="33b4745c8dc542ccad3d4806170ed091"><span><div id="33b4745c8dc542ccad3d4806170ed091" class="notion-header-anchor"></div><a class="notion-hash-link" href="#33b4745c8dc542ccad3d4806170ed091" title="五、把参数继续往两边推，整幅图景就出来了"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">五、把参数继续往两边推，整幅图景就出来了</span></span></h3><div class="notion-text notion-block-bc6857f5894046028d83c455e8e1d3ee">既然 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 可以取任意实数，就没有理由停在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。</div><div class="notion-text notion-block-50f7059930244a94b9e49664fcc6d95d">当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时，较大的那个数会在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 中占据绝对优势，因此</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-12bde672d0614c87bfdfd2898b0d3024">当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时，则有</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f9b4b0ab22274461bd7a1b48ce95b49b">于是，对于 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，可以看到完整的“平均数地图”：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-434f098a2ce944908828f49d3ef102a1">具体到我们熟悉的几个点，就是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-06f4c1acab0f4cd4aa3429bc72caed2f">如果 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，当然所有平均数都等于同一个数。</div><div class="notion-text notion-block-52d74e21bd7e47de9e9b08182b2a3b7a">从这个角度看，<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 就像一个旋钮：</div><ul class="notion-list notion-list-disc notion-block-e319166fe506409194fcff60f0e5d703"><li>把它不断往负方向拧，平均值越来越偏向较小者；</li></ul><ul class="notion-list notion-list-disc notion-block-587b7dcf1b314d7a826d204e73236240"><li>把它不断往正方向拧，平均值越来越偏向较大者；</li></ul><ul class="notion-list notion-list-disc notion-block-cfb92722f19644a8a66173dd5b6852dc"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>、<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>、<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 只是这条连续变化曲线上的几个特别熟悉的位置。</li></ul><div class="notion-text notion-block-5e7febd33c0d41c187034cb5af17a4cb">原本静态的一串公式，就这样变成了一个会“运动”的对象。</div><hr class="notion-hr notion-block-5f1733b15f2c43e4883682cf58453d89"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-a2a9078d12794846a92b45742e54602c" data-id="a2a9078d12794846a92b45742e54602c"><span><div id="a2a9078d12794846a92b45742e54602c" class="notion-header-anchor"></div><a class="notion-hash-link" href="#a2a9078d12794846a92b45742e54602c" title="六、如果给高中生讲，我会让这个公式自己“被发现”出来"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">六、如果给高中生讲，我会让这个公式自己“被发现”出来</span></span></h3><div class="notion-text notion-block-53c7c50d0b3d4c139cc41049242429cf">我不会一上来就说“今天学习幂平均不等式”。那样只是把一个需要记忆的结论，换成另一个更大的结论。</div><div class="notion-text notion-block-7864249246194fe69c8ba009a3a538fc">我更想按下面的顺序让它自然出现。</div><h4 class="notion-h notion-h3 notion-h-indent-1 notion-block-06f1803bd6e84960ae65e49f55861d5b" data-id="06f1803bd6e84960ae65e49f55861d5b"><span><div id="06f1803bd6e84960ae65e49f55861d5b" class="notion-header-anchor"></div><a class="notion-hash-link" href="#06f1803bd6e84960ae65e49f55861d5b" title="第一步：从已经认识的式子出发"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">第一步：从已经认识的式子出发</span></span></h4><div class="notion-text notion-block-9f4cc2b4cb984d0ab7ff86ada9faf40a">先写</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-887b0cb7fa294aa182f0c1cd29065618">再故意改写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-86ff260f5e0143f4b11f7b8871146997">然后把平方平均写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-72b6ac34f3114956ab450311224576b3">问一句：<b>这两个式子是不是其实长得一样？</b></div><h4 class="notion-h notion-h3 notion-h-indent-1 notion-block-38427af9baf94aa6b84ae4381edb852b" data-id="38427af9baf94aa6b84ae4381edb852b"><span><div id="38427af9baf94aa6b84ae4381edb852b" class="notion-header-anchor"></div><a class="notion-hash-link" href="#38427af9baf94aa6b84ae4381edb852b" title="第二步：让学生自己猜参数式"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">第二步：让学生自己猜参数式</span></span></h4><div class="notion-text notion-block-fd8d8d78b76444c5998eaea3d72a70f9">自然猜出</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a060ba215f9d4e5b8a1e32851564c291">然后试 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，调和平均自己冒出来。</div><h4 class="notion-h notion-h3 notion-h-indent-1 notion-block-3a8553b072f243669ff9986afd4faa86" data-id="3a8553b072f243669ff9986afd4faa86"><span><div id="3a8553b072f243669ff9986afd4faa86" class="notion-header-anchor"></div><a class="notion-hash-link" href="#3a8553b072f243669ff9986afd4faa86" title="第三步：故意撞上  的“故障”"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">第三步：故意撞上 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的“故障”</span></span></h4><div class="notion-text notion-block-a846fdd4d835485582f7041054956845">学生会发现这里不能直接代。</div><div class="notion-text notion-block-977c104b5e5e42c58a7fd8d017d536da">这时再告诉他：公式虽然暂时失效，但极限却恰好等于几何平均。</div><div class="notion-text notion-block-9c6e472d5b5541e6b54accc787518ade">这种“缺口被补上”的感觉，比直接宣布 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 更有冲击力。</div><h4 class="notion-h notion-h3 notion-h-indent-1 notion-block-8e97ed60e69848ab9b5f0049998eeb4a" data-id="8e97ed60e69848ab9b5f0049998eeb4a"><span><div id="8e97ed60e69848ab9b5f0049998eeb4a" class="notion-header-anchor"></div><a class="notion-hash-link" href="#8e97ed60e69848ab9b5f0049998eeb4a" title="第四步：把不等式问题改写成函数问题"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">第四步：把不等式问题改写成函数问题</span></span></h4><div class="notion-text notion-block-86facb6ff77346c98a8d392ee24cbe8e">最后再问：</div><blockquote class="notion-quote notion-block-948ea9b7fcf64b6f8bde03f8684a11da"><div>如果 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 随 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 增大而增大，那么原来的整串平均值不等式是不是一下子就都解释了？</div></blockquote><div class="notion-text notion-block-f8f4dbcae1c94365b18a0ebae947d77c">这样，学生看到的就不再是三个等待分别证明的不等式，而是一个统一的问题：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><hr class="notion-hr notion-block-9f0ab993455f4d3fa154da3e54dcc2de"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-e6955dfe323b49168b7d372e0078c78c" data-id="e6955dfe323b49168b7d372e0078c78c"><span><div id="e6955dfe323b49168b7d372e0078c78c" class="notion-header-anchor"></div><a class="notion-hash-link" href="#e6955dfe323b49168b7d372e0078c78c" title="七、严格证明可以分层：高中课堂不必把所有技术一次讲完"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">七、严格证明可以分层：高中课堂不必把所有技术一次讲完</span></span></h3><div class="notion-text notion-block-6a7eeeeb05524ea483c02ad365c1ee56">这里我反而觉得应该区分“理解结构”和“完成严格证明”。</div><div class="notion-text notion-block-fd0971eb4bf542c9bd91c89736cefef7">对于高中课堂，完全可以先用这个函数解释<b>为什么四种平均数天然属于同一条链</b>，然后再用高中范围内熟悉的方法分别证明需要使用的具体不等式。这样既不牺牲严谨，也不会为了证明一般幂平均定理而把课堂拖进过多高等数学。</div><div class="notion-text notion-block-29b6ea5eb88e4cefb86726c70150ebf5">如果作为竞赛拓展、大学先修或者课外讨论，则可以进一步证明 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的完整单调性。</div><div class="notion-text notion-block-31dcdc501ecc4c3a861ab2311ef7d450">一种非常漂亮的做法是令</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-02339d28d3d541928b74200b3be35c27">于是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-3b353b80279b4706a6377ffeaac13c3c">而且</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8a4aada7e88848a8b726e886455551ee">计算二阶导数：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a89d2215237641cf99da6681685df176">所以 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 是凸函数；当 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 时严格凸。</div><div class="notion-text notion-block-61a7d70575ad4f639afaa6414231fea9">现在注意</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-76ea88e81e274b08b05d5021ef9f3e13">它就是函数图像上 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 与 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 两点连线的斜率。</div><div class="notion-text notion-block-f0cc0d44f1ae44569519e58d82beed12">凸函数的割线斜率会随右端点向右移动而增大，因此</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-898cfec7ee6c44e1ba3108eebf107eab">随 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 增大而增大，也就是 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 单调增大，最终得到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-af0f8cdb899144bc9df2f1c29152a77e">我很喜欢这个证明，因为到最后，“平均值不等式”甚至不再主要是一个代数技巧问题，而变成了一个几何事实：<b>凸函数的割线斜率有序。</b></div><hr class="notion-hr notion-block-1afd677493e547aba4b19bf07e9fa050"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-1667140f602e4b1187d500a3b7e6f099" data-id="1667140f602e4b1187d500a3b7e6f099"><span><div id="1667140f602e4b1187d500a3b7e6f099" class="notion-header-anchor"></div><a class="notion-hash-link" href="#1667140f602e4b1187d500a3b7e6f099" title="八、我真正想让学生记住的，不是那四个字母"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">八、我真正想让学生记住的，不是那四个字母</span></span></h3><div class="notion-text notion-block-b26fc0028a184f77b8ceb9039104db1c">如果只是为了考试，当然可以记</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d793add215834de5ae6f0b663b2d9196">它短、快、实用。</div><div class="notion-text notion-block-e78b4fdfa16d4ac0a041dd506ac19b7d">但如果希望在记住结论之外，再多看见一点数学本身的结构，我觉得更值得记住的是：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-930f5b8642f14c729305e34aa8c99a7e">以及</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-0740f5f158a246e084a7d4398dbaf8ad">前者把四个平均数统一成一个对象，后者把一串不等号统一成一个性质。</div><div class="notion-text notion-block-083cceb02f67436ba77704b30160c3c0">于是原来需要记忆的</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-33eb33da65b0443893e38e26bf5a38df">不再是一串孤立符号，而只是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-51bf5712d88344c599b1f66d88a247c5">在高中数学里露出来的一小段。</div><div class="notion-callout notion-yellow_background_co notion-block-1de5b988635b43f2a71a1186aeefb8ae"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="✨">✨</span></div><div class="notion-callout-text"><div class="notion-text notion-block-eb4bb6a07c2d47218585ba9fe27238fd"><b>我最喜欢的不是用一个更复杂的公式替代四个简单公式，而是发现：它们从一开始就不是四件事。</b></div><div class="notion-text notion-block-fa58503d8d00465bb1c7c060ebb08a44">当几个看似零散的结论突然被一个统一对象串起来时，那种“原来如此”的感觉，可能正是数学最值得讲给学生的部分。</div></div></div><hr class="notion-hr notion-block-a5b89489ccd24787987c830dcc787a75"/><div class="notion-text notion-block-395bdafb06db4d68a0a2f9f59ddb30ee"><em>封面图片：Artturi Jalli / Unsplash。</em></div></main></div>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[我只是把零补了回来，却撞见了伯努利数：从等次幂求和到 Faulhaber 公式]]></title>
