I would like to ask whether it would be appropriate to add a short comment and link for Problem #475 pointing to a finite-certificate verification workspace for Graham’s rearrangement problem.
Main repository: https://github.com/Kajin-0/erdos-475
Concise review packet: https://github.com/Kajin-0/erdos-475/blob/main/docs/EXTERNAL_REVIEW_PACKET.md
This workspace focuses on building and verifying finite certificates, not announcing a complete proof. It uses the complement representation B = \mathbb{F}_p^* \setminus A. For each finite complement case, it records a canonical complement representative modulo nonzero scaling, an explicit ordering of A, checks that all non-empty partial sums are distinct modulo p, runs independent Python and Rust verifiers, stores a MANIFEST.sha256 to lock the hash, audits coverage over declared domains, and includes infrastructure to track analytic residue.
The repository currently holds a hash-locked verification package for 136 375 canonical complement instances. The strongly verified finite domains are p=17, |B|=3; p=19, |B|\in{3,4,5}; p=23, |B|\in{3,\dots,9}; p=29, |B|\in{3,\dots,7}; and p=31, |B|\in{3,\dots,6}.
This project does not constitute a full proof of Erdős 475 and is not a request to mark the problem solved. It is a finite-certificate and reduction-ledger workspace. Any theorem-level conclusion would require independently checkable witness artifacts and a proof that the remaining analytic cases lie within the verified finite domain. Some domains still have lower-trust evidence awaiting external artifacts, which are not intended as release-grade certificates. Some documentation and code were produced with AI assistance, but the certificate verification itself uses direct computations and independent checkers.
I would appreciate guidance on whether a comment linking this workspace is appropriate, or whether this belongs solely on the problem page.
I would like to ask whether it would be appropriate to add a short comment and link for Problem #475 pointing to a finite-certificate verification workspace for Graham’s rearrangement problem.
Main repository: https://github.com/Kajin-0/erdos-475
Concise review packet: https://github.com/Kajin-0/erdos-475/blob/main/docs/EXTERNAL_REVIEW_PACKET.md
This workspace focuses on building and verifying finite certificates, not announcing a complete proof. It uses the complement representation B = \mathbb{F}_p^* \setminus A. For each finite complement case, it records a canonical complement representative modulo nonzero scaling, an explicit ordering of A, checks that all non-empty partial sums are distinct modulo p, runs independent Python and Rust verifiers, stores a MANIFEST.sha256 to lock the hash, audits coverage over declared domains, and includes infrastructure to track analytic residue.
The repository currently holds a hash-locked verification package for 136 375 canonical complement instances. The strongly verified finite domains are p=17, |B|=3; p=19, |B|\in{3,4,5}; p=23, |B|\in{3,\dots,9}; p=29, |B|\in{3,\dots,7}; and p=31, |B|\in{3,\dots,6}.
This project does not constitute a full proof of Erdős 475 and is not a request to mark the problem solved. It is a finite-certificate and reduction-ledger workspace. Any theorem-level conclusion would require independently checkable witness artifacts and a proof that the remaining analytic cases lie within the verified finite domain. Some domains still have lower-trust evidence awaiting external artifacts, which are not intended as release-grade certificates. Some documentation and code were produced with AI assistance, but the certificate verification itself uses direct computations and independent checkers.
I would appreciate guidance on whether a comment linking this workspace is appropriate, or whether this belongs solely on the problem page.