陶哲轩博客写数学问题时，通常先把对象定义清楚，再给直觉、反例和证明轮廓。把「The Littlewood-Offord problem in high dimensions and a conjecture of Frankl and Füredi」改写成可阅读的中文笔记，重点是：问题在问什么、已知到哪一步、下一步最容易走偏在哪。原站广告、分享条和导航已去掉。

问题在问什么

Van Vu and I have just uploaded to the arXiv our joint paper “ The Littlewood-Offord problem in high dimensions and a conjecture of Frankl and Füredi “. In this short paper we give a different proof of a high-dimensional Littlewood-Offord result of Frankl and Füredi , and in the process also affirmatively answer one of their open problems.

Let be vectors in , which we normalise to all have length at least . For any given radius , we consider the small ball probability

已知结果和反例

where are iid Bernoulli signs (i.e. they take values or independently with a probability of of each), and ranges over all (closed) balls of radius . The Littlewood-Offord problem is to compute the quantity

where range over all vectors in of length at least one. Informally, this number measures the extent to which a random walk of length (with all steps of size at least one) can concentrate into a ball of radius .

证明或构造的主线

The one-dimensional case of this problem was answered by Erdös . First, one observes that one can normalise all the to be at least (as opposed to being at most ). In the model case when , he made the following simple observation: if a random sum fell into a ball of radius (which in the one-dimensional case, is an interval of length less than ), and one then changed one or more of the signs from to , then the new sum must necessarily lie outside of the ball. In other words, fo

A similar argument works for higher values of , using Dilworth’s theorem instead of Sperner’s theorem, and gives the exact value

阅读时建议盯住的点

whenever and for some natural number , where are the largest binomial coefficients of .

Now consider the higher-dimensional problem. One has the obvious bound

阅读和落地时建议先做的 5 件事

1. 用自己的语言重写定义和结论，不看原文能不能说清对象是什么。
2. 找一个最小反例或边界情形，确认假设少一条会怎样。
3. 把证明拆成可独立检验的引理，每步只保留一个新想法。
4. 若涉及计算或形式化，先写可复现的小例子，再谈一般情形。
5. 记下尚未解决的缺口：缺估计、缺构造，还是缺正确的范畴。

和智能体、形式化工具怎么接

龙虾PRO做 OpenClaw 落地时，数学笔记最有用的部分往往是「可检验的步骤」：定义、反例、引理边界。智能体适合帮忙展开计算和检索，不适合代替你决定哪条假设能扔。

本文侧重全链路风控方法论。落地时请用自身业务单据做回放验证，不要把示例阈值直接当生产策略。 相关：风控体检 · 方案资源