Corridor
Let p be an odd positive integer. Suppose q is a positive divisor of
Status: proved elementary sufficient family / necessary counterexample restriction
Date: 2026-08-15
Claim boundary: this does not prove Erdős–Straus. It adds another exact shifted-factor restriction to the hard-prime counterexample sieve.
Theorem
Let p be an odd positive integer. Suppose q is a positive divisor of
with
Put
Then
is a positive integer and
Thus any 3 mod 4 divisor of p+4 gives an Erdős–Straus decomposition.
In particular, any prime factor
solves p.
Proof
Since q=4m-1 and q | p+4,
The right side is divisible by q. Because q is odd, q | mp+1, so v is an integer.
Now
Substitute q=4m-1:
hence
Divide by mpv:
QED.
Corollary — counterexample restriction
If p is a counterexample, then p+4 has no divisor 3 mod 4. Since
for Mordell-hard p≡1 mod4, this implies
Thus a hard-prime counterexample forces p+4 into the multiplicative semigroup generated by primes 1 mod 4.
Quadratic-residue interpretation
For a hard prime p≡1 mod4, any prime q|p+4 with q≡1 mod4 obeys
Therefore
Since p≡1 mod4, quadratic reciprocity gives
So failure of the p+4 rescue means every prime factor of p+4 is a quadratic residue modulo p.
This aligns with FOUR-P-PLUS-ONE-FILTER.md and the earlier shifted-factor restrictions in FAB-HARD-FIRST-FILTERS.md: an increasingly large collection of nearby forms is forced to factor entirely over the quadratic-residue half modulo a hypothetical hard-prime counterexample.
Finite regression signal
On the exact four-filter survivor population among Mordell-hard primes through 10^7, this theorem resolves 856 of the 2,173 survivors.
Applied after FOUR-P-PLUS-ONE-FILTER.md, the first six exact factor restrictions leave 791 finite survivors through 10^7.
These counts are finite evidence only; the theorem above is universal.