            <link>https://yjy.hauchet.cn/article/power-sums-to-bernoulli-numbers</link>
            <guid>https://yjy.hauchet.cn/article/power-sums-to-bernoulli-numbers</guid>
            <pubDate>Wed, 26 Aug 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[从一个最朴素的等次幂累计求和出发，我用待定系数算到 a=10，并因坚持补齐零系数项而看见整列消失的规律，最终沿着有限差分走到了 Faulhaber 公式与伯努利数。]]></description>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-3c89f85982ee81dd9920cd61baba11cd"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><div class="notion-callout notion-blue_background_co notion-block-47cf30ea6af94992814ac22f2c2553c3"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="🧮">🧮</span></div><div class="notion-callout-text"><div class="notion-text notion-block-ff6fada4c106443889b72501aae78616"><b>这不是我从 Faulhaber 公式出发的一次学习，而是一次反过来的探索。</b></div><div class="notion-text notion-block-5a5bef60c76b4332b483bec50c03708b">我最初只是想算清楚 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。后来为了把多项式写得足够严谨，我把那些通常会被省略的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 系数项也全部补了回来。也正是这些看似“什么都没有”的零，让规律第一次真正显现出来。</div></div></div><div class="notion-table-of-contents notion-gray notion-block-28140c52360448b3aeb741a6aa252a91"><a href="#2386c979c76c4e0797d2b5ccbe203209" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">一、起点：一个再普通不过的等次幂求和</span></a><a href="#de937df0f91d4d17a332c021f366c181" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">二、我一路算到了 </span></a><a href="#e404ecbc161b45f8b07cbff6e3d62baf" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">三、把零补回来以后，公式突然开始“排队”</span></a><a href="#216d16e5ca4f422985c6770cd2a90b9e" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">四、最先出现的三列规律</span></a><a href="#518f933f25604ac995c3fd0204d7abe7" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">五、最神奇的现象：不是某一项为零，而是整列为零</span></a><a href="#0bbcb95b8f0943168fdbde6d306be7f0" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">六、回头看原问题：差分才是多项式背后的骨架</span></a><a href="#96615587181a4ccebab8b4f297b128be" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">七、那些零最终把我带到了伯努利数</span></a><a href="#5606c5cba97c4c7d90de2ce735e29d40" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">八、我最喜欢这次探索的地方</span></a></div><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-2386c979c76c4e0797d2b5ccbe203209" data-id="2386c979c76c4e0797d2b5ccbe203209"><span><div id="2386c979c76c4e0797d2b5ccbe203209" class="notion-header-anchor"></div><a class="notion-hash-link" href="#2386c979c76c4e0797d2b5ccbe203209" title="一、起点：一个再普通不过的等次幂求和"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">一、起点：一个再普通不过的等次幂求和</span></span></h3><div class="notion-text notion-block-770fb844b5e64856a2582b82907c8286">我最开始研究的是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-5c4309ee3925413fb86a30ca88036840">低次情形当然很熟悉：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-61ca1347412343e196d30ccff41a7428">看着这些结果，我产生了一个很自然的猜想：<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 次幂的累计和，似乎总能写成一个 </b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 次多项式。</b></div><div class="notion-text notion-block-8fa5ef0c1b9145ac8d423b5b90d7fbbd">于是我没有先去找一般公式，而是直接设</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c33677adb1864cb0aabaee8d9cfff12d">因为 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，所以常数项应当为零。这样一来，一共有 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 个未知系数，我只要代入 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，就能得到同样数量的方程，再把系数解出来。</div><div class="notion-text notion-block-928b478ec06b4aaa863280ccac2bac9e">现在回头看，这其实就是多项式插值；但当时我的思路很朴素：<b>先假设结果长成一个多项式，然后用足够多的具体值把它钉死。</b></div><hr class="notion-hr notion-block-4fbe458e0ed84552aaced0e05833d15e"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-de937df0f91d4d17a332c021f366c181" data-id="de937df0f91d4d17a332c021f366c181"><span><div id="de937df0f91d4d17a332c021f366c181" class="notion-header-anchor"></div><a class="notion-hash-link" href="#de937df0f91d4d17a332c021f366c181" title="二、我一路算到了 "><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">二、我一路算到了 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span></span></span></h3><div class="notion-text notion-block-19a6eb7bc6e4410bbdccbee640d53bfb">以 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 为例，我先设</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-24b9cf1ff5d7498086d69af3a80d5626">再计算</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ee37a2ff02c341f9ad4838c1f27ab1b8">解出</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-771b354290a04fb6a298007528abfec1">也就是熟悉的</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f52aa956c3564b5e81ecb7698a9f8c95">但真正让我感兴趣的并不是某一个公式，而是：<b>如果继续算下去，这些多项式之间会不会存在某种纵向规律？</b></div><div class="notion-text notion-block-55dc5f8e4cb244b2ba776c1a963f2dfe">于是我把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 全部列了出来。</div><div class="notion-text notion-block-9c9813f989274361881878eb971b60be">这里有一个当时看似只是“书写严谨”的细节：<b>我没有把系数为零的项删掉。</b></div><div class="notion-text notion-block-79a7ab0b2c9d450ba745bd31f92aabd0">例如通常会写</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-5493d70dea3046c790d06ee9c55b3f3c">但我会把它写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8b545712672843afadb54392dba18aa8">因为既然我一开始设的是一个从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 一直排到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的完整多项式，那么即使某一项系数恰好为零，我也希望它仍然占据自己的位置。</div><div class="notion-text notion-block-3182af898a614a9aba6d16ecd3686cfc">后来我才意识到：<b>正是这个习惯，让“缺项”从视觉上的空白变成了真正的数据。</b></div><hr class="notion-hr notion-block-b3b27523dd14423f9d6c4303da4ecdec"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-e404ecbc161b45f8b07cbff6e3d62baf" data-id="e404ecbc161b45f8b07cbff6e3d62baf"><span><div id="e404ecbc161b45f8b07cbff6e3d62baf" class="notion-header-anchor"></div><a class="notion-hash-link" href="#e404ecbc161b45f8b07cbff6e3d62baf" title="三、把零补回来以后，公式突然开始“排队”"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">三、把零补回来以后，公式突然开始“排队”</span></span></h3><div class="notion-text notion-block-0ff9d676e5ee43fca6864ed88676e988">为了看清结构，我把几个结果完整写成：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-342d9fbd98024b3d9fd9e047853ad116">当这些式子严格对齐以后，我发现它们已经不像十个彼此独立的答案，而更像一张系数阵列。</div><div class="notion-text notion-block-582bb65bd2e84629925928af7d250797">如果把每一项按“距离最高次项的位置”来看，也就是统一写成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-4c5ca46271cf43b5b89d352164c6e5df">那么一些规律几乎是自己跳出来的。</div><hr class="notion-hr notion-block-24f88f5749384b30b476e19fd1bb109d"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-216d16e5ca4f422985c6770cd2a90b9e" data-id="216d16e5ca4f422985c6770cd2a90b9e"><span><div id="216d16e5ca4f422985c6770cd2a90b9e" class="notion-header-anchor"></div><a class="notion-hash-link" href="#216d16e5ca4f422985c6770cd2a90b9e" title="四、最先出现的三列规律"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">四、最先出现的三列规律</span></span></h3><div class="notion-text notion-block-351e3773275e4fbb8ae896b0ee845be4">第一列，也就是最高次项系数，依次是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e5cf61ddd148480f9f2fad3e6b889c92">所以自然猜到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-14d1e573e5ba44b58d5e87a01e4149e5">第二列更加整齐：无论 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 是多少，都是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a1b4a836fb034e599efcde370a0a9d19">第三列则是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-026d220a9fff47a196cace305da76c61">统一写成以 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 为分母后，规律变成</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-3d6a0ef960e04cde82bf0287f9afe2ce">于是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-2f38db3440ea41b3a3384368aec4feb7">也就是说，我已经可以从实验中猜出</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-079b4d570d6e475dadc0dc25782d5945">但真正让我觉得“这里面一定藏着东西”的，还不是这些漂亮的分数，而是接下来那些零。