An Erdős problem on random subset sums in finite abelian groups

Jie Ma, Quanyu Tang

A repaired GPT-5.2 Pro argument uses prime cyclic groups to refute f(N)≤log₂N+o(log log N) for the random subset-sum covering threshold. Ma and Tang later prove the stronger lower bound f(p)≥log₂p+(1/(2 log 2)+o(1))log log p; the constant and quantitative refinement are their work.

Sources for this record

What was already known

Erdős and Rényi proved an upper bound of log₂N+O(log log N). Erdős and Hall ruled out an o(log log log N) error but left the wider o(log log N) proposal open.

AI involvement AI-autonomous

The authors attribute the qualitative proof and its repair to GPT-5.2 Pro after human task setting and a request to fix the gap. Tang checked and rewrote the argument; Ma and Tang developed the quantitative strengthening.

The v1 paper explicitly attributes the original qualitative proof and its repair to GPT-5.2 Pro, with no human mathematical step beyond posing the task and requesting a fix. This label does not extend to the human quantitative refinement.

Importance Notable

The qualitative model-origin disproof settles Erdős problem 543; the sharper constant belongs to Ma and Tang.

Limitations

The first model proof had a serious gap that was identified and repaired in a second prompted session. Original chat links expired; attribution rests on the authors’ process account and Tang’s corrected note.

Technical verification Formal (human)

Tang's corrected note reports a step-by-step human repair and check of the qualitative proof, and Ma and Tang report independent verification of the final paper's statements and proofs. This is a source-reported completed human proof check, not a check reproduced here.

Formal checking

Step-by-step correction and independent author proof verification

Checker
Quanyu Tang; Jie Ma and Quanyu Tang
Scope
Qualitative prime-cyclic counterexample to the o(log log N) bound, as retained after correction
Basis
Reported by the linked checker
Evidence source

Checking a formal statement does not establish that it matches the original problem.

Human understanding Full

Tang's corrected note and Ma–Tang's proof overview explain the missed-subset-sum second-moment mechanism and distinguish it from their stronger quantitative argument.

Correction and publication history

2026-10-18 · publication

The qualitative model-origin disproof settles Erdős problem 543; the sharper constant belongs to Ma and Tang.

Read this published revision