As early attempts to the proof of Fermat’s last theorem, many mathematicians solved the problem for small exponents. While these special cases are being studied, Sophie Germain, a French mathematician, came up with the following interesting result. (Look at https://www.agnesscott.edu/lriddle/women/germain.htm for her fascinating and revolutionary story)
Theorem 1: For any odd prime such that is a prime, the equation has no solution in integers with .
Proof : We prove by contradiction. Suppose there exists a solution with . We can assume to be relatively prime for if they are common factors, they can be canceled out to get a primitive triple. Now, using the factorisation and the fact that , we deduce that is a perfect power–Because if there was a prime factor common to both and , then so that –which is not possible for if
(1) divides it has to divide which can’t be true, and if
(2) divides , has to be and it has to divide and hence , which can’t be true.
So we have for some integer Similarly we have and Now the fact that for every integer , implies that divides one of Lets us say it divides (Similar arguments work for other cases also). Therefore Using the above fact once again implies that divides one of . It can’t divide for if it divides them it has to divides or along with which is not possible. Other case can also be dealt using same ideas above. Thus we will have a contradiction.
Now the fact that the first case of FLT is true for the case follows from the above theorem by noting that is a prime. So this theorem proves the first case for a class of primes for which is also a prime. However it is still not known whether there exist infinitely many such primes (now called Sophie Germain primes, by Hardy-Littlewood Heuristics there are about Sophie Germain primes less than ). They are further generalisations of this result about which I will talk sometime later!
Theorem 2 (Stronger):
If is an odd prime and there exists an “auxiliary prime” satisfying: a) there are no consecutive power residues and (b) is not a power residue
then in any solution to the Fermat equation must divide one of , or
The consecutive power residue condition can be rephrased equivalently as “.
Theorem 1 follows from Theorem 2 because the auxilary prime satisfies the conditions of Theorem 2.
Second Condition: and hence there is no solution to .
First Condition: Also because , we have which cannot be
For , the only auxilary primes are , every other doesn’t satisfy the consecutive power resides condition (I). Even for , the condition is false for if $n$ is a multiple of
The initial hope was that for every prime , one can find infinitely many auxiliary primes which implies that at least one of the elements has to be divisible by infinitely many of these auxiliary primes (using reformulation of the second condition given above), which implies FLT for But for any prime , there are at most finitely many auxiliary primes and the strategy fails!
The proof of Theorem 2 is exactly similar to that of Theorem 1 if we just ask for divisibility of by . The divisibility by needs some slightly more analysis.
Sophie Germain and Legendre together founds auxiliary primes for with and hence solved case I of FLT for all these primes!
Some other methods to establish Case I of FLT is to show that a prime which fails Case I satisfies Wieferich prime conditions: divides . That is the Fermat quotient is divisible by . That allows us to computationally check. In fact, can be shown to divide $a^{p-1}-1$ for ,– so all these Wieferich conditions help you to computationally check that FLT case I has to be true for For instance primes less than that divide are only and primes that divide are only Provin such criterion for bases allow to verify the range
(Now that FLT is proved in general, these ideas might not seem that relevant but still they are interesting!)