</div><hr class="notion-hr notion-block-0a4cf7958e5b477785e648541e8bc96e"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-518f933f25604ac995c3fd0204d7abe7" data-id="518f933f25604ac995c3fd0204d7abe7"><span><div id="518f933f25604ac995c3fd0204d7abe7" class="notion-header-anchor"></div><a class="notion-hash-link" href="#518f933f25604ac995c3fd0204d7abe7" title="五、最神奇的现象：不是某一项为零，而是整列为零"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">五、最神奇的现象：不是某一项为零，而是整列为零</span></span></h3><div class="notion-text notion-block-d283a14e6a0f4ab0b000ca51a9f6ddc5">完整对齐后，我发现</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-b2e60125283e4aa29512ff31e1703012">而再往后</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-7810c254b99048b4ba9ffbfef73007ca">……</div><div class="notion-text notion-block-868eb57ae9b0420db72cf7b3ccb5a9d0">也就是说，按相对次数排列时，出现了</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a037380ca5054fbba3f8940a9e46169a">这样的整列。</div><div class="notion-text notion-block-e52019f154c54c6d9f946be08ecbd338">于是一般结构开始呈现出一种非常明显的“隔项”形式：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-795cef2010b943fdac578ea2b26e39b5">继续观察非零列，还能猜出</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-5f90d650b92d4404a88d4e9424ae9d74">然后又是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ac2e7e4c2ee3445a8c4beda28e33762b">再后面则是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c3f0baaadd45464780e82d619f9d334d">所以整个式子开始像这样：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f1dd61fba9c941b8ad371dac5fe506e9">这时问题已经彻底变了。</div><div class="notion-text notion-block-be936f6eebd54ee0b8e8be046b50deae">一开始我想问的是：<b>“</b><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span><b> 怎么算？”</b></div><div class="notion-text notion-block-f421c0eea3fb4accb4bfc83929e9b9f2">到了这里，我真正想问的已经是：<b>“为什么这些系数会这样排列？为什么偏偏有些列永远是零？”</b></div><hr class="notion-hr notion-block-f96053dcbf984a30aa698db3f76bdc7b"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-0bbcb95b8f0943168fdbde6d306be7f0" data-id="0bbcb95b8f0943168fdbde6d306be7f0"><span><div id="0bbcb95b8f0943168fdbde6d306be7f0" class="notion-header-anchor"></div><a class="notion-hash-link" href="#0bbcb95b8f0943168fdbde6d306be7f0" title="六、回头看原问题：差分才是多项式背后的骨架"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">六、回头看原问题：差分才是多项式背后的骨架</span></span></h3><div class="notion-text notion-block-e5d29b9fce1842aaa5a86db65aebedc8">真正统一这些现象的是一个极其简单的关系：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-cf2c36469da84ae68bd3390939ea1452">因为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-ffd460c1e346489fb5a5df05f10482f9">而</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-e0acbb13d39a49629a2d38d86884a03b">相减以后只剩下 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。</div><div class="notion-text notion-block-2c762517176648e786390674a02fd4b4">这就是有限差分。</div><div class="notion-text notion-block-f4685c9e41b14cdfb95e909dffd69870">一个 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 次多项式做一次差分会降成 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 次多项式，所以这个关系首先解释了：为什么我最开始把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 猜成 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 次多项式，并不是偶然碰对。</div><div class="notion-text notion-block-6875031f216a4340bd4e958dbb133871">更重要的是，我甚至不需要再给每一个 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 单独代数值求方程组。只要设</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-9f3fc189a18444c4bc7b33b14e6b8732">再利用</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-451e1f6d27134cdb8e9d6c80ce3e7d32">把</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d0df1cdfe5744981adea5fac3bd84c89">左右两边按幂次比较，就可以逐项递推出全部系数。</div><div class="notion-text notion-block-ae08ccc579a0421d910c9f06e4f0edfb">最高次项给出</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f8ce58381bc04671b2910ef3801535fa">于是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-3163990a2e3e48f3b49ba0612937e485">下一项会得到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-3d89b5005ba1439191eff72359952bb4">再下一项得到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-24eed7795397452a8bf2c184e9dfd238">也就是说，我原来从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 一直算到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 才观察到的纵向规律，其实都可以从一个差分恒等式里系统地推出。</div><hr class="notion-hr notion-block-c7eb4927bf2045afa9e07a410523cbb6"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-96615587181a4ccebab8b4f297b128be" data-id="96615587181a4ccebab8b4f297b128be"><span><div id="96615587181a4ccebab8b4f297b128be" class="notion-header-anchor"></div><a class="notion-hash-link" href="#96615587181a4ccebab8b4f297b128be" title="七、那些零最终把我带到了伯努利数"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">七、那些零最终把我带到了伯努利数</span></span></h3><div class="notion-text notion-block-d62600fd8785496aa41c9de2c059fda6">继续整理系数以后，会出现一组特殊的常数：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8d7d9000496445aa972c6c4919b33920">它们就是伯努利数：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-0553a37bfc0146b3aa8f2edec9cc54b5">其中最对应我当时观察的性质是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-af763ecc41ae42ddae362cd1ac07927d">也就是说，我看到的</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-0a96e834a8834620a2189be2c1468bb9">那些整列为零的现象，正是奇数编号伯努利数消失在幂和公式中的投影。</div><div class="notion-text notion-block-e0c904bdb76c41f5ba3f37609a9a5dd9">最终，所有的等次幂求和都可以统一写成 Faulhaber 公式：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-22c748205c4142718b02dbb347bd43ed">这里采用 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的约定。</div><hr class="notion-hr notion-block-75d94b489a62479791114c1e2694bd38"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-5606c5cba97c4c7d90de2ce735e29d40" data-id="5606c5cba97c4c7d90de2ce735e29d40"><span><div id="5606c5cba97c4c7d90de2ce735e29d40" class="notion-header-anchor"></div><a class="notion-hash-link" href="#5606c5cba97c4c7d90de2ce735e29d40" title="八、我最喜欢这次探索的地方"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">八、我最喜欢这次探索的地方</span></span></h3><div class="notion-text notion-block-06d1e10e61cc4c14b2a55eacd1e1c3e0">现在再看这条路线，它其实非常有“实验数学”的味道：</div><ol start="1" class="notion-list notion-list-numbered notion-block-cb2fc72183dd4809a53e1950cba75197" style="list-style-type:decimal"><li>从几个低次公式出发，猜一个可能的结构；</li></ol><ol start="2" class="notion-list notion-list-numbered notion-block-0ee459175ef94af99133ef3c6212d81e" style="list-style-type:decimal"><li>用待定系数和具体数据不断生成样本；</li></ol><ol start="3" class="notion-list notion-list-numbered notion-block-7236564389c14ee5a49addf2b655459a" style="list-style-type:decimal"><li>把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 全部算出来；</li></ol><ol start="4" class="notion-list notion-list-numbered notion-block-196fcb25750c4df884b3a45e3e7aa938" style="list-style-type:decimal"><li>为了多项式项数上的严谨，不省略系数为零的项；</li></ol><ol start="5" class="notion-list notion-list-numbered notion-block-2d1931e2c3094485aa98a76b79109b73" style="list-style-type:decimal"><li>把所有系数严格对齐以后，开始纵向观察；</li></ol><ol start="6" class="notion-list notion-list-numbered notion-block-93cca06f8a664911a01dd1ad4d90cb44" style="list-style-type:decimal"><li>发现 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>、<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>、<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 等规律；</li></ol><ol start="7" class="notion-list notion-list-numbered notion-block-2249adeb12f146e1bb2a97796db6443c" style="list-style-type:decimal"><li>更进一步发现某些相对位置不是偶然缺项，而是<b>整列恒为零</b>；</li></ol><ol start="8" class="notion-list notion-list-numbered notion-block-6e29b9df90964d7d8534d265d4be6e69" style="list-style-type:decimal"><li>再用有限差分解释这些规律为什么必然出现；</li></ol><ol start="9" class="notion-list notion-list-numbered notion-block-0c743aebb4f646b8bb863ef943f2b733" style="list-style-type:decimal"><li>最后才发现，这套结构早已有名字——Faulhaber 公式与伯努利数。</li></ol><div class="notion-callout notion-yellow_background_co notion-block-945505bce4a54517a6e247f712c46ec4"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="0️⃣">0️⃣</span></div><div class="notion-callout-text"><div class="notion-text notion-block-ed5363d0197c4c5b87e5c39a5fc7022c">我觉得这次探索里最有意思的一句话是：<b>在化简答案时，零往往意味着“没有东西”；但在寻找结构时，零本身也可能是一条信息。</b></div><div class="notion-text notion-block-c96811153c90435ca0743d22195c4860">如果当时我像通常那样把所有 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 都删掉，我看到的可能只是一堆缺项的多项式；正因为我坚持把它们补齐，那些“什么都没有的位置”才排成了一列，最后指向了更深的规律。</div></div></div><div class="notion-text notion-block-d1a573467f444ba08b76a259c65636f8">所以我最初只是想找一种计算</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-bc702cdfa2c548de97ddbdb5b16c7dee">的方法。</div><div class="notion-text notion-block-1b624cedaca24a649e775b7093a86d60">结果一路算下去，却从待定系数走到了多项式插值，从多项式插值走到了有限差分，又从有限差分撞见了伯努利数。</div><div class="notion-text notion-block-ce4c1f7c680f43c5b38f94f28c97ea39">数学里让我着迷的，往往就是这种时刻：<b>你以为自己只是在认真算一道题，直到某个微不足道的细节突然排列起来，告诉你背后还有一整套结构。</b></div></main></div>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[当宇宙装不下一个数字：从古戈尔、葛立恒数到 TREE(3)]]></title>
            <link>https://yjy.hauchet.cn/article/large-numbers-googol-graham-tree3</link>
            <guid>https://yjy.hauchet.cn/article/large-numbers-googol-graham-tree3</guid>
