第 1 讲 · 47 页课件
测度与积分
Measure and Integration for Probability
先弄清「哪些集合可以谈概率」(σ-代数),再把「长度」推广成测度,最后建立勒贝格积分,并学会用 MCT、Fatou 引理、DCT 判断极限和积分能不能交换。
精讲已完成 47 / 47 页 · 已看懂 0 / 47 页
开篇
第 1 部分 概率空间与可测事件Probability spaces and measurable events
- 04 路线图:进入第 1 部分Where we are
- 05 概率模型从「所有可能的世界」开始A probabilistic model starts with complete possible worlds
- 06 事件:把满足某个性质的结果收集起来Events collect outcomes for which a property holds
- 07 板书续页:无穷次抛硬币、上极限与下极限Handwritten page: infinite coin tosses, limsup and liminf
- 08 概率的分配必须尊重可数拆分(可数可加性)Probability weights must respect countable decompositions
- 09 σ-代数:对可数次「或、且、非」封闭的事件族A sigma-algebra is closed under countable event logic
- 10 为什么不总是取全部子集 2^Ω?Why not always choose 2^Ω?
- 11 生成的 σ-代数:只添加「被迫加入」的集合The generated sigma-algebra adds exactly what is forced
- 12 测度空间:把「事件范围」和「重量」分开Measure spaces separate the event domain from the weights
- 13 测度的从下连续性Continuity from below
- 14 从上连续性,以及它必需的有限条件Continuity from above and its necessary hypothesis
第 2 部分 从乔丹容量到外测度From Jordan content to outer measure
第 3 部分 勒贝格可测集与勒贝格测度Lebesgue measurable sets and measure
- 19 路线图:进入第 3 部分Where we are
- 20 卡拉泰奥多里判别:任意切一刀,总成本不变The Carathéodory test asks whether every cut preserves cost
- 21 维塔利定理:长度无法定义在所有子集上Vitali theorem: length cannot live on every subset
- 22 可数可加性导出矛盾(维塔利定理的证明)Countable additivity forces the contradiction
- 23 构造完成:勒贝格测度正好满足概率的需要The construction delivers exactly what probability needs
第 4 部分 勒贝格积分Lebesgue integration
- 24 路线图:进入第 4 部分Where we are
- 25 达布和:把黎曼可积的判别写清楚Darboux sums make the Riemann criterion explicit
- 26 勒贝格保留「台阶」,但换了台阶的底Lebesgue keeps the staircases but changes their base sets
- 27 有界且黎曼可积 ⇒ 勒贝格可积,积分值相同Bounded Riemann integrability implies Lebesgue integrability
- 28 反过来不成立:狄利克雷函数Lebesgue integrability does not imply Riemann integrability
- 29 可测函数:积分的「合法输入」Measurable functions are the inputs to the integral
- 30 示性函数与简单函数:积分的「原子」Indicators are the atomic measurable functions
- 31 简单函数的积分与写法无关The simple integral is representation independent
- 32 简单函数积分的线性与单调性Linearity and monotonicity for simple functions
- 33 从下方逼近:把积分推广到非负可测函数Extend the simple integral by lower approximation
- 34 只要可测,就能用递增的简单函数逼近Measurability alone gives an increasing simple approximation
第 5 部分 MCT、Fatou 引理、DCT、BCTMCT, Fatou's Lemma, DCT, BCT
- 35 路线图:进入第 5 部分Where we are
- 36 单调收敛定理(MCT)Monotone convergence theorem
- 37 板书续页:单调收敛定理的证明(续)Handwritten page: proof of MCT, continued
- 38 用 MCT 把可加性推广到所有非负可测函数MCT extends simple-function additivity to all f, g ≥ 0
- 39 只有逐点收敛时,「质量」会丢失Pointwise convergence alone does not preserve mass
- 40 尾部下确界找回单调性:liminf 与 limsupTail infima recover the monotonicity needed by MCT
- 41 Fatou 引理:积分的下极限,至少是下极限的积分Fatou's lemma
- 42 板书续页:Fatou 引理的应用——单调函数导数的积分Handwritten page: an application of Fatou's lemma
- 43 在 L¹ 里:零测集上改值、线性组合都无害On L¹, null changes and linear combinations are harmless
- 44 控制收敛定理(DCT):有可积的「屋顶」,极限和积分就能交换Dominated convergence theorem
- 45 板书续页:控制收敛定理的证明(收尾)与使用方法Handwritten page: proof of DCT, continued
- 46 有界收敛定理(BCT):有限测度 + 一致有界,极限就能换积分Bounded convergence is a finite measure corollary
- 47 板书总结:一页看懂整讲Handwritten page: the whole topic on one page