← All papers
First page of No new Goormaghtigh primes up to $10^{700}$

No new Goormaghtigh primes up to $10^{700}$

Jon Grantham

math.NT Sep 19, 2024 · v3
A Lean 4 development verifies the abc-conditional finiteness argument and checks coverage and survivor rejection for the bounded computational search.
The Goormaghtigh conjecture states that the only two numbers which have two non-trivial representations as repunits are $31$ and $8191$. We call such a prime number a {\it Goormaghtigh prime}. We show that there are no other Goormaghtigh primes less than $10^{700}$.

The Goormaghtigh conjecture says 31 and 8191 are the only numbers with two non-trivial repunit representations. Both are prime, so the paper asks whether any other 'Goormaghtigh primes' exist below 10^700.

Known lower bounds on m, n and y from Bennett et al. restrict the cases to prime exponents with m ≥ 53, n ≥ 7 and y > 10^5. A C program then sieves the remaining triples (m,n,x) modulo small primes, using precomputed repunit residue tables, and treats x=2 separately. Assuming the abc conjecture, an argument suggested by Pomerance shows there are finitely many Goormaghtigh numbers with neither representation of length three or four. A Lean 4 formalization verifies this abc argument, the coverage of the certified search intervals, the rejection of every logged survivor, and the x=2 branch.

There are no Goormaghtigh primes below 10^700 other than 31 and 8191. Conditionally on abc, there are only finitely many Goormaghtigh numbers with n ≥ 5. Code, data and the Lean proofs are released publicly.