            <pubDate>Tue, 04 Aug 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[从古戈尔与幂塔出发，穿过斯奎斯数、葛立恒数、TREE(3)、Busy Beaver 与拉约数，理解人类如何定义远超物质宇宙承载能力的有限大数。]]></description>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-3b29f85982ee81a299d6fb3798b8048f"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><div class="notion-callout notion-blue_background_co notion-block-476c93ec85b647eebce66fd75343fd85"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="🔢">🔢</span></div><div class="notion-callout-text"><div class="notion-text notion-block-be9ae947310048ed8a74e5cc4aa8a4b5"><b>“有限与无限”系列 · 上篇</b></div><div class="notion-text notion-block-7d6e28162a6548109503b4e5d6b0a835">有些数大到无法写完，有些数大到无法用普通指数描述，还有些数之所以巨大，并不是因为它们堆叠了更多的零，而是因为它们来自组合结构、递归过程与可计算性的边界。</div></div></div><div class="notion-text notion-block-cc08ccf7b18f457eb4f661933a5e1646">人类很早就开始给大数命名。万、亿、兆、京、垓、秭、穰、沟、涧、正、载，已经足以覆盖日常生活与绝大多数科学计数。佛教典籍中还出现极、恒河沙、阿僧祇、那由他、不可思议、无量大数等名称。不同典籍和时代对这些数名的具体进位规则并不完全一致，但无论采用哪一种传统，它们在现代大数理论面前都仍只是有限数世界里相对温和的一角。</div><div class="notion-text notion-block-372bda842c504c6dbfe602ccba89d7f1">真正有趣的问题不是“还能再加多少个零”，而是：<b>我们能否发明一种有限的语言，准确指向一个远远大于任何可实际书写数字的整数？</b></div><div class="notion-table-of-contents notion-gray notion-block-4d25a638ef2b4f1ab3a80485fcaf87a1"><a href="#db63ef7bc01f443e868348480493a576" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">一、古戈尔：宇宙中的粒子也不够用</span></a><a href="#6c227a1ebf984bb78a2346fd129fbb06" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">二、指数塔：重复乘法已经不够用了</span></a><a href="#badc5391f7724cb087187cdce8ab261f" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">三、几个曾震撼公众的著名大数</span></a><a href="#59c3691c8e43468d91208d8f096cf8fd" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:24px">1. 斯奎斯数：从解析数论中诞生</span></a><a href="#1dc61554bf4547d79d0ae1c3b81ab661" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:24px">2. 莫泽数：五边形里的递归</span></a><a href="#eeef4e5c19754e53a4d8e6cc1316f72d" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:24px">3. 葛立恒数：曾进入吉尼斯纪录的证明上界</span></a><a href="#9ef188dde7f94ef7824d7fed340376f5" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">四、TREE(3)：葛立恒数只是它脚下的尘埃</span></a><a href="#100dbac4d8314998a7bd4688bf47c54a" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:24px">TREE(3) 与葛立恒数谁更大？</span></a><a href="#a7f7802587a44f44baad06ac24c061a9" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">五、SCG(13)：简单图游戏中的又一次爆炸</span></a><a href="#186ba3b6d9574e1aa2ad2580587755d3" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">六、Busy Beaver：增长快到任何算法都追不上</span></a><a href="#e6f5dedc39ff4878af4f1b875e99339d" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">七、拉约数：比拼的不是增长，而是语言能表达什么</span></a><a href="#d8d47d513de047e8b7c0c98f0819e8b5" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">八、为什么不存在最大的有限数？</span></a><a href="#eac7c80713ca46c4ac0b66def43ea1cc" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">九、再大的有限数，也没有碰到无限</span></a><a href="#9b2889e319de43f3b9ad7a24e9f64c64" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">参考资料</span></a></div><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-db63ef7bc01f443e868348480493a576" data-id="db63ef7bc01f443e868348480493a576"><span><div id="db63ef7bc01f443e868348480493a576" class="notion-header-anchor"></div><a class="notion-hash-link" href="#db63ef7bc01f443e868348480493a576" title="一、古戈尔：宇宙中的粒子也不够用"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">一、古戈尔：宇宙中的粒子也不够用</span></span></h3><div class="notion-text notion-block-5820c63238f340469de45b6582ce0b33">1920 年代，美国数学家爱德华·卡斯纳请年幼的侄子米尔顿·西罗塔为 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 起名。孩子给出的名字是 <b>googol</b>，中文通常译作“古戈尔”。</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-6b0ee846fa9144f899bc234e6e34b04d">它就是 1 后面跟 100 个零。这个数虽然巨大，却仍可完整写在一张纸上。它的重要性不在于数学结构有多复杂，而在于它第一次以一个生动名字提醒公众：数学中的数可以远远超过物理世界中可计数对象的数量。</div><div class="notion-text notion-block-5f4bbb811a164b78afb1e15338f8d349">可观测宇宙中的粒子数常被粗略估计在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 数量级附近。即使这个估计只用于直觉比较，也足以说明：我们无法拿宇宙中的每个粒子去对应古戈尔中的每一个单位。</div><div class="notion-text notion-block-8e56b0ee9fde44049bf0df9127938224">随后出现的是 <b>古戈尔普勒克斯</b>：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8590c967904d42778aec9acea74ca078">它是 1 后面跟一个古戈尔个零。不要把它误解为 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；指数运算从右往左结合，<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 比 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 大得不可同日而语。</div><div class="notion-text notion-block-8d8570d53a80477ba328dc4e14a671e3">如果每秒写十亿个零，从宇宙诞生一直写到今天，也远远写不完一个古戈尔普勒克斯。于是我们第一次遇到大数理论的核心事实：</div><blockquote class="notion-quote notion-block-4f990d9ff7a94e3d909b627f95591d9d"><div><b>数字本身无法展开，不妨碍它被精确定义。</b></div></blockquote><hr class="notion-hr notion-block-9196dd55e64f472e8d915ef01a5eb0c4"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-6c227a1ebf984bb78a2346fd129fbb06" data-id="6c227a1ebf984bb78a2346fd129fbb06"><span><div id="6c227a1ebf984bb78a2346fd129fbb06" class="notion-header-anchor"></div><a class="notion-hash-link" href="#6c227a1ebf984bb78a2346fd129fbb06" title="二、指数塔：重复乘法已经不够用了"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">二、指数塔：重复乘法已经不够用了</span></span></h3><div class="notion-text notion-block-f788f97e971c42598e560c1eeb39d589">加法是重复计数，乘法是重复加法，乘方是重复乘法。若继续重复乘方，就得到<b>幂塔</b>。</div><div class="notion-text notion-block-d4ca833961ff4de09fb2d92f22cfbd9c">例如：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d4ced8f314684235973086b62e02c0b1">这里的 4 表示幂塔高度，而不是指数前的乘数。只增加一层，规模就会发生远超普通指数增长的跃迁。</div><div class="notion-text notion-block-9f926d2f5f884b70a5d8568f22396e15">为了压缩这种结构，唐纳德·克努特提出了<b>上箭头记号</b>（以下采用递归定义，避免公式渲染歧义）：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-61b6d60fd4fb4396b3d7b194ed51af01">双箭头表示幂塔（tetration）；为避免展开式在不同渲染器中产生歧义，以下面的递归定义为准：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-8ee2bbba89474e4bb9b1a390b88d6972">三箭头则表示重复进行双箭头运算，可递归定义为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-9bc31d81ebe74745acb49a4a89f6b0be">箭头数量每增加一个，增长层级都会发生质变。</div><div class="notion-text notion-block-3f169b15a1fb4e00b7f89685cdaea6cb">例如：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-57a5d0960ac94d8b8cabe2bc000fd35e">即使这个数已经极其庞大，它与后文中的葛立恒数相比仍只是起跑线附近的一粒尘埃。</div><div class="notion-text notion-block-0cd683bf0d314026a431bcf41703ffe9">类似的压缩记号还包括康威链式箭号、超运算、斯坦豪斯—莫泽记号等。大数记号的意义并不是制造视觉奇观，而是把“不断重复某种增长操作”变成一个可递归定义、可严格推理的数学对象。</div><hr class="notion-hr notion-block-3353fc3a2bea42c9a9599aa8eda5f869"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-badc5391f7724cb087187cdce8ab261f" data-id="badc5391f7724cb087187cdce8ab261f"><span><div id="badc5391f7724cb087187cdce8ab261f" class="notion-header-anchor"></div><a class="notion-hash-link" href="#badc5391f7724cb087187cdce8ab261f" title="三、几个曾震撼公众的著名大数"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">三、几个曾震撼公众的著名大数</span></span></h3><h4 class="notion-h notion-h3 notion-h-indent-1 notion-block-59c3691c8e43468d91208d8f096cf8fd" data-id="59c3691c8e43468d91208d8f096cf8fd"><span><div id="59c3691c8e43468d91208d8f096cf8fd" class="notion-header-anchor"></div><a class="notion-hash-link" href="#59c3691c8e43468d91208d8f096cf8fd" title="1. 斯奎斯数：从解析数论中诞生"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">1. 斯奎斯数：从解析数论中诞生</span></span></h4><div class="notion-text notion-block-fc1987b28ba7425fa32f23601c554aa7">斯奎斯数与素数计数函数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 和对数积分 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的比较有关。早期结果给出了一个极其巨大的上界，用来保证两条函数曲线最终会出现交叉。</div><div class="notion-text notion-block-79240e1813ba47b18ed6bf2184207386">在假设黎曼猜想成立时，斯奎斯最初得到的上界常被写作大约</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a1c759388fa6465cbff4adf49bfa3f0a">不假设黎曼猜想时，早期上界还要更大。后来这个上界被大幅改进，所以“斯奎斯数”更适合被理解为大数史上的标志：一个严肃的数论问题，曾自然产生出远超日常想象的数值上界。</div><h4 class="notion-h notion-h3 notion-h-indent-1 notion-block-1dc61554bf4547d79d0ae1c3b81ab661" data-id="1dc61554bf4547d79d0ae1c3b81ab661"><span><div id="1dc61554bf4547d79d0ae1c3b81ab661" class="notion-header-anchor"></div><a class="notion-hash-link" href="#1dc61554bf4547d79d0ae1c3b81ab661" title="2. 莫泽数：五边形里的递归"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">2. 莫泽数：五边形里的递归</span></span></h4><div class="notion-text notion-block-cf451ba7c6b0496d99410f2800d5372f">斯坦豪斯提出用几何图形表示大数：把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 放入三角形表示 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，放入正方形表示重复 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 次三角形操作，放入圆形则继续重复正方形操作。</div><div class="notion-text notion-block-d06059a22f6f45b9a5bcb3d017a27a33">莫泽把这个体系向外推进，并定义了著名的<b>莫泽数</b>。它远远超过古戈尔普勒克斯，也超过大多数仅靠固定高度幂塔构造的数字。</div><div class="notion-text notion-block-d67d472f9f99404e8bfbc979a3a28eb2">莫泽数的启示是：真正的大数往往来自<b>递归定义的操作次数本身也由巨大数字控制</b>。</div><h4 class="notion-h notion-h3 notion-h-indent-1 notion-block-eeef4e5c19754e53a4d8e6cc1316f72d" data-id="eeef4e5c19754e53a4d8e6cc1316f72d"><span><div id="eeef4e5c19754e53a4d8e6cc1316f72d" class="notion-header-anchor"></div><a class="notion-hash-link" href="#eeef4e5c19754e53a4d8e6cc1316f72d" title="3. 葛立恒数：曾进入吉尼斯纪录的证明上界"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">3. 葛立恒数：曾进入吉尼斯纪录的证明上界</span></span></h4><div class="notion-text notion-block-d2593c62ec914443bf69361f6d00dc6c">葛立恒数来自拉姆齐理论中的一个高维超立方体染色问题。它并不是为了争夺“最大数字”而编造出来的，而是一个数学证明中得到的有限上界。</div><div class="notion-text notion-block-25ff35edc5cc4db29f4478353b5e84d2">先定义</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-abfc76c4811843a29e7311e0625feaad">接着递归定义</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-83652e1271554e2bbcdd8ea1ecc94fe3">最后令</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a2f501b9b2a84337a254916e2ef75d19">这就是通常所说的葛立恒数。