# Lucas-Lehmer test (proof)

• Jun 21st 2010, 07:15 AM
jamix
Lucas-Lehmer test (proof)
Recall the Lucas Lehmer primality test for Mersenne primes which claims the following:

$2^p - 1$ is prime iff $s_{p-1} \equiv 0 mod{p}$

where $s_1 = 4$ and $s_i \equiv s_{i-1}^2 - 2 mod{p}$.

Does anyone know the details to the proof of this important theorem? I've been trying to work it out on my own, but its tough.
• Jun 21st 2010, 07:16 AM
chiph588@
• Jun 21st 2010, 09:53 AM
jamix
The proof on wiki was much easier to follow than many others I've seen, thanks. Who would have though that one could consider the orders of elements in the integral domain $Z_p + Z_p \cdot \sqrt{3}$ in order to solve this!?

Thanks again.
• Jun 21st 2010, 05:26 PM
jamix
I'm finding that for prime Mersennes $q = 2^p - 1$, one doesn't necessarily need to have $s_0 = 4$ in order for $s_{p-2} \equiv 0 mod(q)$.

Consider the prime $2^7 - 1$ for instance. If $s_0$ is in the following set we have that $s_{p-2} \equiv 0 mod(q)$.

{3,4,10,18,21,27,37,38,43,44,46,49,51,52,54,63,64, 73,75,76,78,81,83,84,89,90,100,106,109,117,123,124 }

I'm guessing there is a probalistic reason for this occurence. Anyone wanna try to work through it with me?