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 Lucas-Lehmer test for Mersenne primes」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

In the last few weeks, the Great Internet Mersenne Prime Search (GIMPS) announced the discovery of two new Mersenne primes , both over ten million digits in length, including one discovered by the computing team right here at UCLA (see this page for more information). [I was not involved in this computing effort.] As for the question “Why do we want to find such big primes anyway?”, see this page , though this is not the focus of my post today.

The GIMPS approach to finding Mersenne primes relies of course on modern computing power, parallelisation, and efficient programming, but the number-theoretic heart of it – aside from some basic optimisation tricks such as fast multiplication and preliminary sieving to eliminate some obviously non-prime Mersenne number candidates – is the Lucas-Lehmer primality test for Mersenne numbers, which is much faster for this special type of number than any known general-purpose (dete

已知结果和反例

We begin with a general discussion of how to tell when a given number n (which is not necessarily of the Mersenne form ) is prime or not. One should think of n as being moderately large, e.g. (which is broadly the size of the Mersenne primes discovered recently).

Our starting point will be Lagrange’s theorem , which asserts that

证明或构造的主线

for any finite group G and any , thus the order of a in G divides |G|. Specialising this to the multiplicative group of a finite field of prime order p, we obtain Fermat’s little theorem

for prime and coprime to ; applying it instead to the multiplicative group of a cyclic group of order , we obtain Euler’s theorem

阅读时建议盯住的点

whenever is coprime to n, where is the Euler totient function of n.

Fermat’s little theorem (2) already gives a necessary condition for the primality of a candidate prime : take any coprime to (typically one picks a small number such as or ), and compute modulo . If it is not equal to 1, then cannot be prime. This is a (barely) feasible test to execute for as large as , because one can compute exponents such as relatively quickly, by the trick of repeatedly squaring a modulo n to obtain , and then decomposing into binary to compute . (If is a

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