</div><div class="notion-text notion-block-f9472f40290e4a858f64af9a20454700">注意它的恐怖之处：第二步中的箭头数量不是 4、100 或古戈尔，而是前一步产生的整个巨大数字 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。仅仅 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 就已无法用通常幂塔直观展开；从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 迭代到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，增长层级不断自我放大。</div><div class="notion-text notion-block-120d456585cb4c7eb9a9ea9ec403f175">尽管如此，葛立恒数仍是一个普通的有限整数。它有确定的个位数，事实上也可以借助模运算求出末尾若干位。巨大并不等于模糊。</div><hr class="notion-hr notion-block-6f862846371d42aea9490f9604804ebc"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-9ef188dde7f94ef7824d7fed340376f5" data-id="9ef188dde7f94ef7824d7fed340376f5"><span><div id="9ef188dde7f94ef7824d7fed340376f5" class="notion-header-anchor"></div><a class="notion-hash-link" href="#9ef188dde7f94ef7824d7fed340376f5" title="四、TREE(3)：葛立恒数只是它脚下的尘埃"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">四、TREE(3)：葛立恒数只是它脚下的尘埃</span></span></h3><div class="notion-text notion-block-ba5eaf6d779c4ad3bdd661d4c8190d2b">如果只知道一个著名大数，很多人会想到葛立恒数；但在现代大数文化中，<b>TREE(3)</b> 才是更具冲击力的名字。</div><div class="notion-text notion-block-58a55ea355224ee7a9071d7df5ec7ccf">TREE 函数来自图论与克鲁斯卡尔树定理。我们考虑带有有限颜色标记的有限根树，并构造一个序列，要求第 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 棵树至多有 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 个顶点，同时任何较早的树都不能以保持结构与标记的方式嵌入较晚的树。</div><div class="notion-text notion-block-27555fc81e0443c0964161871ce4c7d7"><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 定义为使用 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 种标记时，这类坏序列能够达到的最大长度。</div><div class="notion-text notion-block-c6b1dd42193a40eda24212c01bd78d5f">前两个值小得出奇：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-24d1cbc1cb6342c8b2cd2356c1abfc06">然而</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-6610cef2c0c3421c9f47b2746413b126">却发生了难以想象的爆炸。它不仅大于葛立恒数，而且大到克努特上箭头、康威链式箭号等常见记号若不引入更高层递归，也很难给出有意义的尺度比较。</div><div class="notion-text notion-block-be01880e381d44eab7ab8361989098e6">这种突变揭示了组合数学中的一种深层现象：</div><ul class="notion-list notion-list-disc notion-block-f6766e57c7274594aea70b073d587840"><li>局部规则非常简单；</li></ul><ul class="notion-list notion-list-disc notion-block-0425ca00d5ad4f54b695c5942ece2933"><li>对象只是有限的树；</li></ul><ul class="notion-list notion-list-disc notion-block-6920b6cec5254134a2e6a4f38d1f838b"><li>问题只问一个最长序列；</li></ul><ul class="notion-list notion-list-disc notion-block-3a3c6243c7ee4e9eafe4bfc57f00807d"><li>但“避免嵌入”的全局约束会制造出超乎想象的有限长度。</li></ul><div class="notion-text notion-block-a0c2d30f92cc4b12a074ba76f59827e7">TREE(3) 的巨大，不来自故意堆叠指数，而来自一个自然组合定理背后的序结构。</div><h4 class="notion-h notion-h3 notion-h-indent-1 notion-block-100dbac4d8314998a7bd4688bf47c54a" data-id="100dbac4d8314998a7bd4688bf47c54a"><span><div id="100dbac4d8314998a7bd4688bf47c54a" class="notion-header-anchor"></div><a class="notion-hash-link" href="#100dbac4d8314998a7bd4688bf47c54a" title="TREE(3) 与葛立恒数谁更大？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">TREE(3) 与葛立恒数谁更大？</span></span></h4><div class="notion-text notion-block-6aa67a17040746f8bb98faddf60ee19a">结论非常明确：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-6c01423bd7804d739967578322ad2c04">这里的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 不是严格运算符，只是强调二者差距巨大。把葛立恒数乘方、做幂塔、甚至进行许多层熟悉的快速增长操作，仍不足以逼近 TREE(3) 所处的层级。</div><hr class="notion-hr notion-block-e5e511b4f30a4ca79a86223818c2d4a5"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-a7f7802587a44f44baad06ac24c061a9" data-id="a7f7802587a44f44baad06ac24c061a9"><span><div id="a7f7802587a44f44baad06ac24c061a9" class="notion-header-anchor"></div><a class="notion-hash-link" href="#a7f7802587a44f44baad06ac24c061a9" title="五、SCG(13)：简单图游戏中的又一次爆炸"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">五、SCG(13)：简单图游戏中的又一次爆炸</span></span></h3><div class="notion-text notion-block-66a08bab686a47eab1e4b24c2ca4d8cb">哈维·弗里德曼研究有限图与有限形式系统时提出过若干快速增长函数，其中常被大数爱好者提及的是 <b>SCG(13)</b>。</div><div class="notion-text notion-block-9981512e6f72421485388f575fc4d131">SCG 可粗略理解为某类“带标签子立方图序列”的最大长度问题：每一步只能使用受限规模的有限图，同时要求早先图不能嵌入后来的图。它与 TREE 函数一样，把一个看似朴素的有限组合限制推到极端。</div><div class="notion-text notion-block-cd84e4b8e6a54c18aad772b58475d5d7">SCG(13) 通常被认为远大于 TREE(3)。但这里需要保持严谨：不同资料可能采用略有差异的函数定义与记号约定，因此比较时必须确认双方使用的是同一版本。</div><div class="notion-text notion-block-7f16d966badd4612aaf4112163c33243">它再次说明，大数排行榜不是一条靠多写几个箭头就能统一排列的直线。不同大数往往来自不同的形式系统、序数分析和组合原理；真正的比较需要把它们翻译进共同的增长层级。</div><hr class="notion-hr notion-block-6a889ba857264c68ab2cf0ddef9b1682"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-186ba3b6d9574e1aa2ad2580587755d3" data-id="186ba3b6d9574e1aa2ad2580587755d3"><span><div id="186ba3b6d9574e1aa2ad2580587755d3" class="notion-header-anchor"></div><a class="notion-hash-link" href="#186ba3b6d9574e1aa2ad2580587755d3" title="六、Busy Beaver：增长快到任何算法都追不上"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">六、Busy Beaver：增长快到任何算法都追不上</span></span></h3><div class="notion-text notion-block-4848c474d9ba41f99d28e35c6fca1fd7">前面的数虽然巨大，却仍可由明确递归规则计算，至少原则上如此。<b>忙碌海狸函数</b>则把我们带到可计算性的边界。</div><div class="notion-text notion-block-08eb0462ac8a441a993c7fb2e2338f34">考虑只有 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 个非停机状态的图灵机。让这些机器从空白纸带开始运行，并只考察最终会停机的机器。定义：</div><ul class="notion-list notion-list-disc notion-block-a71e5e1c12584127b7514d561ab02693"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：所有这类停机机器中，停机前写出最多数量的 1；</li></ul><ul class="notion-list notion-list-disc notion-block-138211ba5bb5471facf332a4ce6d78f2"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>：所有这类停机机器中，运行步数最多者的步数。</li></ul><div class="notion-text notion-block-b2f08790b8924fe4aebcae7d819e1d29">因为候选机器数量有限，最大值一定存在。但不存在一个算法能对任意 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 都正确计算 Busy Beaver 函数；否则便可据此解决停机问题。</div><div class="notion-text notion-block-d67610d9aca4451db7c276d8b40ff4ae">这意味着 Busy Beaver 最终会超过<b>每一个可计算函数</b>。无论你设计怎样疯狂的递归、幂塔或箭头系统，只要它仍对应一个可计算函数，Busy Beaver 在足够大的输入处终将把它甩在身后。</div><div class="notion-text notion-block-aa030d5ef4094eb38e0e039a17ce0997">这里必须区分两件事：</div><ul class="notion-list notion-list-disc notion-block-4233e6cc376e429588e58afc995b6462"><li>每个具体的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 都是有限整数；</li></ul><ul class="notion-list notion-list-disc notion-block-a04138c1dcf74692b1d21958aa9577b4"><li>整个函数却不可计算。</li></ul><div class="notion-text notion-block-cc6b1dab2d53454bb4eedd75883c020f">大数由此从“写不下”升级为“没有统一算法算得出”。</div><hr class="notion-hr notion-block-adf44e2c5dc24d87afc8fef456e5c03a"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-e6f5dedc39ff4878af4f1b875e99339d" data-id="e6f5dedc39ff4878af4f1b875e99339d"><span><div id="e6f5dedc39ff4878af4f1b875e99339d" class="notion-header-anchor"></div><a class="notion-hash-link" href="#e6f5dedc39ff4878af4f1b875e99339d" title="七、拉约数：比拼的不是增长，而是语言能表达什么"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">七、拉约数：比拼的不是增长，而是语言能表达什么</span></span></h3><div class="notion-text notion-block-7007f5a2e30344c184bbb6c28353a6a6">2007 年，MIT 举行了一场“大数决斗”。规则不是无限制地说“对方的数字加一”，而是要求参与者用有限符号和规定语言定义一个自然数。</div><div class="notion-text notion-block-31f25161b3aa4f2b844aff709048a35e">阿古斯丁·拉约最终给出的思想，可简化为：</div><blockquote class="notion-quote notion-block-de6311228eee4f388af6811b9edd594a"><div>取一个最小自然数，使它大于所有能够用不超过一个古戈尔个符号、在指定集合论语言中唯一描述的有限自然数。</div></blockquote><div class="notion-text notion-block-b00bb57015404aefb1152852cf74e938">这便产生了通常所称的<b>拉约数</b>。</div><div class="notion-text notion-block-e0bdc0c2f8e7433182662a19f936377e">拉约数的威力来自元语言：它不再逐层构造一个大数，而是把某个形式语言在长度限制内能够描述的所有数字一次性收集起来，然后跳到它们之上。</div><div class="notion-text notion-block-a3e0df0d6a854d258b2a3fb0fbb96494">不过，“拉约数是最大的数”是错误说法。任何有限数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 都有 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。拉约数只是在指定语言、指定符号预算与指定语义规则下，压过所有可在该限制内定义的数。改变语言或增加符号预算，又可以定义更大的数。</div><hr class="notion-hr notion-block-326c4f0c9f1948dcb525da2f69f42d8a"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-d8d47d513de047e8b7c0c98f0819e8b5" data-id="d8d47d513de047e8b7c0c98f0819e8b5"><span><div id="d8d47d513de047e8b7c0c98f0819e8b5" class="notion-header-anchor"></div><a class="notion-hash-link" href="#d8d47d513de047e8b7c0c98f0819e8b5" title="八、为什么不存在最大的有限数？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">八、为什么不存在最大的有限数？</span></span></h3><div class="notion-text notion-block-499701c737214772b3a5c31971f3e954">无论一个数多大，只要它是有限整数，就可以加一：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-78fb95a199f24c199fa8f7699f82887a">因此不存在“最大的自然数”。所谓著名大数，通常属于以下几类：</div><ol start="1" class="notion-list notion-list-numbered notion-block-a776ca73a3824d6786f0ba4ce9bd31e8" style="list-style-type:decimal"><li><b>文化命名型</b>：古戈尔、古戈尔普勒克斯；</li></ol><ol start="2" class="notion-list notion-list-numbered notion-block-4f6edbaa05364482ba9afb56b27c9572" style="list-style-type:decimal"><li><b>记号构造型</b>：幂塔、莫泽数；</li></ol><ol start="3" class="notion-list notion-list-numbered notion-block-f6eebfa51fda4594842a582e90cfca07" style="list-style-type:decimal"><li><b>数学问题上界型</b>：斯奎斯数、葛立恒数；</li></ol><ol start="4" class="notion-list notion-list-numbered notion-block-1fe185791abf44b18d1e1064dd72955a" style="list-style-type:decimal"><li><b>有限组合极值型</b>：TREE(3)、SCG(13)；</li></ol><ol start="5" class="notion-list notion-list-numbered notion-block-2a0a92aab67141e7bfff400f4516f0ed" style="list-style-type:decimal"><li><b>可计算性边界型</b>：Busy Beaver；</li></ol><ol start="6" class="notion-list notion-list-numbered notion-block-dbc666edd4ff4b0080013866690d3b81" style="list-style-type:decimal"><li><b>语言支配型</b>：拉约数。</li></ol><div class="notion-text notion-block-5c2b9afe857348bf8fa8ee9c525d50b6">它们不是按“零的数量”排成一列，而是代表了人类制造和理解大数的不同方法。</div><hr class="notion-hr notion-block-8d701f7fff4b4850aa98941c492bf28b"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-eac7c80713ca46c4ac0b66def43ea1cc" data-id="eac7c80713ca46c4ac0b66def43ea1cc"><span><div id="eac7c80713ca46c4ac0b66def43ea1cc" class="notion-header-anchor"></div><a class="notion-hash-link" href="#eac7c80713ca46c4ac0b66def43ea1cc" title="九、再大的有限数，也没有碰到无限"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">九、再大的有限数，也没有碰到无限</span></span></h3><div class="notion-text notion-block-c4161cd68dc242199364f4f048a2653c">设 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 是古戈尔普勒克斯、葛立恒数、TREE(3)、拉约数，或者任何别的有限整数。集合</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c8250dfb6f104fcf89cae1b8bfb9a05f">仍然只有有限个元素。只要从 1 开始逐个计数，总会在有限步后结束。</div><div class="notion-text notion-block-99fd22ec653747f9b59e8b891359fdce">自然数集合</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-16e07ba057474e28ac4ada3f88d4b11e">则不存在最后一个元素。它不是“数到了某个极大的终点”，而是根本没有终点。</div><div class="notion-text notion-block-5ce8b8cf0580490cb7fdca3539c30cd6">所以：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-87485eeeccc040a0b81fbba181fc45fa">这里并不是说二者与 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 很接近。恰恰相反，有限与无限之间没有一条通过不断变大便能跨越的数值坡道。<b>所有有限整数，无论多大，在最小的无限基数面前都属于同一个有限阵营。</b></div><div class="notion-callout notion-purple_background_co notion-block-c27405381fe24252a23f68b872457d7d"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="♾️">♾️</span></div><div class="notion-callout-text"><div class="notion-text notion-block-81b2ba3946594cf5a30149d0e13f8839">下一篇将正式进入无限：为什么自然数和偶数一样多？为什么实数比自然数更多？为什么无限不止一种大小？以及为什么不存在最大的无限？</div></div></div><hr class="notion-hr notion-block-61634c0fe958407cac3e260023536253"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-9b2889e319de43f3b9ad7a24e9f64c64" data-id="9b2889e319de43f3b9ad7a24e9f64c64"><span><div id="9b2889e319de43f3b9ad7a24e9f64c64" class="notion-header-anchor"></div><a class="notion-hash-link" href="#9b2889e319de43f3b9ad7a24e9f64c64" title="参考资料"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">参考资料</span></span></h3><ul class="notion-list notion-list-disc notion-block-d3041552f3f1457ca3a963256615db49"><li>Agustín Rayo, <a class="notion-link" href="https://web.mit.edu/arayo/www/bignums.html" target="_blank" rel="noopener noreferrer">Big Number Duel</a></li></ul><ul class="notion-list notion-list-disc notion-block-895117ed34f54cb68b7f817c6b1c769f"><li>Stanford Encyclopedia of Philosophy, <a class="notion-link" href="https://plato.stanford.edu/entries/infinity/" target="_blank" rel="noopener noreferrer">Infinity</a></li></ul><ul class="notion-list notion-list-disc notion-block-6d39e56128904b71818f9b7aa0ccd1c1"><li>Harvey Friedman 关于有限组合陈述与巨大有限数的相关研究</li></ul><ul class="notion-list notion-list-disc notion-block-f0a2b7c014cb4976a19c46a1e80a8ab0"><li>Ronald Graham 与 Bruce Rothschild 关于拉姆齐理论及相关上界的研究</li></ul></main></div>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[无限不只是永远数不完：从希尔伯特旅馆到不可数无穷]]></title>
            <link>https://yjy.hauchet.cn/article/infinity-countable-uncountable-aleph</link>
            <guid>https://yjy.hauchet.cn/article/infinity-countable-uncountable-aleph</guid>
            <pubDate>Tue, 04 Aug 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[无限不是一个极大的有限数。从希尔伯特旅馆、康托尔对角线、阿列夫数到连续统假设，理解不同层级的无限及其令人反直觉的运算规则。]]></description>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-3b29f85982ee814b89bfc683dc7cea61"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><div class="notion-callout notion-purple_background_co notion-block-35fcb5a680824c639ab3bbf1615b11ec"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="♾️">♾️</span></div><div class="notion-callout-text"><div class="notion-text notion-block-d5c7c489d2e547e5b8902d0f6d1aa9c8"><b>“有限与无限”系列 · 下篇</b></div><div class="notion-text notion-block-229a95b94072460eab9fc877610c196f">葛立恒数、TREE(3) 与拉约数即使巨大到无法想象，也仍然只是有限整数。无限不是把有限数继续放大，而是一次概念层面的跃迁。</div></div></div><div class="notion-text notion-block-51c2cda96d84424792db037af2b29592">在日常语言中，“无限”常被用来表示“非常多”“看不到尽头”或“永远持续”。数学却要求更精确：无限集合具有怎样的大小？两个无限集合如何比较？一种无限之上是否还有更大的无限？</div><div class="notion-text notion-block-ac68c6fd97f840c38d6bb8f61c818cba">十九世纪以前，许多思想家只愿意把无限看成一个永远不能完成的过程。康托尔则大胆地把无限集合当作可以整体研究的对象，并由此发现：<b>无限不仅存在不同大小，而且没有最大的无限。</b></div><div class="notion-table-of-contents notion-gray notion-block-a189877aae92487e8e5817e045e0a518"><a href="#1e36df6faf08430ea0bde19db1efde91" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">一、无限不是一个写着 ∞ 的超级数字</span></a><a href="#6ab9ecc207b14abebc98d28abf24f269" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">二、伽利略悖论：平方数竟和自然数一样多</span></a><a href="#bc00f843e49a47ee851506b193d8cbf0" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">三、希尔伯特旅馆：满房也能继续住人</span></a><a href="#962a8c49751f4fe1b9c480e6d235efea" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">四、有理数看似密集，却仍然可数</span></a><a href="#44b2da5c2fb64067bd6abce3b1464f93" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">五、康托尔对角线：实数比自然数更多</span></a><a href="#7f5e4bf84d77454596a86e3b91cdc827" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">六、连续统：实数到底有多大？</span></a><a href="#b871ab7ef09746a0820dd1bcf529b221" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">七、阿列夫数：给无限层级编号</span></a><a href="#e375618a6d304bb4952c219e92fe0913" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">八、连续统假设：为什么数学不能简单回答是或否？</span></a><a href="#2b6ad7c2d7bf4b0d95289869aae65ed9" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">九、基数与序数：大小和顺序不是一回事</span></a><a href="#f798440d669e4c688f5ef0b031206ada" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">十、0.999…=1 与无穷小：别把不同无限混为一谈</span></a><a href="#7888a826933e4f3fb54d489c6de54ffe" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">十一、绝大多数实数都无法被描述</span></a><a href="#c4d2ea801be94e70b0f715919dac85f4" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">十二、有限再大，也不会逐渐变成无限</span></a><a href="#9f35343f2271440abdf6f88cb2dad0a8" class="notion-table-of-contents-item"><span class="notion-table-of-contents-item-body" style="display:inline-block;margin-left:0">参考资料</span></a></div><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-1e36df6faf08430ea0bde19db1efde91" data-id="1e36df6faf08430ea0bde19db1efde91"><span><div id="1e36df6faf08430ea0bde19db1efde91" class="notion-header-anchor"></div><a class="notion-hash-link" href="#1e36df6faf08430ea0bde19db1efde91" title="一、无限不是一个写着 ∞ 的超级数字"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">一、无限不是一个写着 ∞ 的超级数字</span></span></h3><div class="notion-text notion-block-199f950362504967a41b8ec07e9e888a">符号 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 在微积分中经常出现，例如</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-16fc8acfb7164f0889df6c306911765c">这里的 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 并不表示 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 最终取到某个叫“无穷大”的实数，而表示：对任意给定的界限，<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 都可以超过它。</div><div class="notion-text notion-block-7786b11762b54cafae4eee7b455d3e08">同样，</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-23bca0e2d392419e9915743d7e9bdaef">表示一个无限序列的部分和趋近于 1，而不是在某一步真的把“第无穷项”加了进去。</div><div class="notion-text notion-block-50fa74bbc8344fe5937bff19e9897bde">数学中的无限至少有两种常见理解：</div><ul class="notion-list notion-list-disc notion-block-0da0010e645b4b9fa9d3ece26f2a831b"><li><b>潜无限</b>：一个过程可以无止境继续，例如不断给自然数加一；</li></ul><ul class="notion-list notion-list-disc notion-block-ec804cd2de8b418281edfd88aae1d45b"><li><b>实无限</b>：把全部自然数看成一个完整集合 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，并研究它的整体性质。</li></ul><div class="notion-text notion-block-b030b9c278ba4a6db148ad1c86c888ea">现代集合论主要研究后者，但潜无限仍是极限、算法与过程直觉的重要来源。</div><hr class="notion-hr notion-block-3c633e2adf08469c9a39d9c594f6459f"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-6ab9ecc207b14abebc98d28abf24f269" data-id="6ab9ecc207b14abebc98d28abf24f269"><span><div id="6ab9ecc207b14abebc98d28abf24f269" class="notion-header-anchor"></div><a class="notion-hash-link" href="#6ab9ecc207b14abebc98d28abf24f269" title="二、伽利略悖论：平方数竟和自然数一样多"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">二、伽利略悖论：平方数竟和自然数一样多</span></span></h3><div class="notion-text notion-block-4f22455bdc4145609530c3ab89573b86">自然数为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-7d098af489054456b74daa89670c4a4b">平方数为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-921f1b6d1ce749e797e9822287b72b08">平方数显然只是自然数的一部分。有限集合中，真子集一定更小；但我们可以建立对应：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-aa7c5f8812244991b2e662b6807bc0e1">每个自然数恰好对应一个平方数，每个平方数也恰好来自一个自然数。因此两者之间存在双射。</div><div class="notion-text notion-block-f253cd527c664615a6026a4fea8c22b2">这说明在无限集合中，<b>真子集可以和整体一样大</b>。</div><div class="notion-text notion-block-ea334f0851f34c539e886f7698536f51">康托尔把“能够建立一一对应”作为集合大小相同的标准。自然数集合的基数记作</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-4f3eb08a54c84916b725826355d31b27">读作“阿列夫零”，它是最小的无限基数。</div><div class="notion-text notion-block-5c38a66bd2604967a7908cb12af9a3c0">偶数集合、奇数集合、整数集合都与自然数集合等势：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-2c9300d73da74f7f8b05ad0e381a4630">这不是说它们作为集合完全相同，而是说它们拥有相同的基数。</div><hr class="notion-hr notion-block-78d93528a4514c0b96daea972133b390"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-bc00f843e49a47ee851506b193d8cbf0" data-id="bc00f843e49a47ee851506b193d8cbf0"><span><div id="bc00f843e49a47ee851506b193d8cbf0" class="notion-header-anchor"></div><a class="notion-hash-link" href="#bc00f843e49a47ee851506b193d8cbf0" title="三、希尔伯特旅馆：满房也能继续住人"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">三、希尔伯特旅馆：满房也能继续住人</span></span></h3><div class="notion-text notion-block-9478a94554f94f5c9cdcb77c0bc67f66">设有一家旅馆，房间编号为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f98e761baa7c43919952f93b329f86b9">每个房间都住着一位客人，因此旅馆已满。这时又来一位新客人。</div><div class="notion-text notion-block-319f018b8de24958a672f0e6beb008be">有限旅馆只能拒绝他；无限旅馆却可以让原来住在 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 号房的客人搬到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 号房。于是 1 号房被空出。</div><div class="notion-text notion-block-c54e0dc50e014a9e9c184208f35dcf06">若一次来了可数无穷多个新客人，则可让原客人从 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 号房搬到 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 号房，把所有奇数房间留给新客人。</div><div class="notion-text notion-block-f282f9fc1c4a45c0b3df495d98cfc41b">希尔伯特旅馆并不是现实建筑方案，而是对可数无限算术的形象展示：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-fa97560e86a14b00b8d349f93b9c0d2f">无限基数的加法和乘法与有限数不同。增加有限个、甚至可数无穷多个元素，可能仍不会改变其基数。</div><hr class="notion-hr notion-block-45c0f25f595d4ab0a8aa504a5c9ff62b"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-962a8c49751f4fe1b9c480e6d235efea" data-id="962a8c49751f4fe1b9c480e6d235efea"><span><div id="962a8c49751f4fe1b9c480e6d235efea" class="notion-header-anchor"></div><a class="notion-hash-link" href="#962a8c49751f4fe1b9c480e6d235efea" title="四、有理数看似密集，却仍然可数"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">四、有理数看似密集，却仍然可数</span></span></h3><div class="notion-text notion-block-de360daf7e294c5d8fd7065da249b73d">有理数集合</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-eb62623e56f846aabbd4ed54cada1e7c">在数轴上非常密集：任意两个不同实数之间都有有理数。直觉上，它似乎应比自然数“多得多”。</div><div class="notion-text notion-block-e2f1e0339a5e4ec483d8de1da46ad7cf">然而可以把所有分数排在二维格点中，再沿对角线依次扫描，跳过重复表示，便能把每个有理数安排到某个自然数编号上。因此</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-67222ee8ea4c4c03bf0d3b26557d9289">可数并不意味着稀疏，也不意味着能在有限时间列完；它只表示存在一个序列</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a669f1f88ac04655a8f74c50cd2344cf">能够使每个元素恰好在某个有限位置出现。</div><div class="notion-text notion-block-c8168515e9b244bd8812f610dd3d68c1">这也是为什么：</div><ul class="notion-list notion-list-disc notion-block-54f94a3935ac4250b65c14165f4d99cd"><li>全体有限长度字符串可数；</li></ul><ul class="notion-list notion-list-disc notion-block-5c6337daa2064550b759a17944ccf13c"><li>全体有限程序可数；</li></ul><ul class="notion-list notion-list-disc notion-block-cb591a12f5ad4a5caef1d1dbaa97f8f4"><li>全体代数数可数；</li></ul><ul class="notion-list notion-list-disc notion-block-4c267d2a35de49f6bf7e57c4a371f839"><li>全体可由有限文本直接描述的对象至多可数。</li></ul><div class="notion-text notion-block-619ffb9ef68b42949cd7f83533d5573b">最后一点将在后文产生耐人寻味的结果：实数不可数，因此绝大多数实数无法被任何有限文字单独命名。</div><hr class="notion-hr notion-block-5a4dd897bc904f1f8111dc46e0f2e983"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-44b2da5c2fb64067bd6abce3b1464f93" data-id="44b2da5c2fb64067bd6abce3b1464f93"><span><div id="44b2da5c2fb64067bd6abce3b1464f93" class="notion-header-anchor"></div><a class="notion-hash-link" href="#44b2da5c2fb64067bd6abce3b1464f93" title="五、康托尔对角线：实数比自然数更多"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">五、康托尔对角线：实数比自然数更多</span></span></h3><div class="notion-text notion-block-ffb23754caa4416b902d36a018008948">考虑区间 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 内的实数。假设它们可以按顺序全部列出：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-5f413d8098534619b431fec3db97d8d9">现在构造一个新实数</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-1d19889fba3c4921886ecf4cc949e3e7">规定 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 与第 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 个数的第 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 位数字不同。例如，若 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 就令 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，若 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 就令 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。这样可以避开 0.4999… 与 0.5000… 的双重表示问题。</div><div class="notion-text notion-block-8b2160b7316a4205bb7323b0f0c8378c">于是：</div><ul class="notion-list notion-list-disc notion-block-aecafe3bb3c5491786c2c030aab0e099"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 与 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 在第 1 位不同；</li></ul><ul class="notion-list notion-list-disc notion-block-688891d47e444036a05c1818620c2bd5"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 与 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 在第 2 位不同；</li></ul><ul class="notion-list notion-list-disc notion-block-85a2712eb2d943a9850ffca4f7ba958f"><li>一般地，<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 与 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 在第 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 位不同。</li></ul><div class="notion-text notion-block-c86d25c4c990447ebd0f026e0cfd2005">所以 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 不在列表中。这与“列表包含全部实数”的假设矛盾。</div><div class="notion-text notion-block-1e382986f973488490fa1d6ea902f176">因此</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-f8f793b633f54841a956f69cf755e5d0">这是数学史上最震撼的结论之一：<b>无限也有大小之分。</b></div><hr class="notion-hr notion-block-a41af57ccd1e4d61b11388a6730e6fbb"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-7f5e4bf84d77454596a86e3b91cdc827" data-id="7f5e4bf84d77454596a86e3b91cdc827"><span><div id="7f5e4bf84d77454596a86e3b91cdc827" class="notion-header-anchor"></div><a class="notion-hash-link" href="#7f5e4bf84d77454596a86e3b91cdc827" title="六、连续统：实数到底有多大？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">六、连续统：实数到底有多大？</span></span></h3><div class="notion-text notion-block-313020472a704db88d0bcad72bbd9cf9">实数集合的基数称为<b>连续统基数</b>，常记作</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-c76d2d0ad3354933a693e9d8ebf021a2">一个实数可以由它的小数位序列编码，而一个二进制序列又可对应自然数集合的一个子集。因此</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-06565768593f40b89d51479f0f079a19">其中 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 表示自然数集合的幂集，即所有自然数子集组成的集合。</div><div class="notion-text notion-block-e3a78c56c43347d1aeb5b6742844fb59">康托尔证明，对任何集合 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，其幂集都严格更大：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d7636422bb4e47ecacaa545da7ad0901">证明再次使用对角思想。假设存在满射 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，定义</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-172eae598c1341538cc2ac89d8ac7fae">若 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，则</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-645610ab1add42db9882ebb7e9365d32">产生矛盾。因此没有任何函数能把 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的元素完整覆盖到所有子集。</div><div class="notion-text notion-block-248e2db5998c422facf0f0e005590399">由此立刻得到无穷阶梯：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-39ebe7ab67e145da850681087fd37368">所以不存在最大的基数，也不存在“最大的无限”。</div><hr class="notion-hr notion-block-a6435d7f717144fba3d6665cd674b62f"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-b871ab7ef09746a0820dd1bcf529b221" data-id="b871ab7ef09746a0820dd1bcf529b221"><span><div id="b871ab7ef09746a0820dd1bcf529b221" class="notion-header-anchor"></div><a class="notion-hash-link" href="#b871ab7ef09746a0820dd1bcf529b221" title="七、阿列夫数：给无限层级编号"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">七、阿列夫数：给无限层级编号</span></span></h3><div class="notion-text notion-block-74384b59573c49c5bf891dc0771de819">最小无限基数是 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。紧接着比它大的最小基数记为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-40d131009f5a4a76b83c14051602cd80">再下一个是</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-3fecdcd41c6140beb4260df5825454ae">如此继续。更一般地，对任意序数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，都可以定义 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。</div><div class="notion-text notion-block-af280f2f7af84a8ba9da551c168b6726">需要注意：</div><ul class="notion-list notion-list-disc notion-block-46eed89c2f104365b7c8003b153832aa"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的定义是“最小的不可数基数”；</li></ul><ul class="notion-list notion-list-disc notion-block-a73de900f6a54cefab264f1057d65a7e"><li><span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 的定义是“自然数幂集的基数”；</li></ul><ul class="notion-list notion-list-disc notion-block-bd9332b8a8b3416a8102f0c1e2d4926d"><li>二者是否相等，并不是定义直接保证的。</li></ul><div class="notion-text notion-block-8f643bb001264ceca8beff5b9bee957c">命题</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-472c46cb552a492499b1fe88f79a102b">就是著名的<b>连续统假设</b>。</div><div class="notion-text notion-block-ae5497bee2ed4434aef712bb706f01d8">它声称：可数无限与实数的无限之间，不存在第三种基数。</div><hr class="notion-hr notion-block-e453eb907a284e02beab1b2699bcd811"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-e375618a6d304bb4952c219e92fe0913" data-id="e375618a6d304bb4952c219e92fe0913"><span><div id="e375618a6d304bb4952c219e92fe0913" class="notion-header-anchor"></div><a class="notion-hash-link" href="#e375618a6d304bb4952c219e92fe0913" title="八、连续统假设：为什么数学不能简单回答是或否？"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">八、连续统假设：为什么数学不能简单回答是或否？</span></span></h3><div class="notion-text notion-block-8ae4e3ab35454bf3b58632b6e29708d9">康托尔提出连续统假设后，许多数学家试图证明或反驳它。希尔伯特在 1900 年把它列为著名的第一问题。</div><div class="notion-text notion-block-89b9305bca73434d829447eb979ce26d">后来发生了极具哲学意味的转折：</div><ul class="notion-list notion-list-disc notion-block-3d56d9836ae04143ba256e4eb56f421f"><li>哥德尔证明，若 ZFC 集合论是一致的，那么 ZFC 加上连续统假设仍然一致；</li></ul><ul class="notion-list notion-list-disc notion-block-fd3e2aa6cdb54623bc56469b9c58fb1f"><li>科恩证明，若 ZFC 是一致的，那么 ZFC 加上连续统假设的否定也仍然一致。</li></ul><div class="notion-text notion-block-17a08ec5436a432ebab2dd76f8aaf89a">因此，在通常的 ZFC 公理体系中，连续统假设既不能被证明，也不能被否证。它是相对于 ZFC <b>独立</b>的。</div><div class="notion-text notion-block-8c0eb7b6e81343fdafe3bb8fb49f7b4b">这不意味着命题没有意义，也不意味着数学家意见随意。它意味着：现有公理没有提供足够信息决定答案。若要决定它，必须加入新的公理原则，并讨论这些原则为什么应被接受。</div><div class="notion-text notion-block-a1164eb207674119a5275fecfe9cb9c2">无限研究由此触及数学基础的核心：</div><blockquote class="notion-quote notion-block-43ce8359b6ee4508a192cc8a6ed0cb9c"><div>数学真理究竟只是某套公理的推论，还是存在一个等待我们发现的唯一集合宇宙？</div></blockquote><div class="notion-text notion-block-1743be6ea2ac4a2794856968cc720397">不同集合论哲学会给出不同侧重，但独立性定理本身是严格的数学结果。</div><hr class="notion-hr notion-block-59da94ce3dec46488996cd6fe5d0cd3f"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-2b6ad7c2d7bf4b0d95289869aae65ed9" data-id="2b6ad7c2d7bf4b0d95289869aae65ed9"><span><div id="2b6ad7c2d7bf4b0d95289869aae65ed9" class="notion-header-anchor"></div><a class="notion-hash-link" href="#2b6ad7c2d7bf4b0d95289869aae65ed9" title="九、基数与序数：大小和顺序不是一回事"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">九、基数与序数：大小和顺序不是一回事</span></span></h3><div class="notion-text notion-block-4af325523bea410381c3a7814e3b490a">基数回答“有多少个”，序数回答“排在什么顺序”。有限情况下二者很容易混在一起，但无限情况下差别巨大。</div><div class="notion-text notion-block-dafb7ba117d045c897b716451c8da9ed">自然数的标准顺序类型记作</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a65a733aa1534749bed42620ecb9a28d">在末尾再接一个元素，得到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-789f3ee9edea4b15b310a121ef3cec95">把两个自然数序列首尾连接，得到</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-794422644cff4325af99ecc2e815a133">这些序数的基数都仍是 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，但顺序结构不同。</div><div class="notion-text notion-block-4b8ac70178eb48cab288b67a1df5caf9">序数加法甚至不满足交换律：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-587070c35c624cacb5bf0505e213cc66">因为在一个新元素之后接上无限序列，整体仍与 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 同序；而</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-73757298c1894de08d716e8fd518f74c">因为它确实拥有一个最后元素。</div><div class="notion-text notion-block-a03a1c57d50a4cd8a96535849fd891d3">所以“无穷加一还是无穷”只有在说明运算类型后才准确：</div><ul class="notion-list notion-list-disc notion-block-94ab6de055644e84b60e69d6c1973fa3"><li>对无限基数，<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>；</li></ul><ul class="notion-list notion-list-disc notion-block-616236b4baf14785ae6453265d7382b2"><li>对序数，<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。</li></ul><hr class="notion-hr notion-block-5be04a9c37dd4b1d8fb22ae7ecb1983e"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-f798440d669e4c688f5ef0b031206ada" data-id="f798440d669e4c688f5ef0b031206ada"><span><div id="f798440d669e4c688f5ef0b031206ada" class="notion-header-anchor"></div><a class="notion-hash-link" href="#f798440d669e4c688f5ef0b031206ada" title="十、0.999…=1 与无穷小：别把不同无限混为一谈"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">十、0.999…=1 与无穷小：别把不同无限混为一谈</span></span></h3><div class="notion-text notion-block-16c0c3db8de44de998b00d7a0325e9ca">无限小数</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d5ceae32ccf740b1a352bba6001da113">表示数列</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-a8c561e29dc74af281bcca5ff5c9baa1">的极限。因为</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-76361005425245e6a79063f70c58eec7">所以</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-d176e812b5ff43fb83c2e12f3c4bc70f">这里没有一个“永远差一点但最终不等于”的神秘数；在实数体系中，两者是同一个数的不同表示。</div><div class="notion-text notion-block-427be408f011456d8f2f117dcde50aa8">而“无穷小”又是另一概念。标准实分析通常通过极限讨论趋于零的变量；非标准分析则可以在扩展数系中严格引入非零无穷小。二者都能严谨，但不能把无穷小、无限基数、发散到无穷大和无限过程当作同一个对象。</div><hr class="notion-hr notion-block-47a5fc480d6a4082894c58855fe0ddb4"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-7888a826933e4f3fb54d489c6de54ffe" data-id="7888a826933e4f3fb54d489c6de54ffe"><span><div id="7888a826933e4f3fb54d489c6de54ffe" class="notion-header-anchor"></div><a class="notion-hash-link" href="#7888a826933e4f3fb54d489c6de54ffe" title="十一、绝大多数实数都无法被描述"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">十一、绝大多数实数都无法被描述</span></span></h3><div class="notion-text notion-block-5075ea334b8145b6be31b7388ab97f41">有限字母表上的所有有限字符串是可数的。因此，人类能够写出的所有有限定义、程序、公式和书籍合在一起，也至多只有可数多个。</div><div class="notion-text notion-block-c2cd0a7fed4c49a18394de185351eded">但实数集合不可数。</div><div class="notion-text notion-block-13391993cd1e45a1a2dd72760292ad97">所以从基数角度看，能够被有限语言唯一描述的实数只占实数中的可数部分；<b>绝大多数实数没有任何有限名称。</b></div><div class="notion-text notion-block-3b73613fcf204d569497a795a056d91b">这句话不表示我们知道某个具体“不可描述实数”然后又描述了它。它是一个整体计数结论：可描述对象的数量不足以覆盖全部实数。</div><div class="notion-text notion-block-07447efec3f74c8bb8583c689751b23b">类似地，可计算实数也是可数的。圆周率 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>、自然常数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>、<span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span> 都可计算，但几乎所有实数都不可计算。</div><div class="notion-text notion-block-b44fe9cf7ff844f492101b2048354e12">这使无限与信息理论发生连接：有限程序可以产生极其复杂的无限序列，但有限程序的总数仍太少，无法覆盖所有无限序列。</div><hr class="notion-hr notion-block-44df245db9e8433e8bfbd58813e71f61"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-c4d2ea801be94e70b0f715919dac85f4" data-id="c4d2ea801be94e70b0f715919dac85f4"><span><div id="c4d2ea801be94e70b0f715919dac85f4" class="notion-header-anchor"></div><a class="notion-hash-link" href="#c4d2ea801be94e70b0f715919dac85f4" title="十二、有限再大，也不会逐渐变成无限"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">十二、有限再大，也不会逐渐变成无限</span></span></h3><div class="notion-text notion-block-7294724f063e4b24bef369ca9f9e9bf4">上一篇介绍了古戈尔、葛立恒数、TREE(3)、Busy Beaver 和拉约数。它们之间差距巨大，但都满足同一个事实：</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-be838cb296d049f692fcb0a1c1fc5851">把一个有限数不断加大，只会得到另一个有限数。不存在某个“最大的有限数”作为通往无限的门槛。</div><div class="notion-text notion-block-aa633807efa14840b42928fe6774c25d">更准确地说，自然数序列</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-48138cc2b37746e3bab9b9ddfdeb6522">的每一项都是有限数，但整个集合的基数是 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>。无限不是序列中某个遥远成员，而是关于<b>整体无终止性与集合对应关系</b>的结构概念。</div><div class="notion-text notion-block-48871b9a715e4e9b89e40e04eb923a83">同样，无限也没有终极顶点。对任意基数 <span role="button" tabindex="0" class="notion-equation notion-equation-inline"><span></span></span>，康托尔定理都给出更大的基数</div><span role="button" tabindex="0" class="notion-equation notion-equation-block"><span></span></span><div class="notion-text notion-block-6d379c40911047dd9f8ab14985bc350b">因此数学中的无限不是一个孤零零的符号，而是一座没有最高层的阶梯。</div><div class="notion-callout notion-blue_background_co notion-block-e1301dc966864be4971714f8bbe3e886"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="✦">✦</span></div><div class="notion-callout-text"><div class="notion-text notion-block-03e9595db891417b89e1627039623bc4"><b>有限大数让我们看到表示法的力量；无限理论则迫使我们重新定义“多少”。</b>
前者问：一个有限公式能压缩多大的整数？后者问：即使无法逐个数完，我们还能否严格比较两个整体的大小？</div></div></div><hr class="notion-hr notion-block-aa6874b7d3f044b491296fa515892ff4"/><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-9f35343f2271440abdf6f88cb2dad0a8" data-id="9f35343f2271440abdf6f88cb2dad0a8"><span><div id="9f35343f2271440abdf6f88cb2dad0a8" class="notion-header-anchor"></div><a class="notion-hash-link" href="#9f35343f2271440abdf6f88cb2dad0a8" title="参考资料"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">参考资料</span></span></h3><ul class="notion-list notion-list-disc notion-block-1f7d66d45f8c4df280f6cec5b69ab266"><li>Stanford Encyclopedia of Philosophy, <a class="notion-link" href="https://plato.stanford.edu/entries/infinity/" target="_blank" rel="noopener noreferrer">Infinity</a></li></ul><ul class="notion-list notion-list-disc notion-block-3cf1c79676514a65be6e62bdedf7ec6e"><li>Stanford Encyclopedia of Philosophy, <a class="notion-link" href="https://plato.stanford.edu/entries/continuum-hypothesis/" target="_blank" rel="noopener noreferrer">The Continuum Hypothesis</a></li></ul><ul class="notion-list notion-list-disc notion-block-2ddbca14c1c5412e99b9792d48453b1c"><li>Georg Cantor 关于集合基数、对角线方法与超限数的经典工作</li></ul><ul class="notion-list notion-list-disc notion-block-bf207fef24c743ae9a4299ce12335cbc"><li>David Hilbert, <em>On the Infinite</em></li></ul></main></div>]]></content:encoded>
        </item>
    </channel>
</rss>