← All problems

Potential density of integral points in relative character varieties

21 years old · documented by

Let YY be a smooth complex variety equipped with a smooth projective simple normal crossings compactification Y‾\overline{Y}, and let D=Y‾∖YD=\overline{Y}\setminus Y have components D1,…,DnD_1,\ldots,D_n. Let GG be a Chevalley group over Z\mathbb{Z}, let KK be a number field with ring of integers OK\mathscr{O}_K, and let XG,C‾(Y)X_{G,\underline{C}}(Y) denote the relative character variety of representations of π1(Y)\pi_1(Y) into GG whose local monodromy around each DiD_i has prescribed image CiC_i in (G/ad⁡G)(OK)(G/_{\operatorname{ad}}G)(\mathscr{O}_K), where C‾=(Ci)i=1,…,n\underline{C}=(C_i)_{i=1,\ldots,n}. Fix

C‾∈(G/ad⁡G)(OK)n.\underline{C}\in (G/_{\operatorname{ad}}G)(\mathscr{O}_K)^n.

Potential-density conjecture. Integral points are potentially Zariski-dense in the KK-scheme XG,C‾(Y)KX_{G,\underline{C}}(Y)_K: there exists a finite extension L/KL/K such that the Zariski closure of the OL\mathscr{O}_L-points of XG,C‾(Y)X_{G,\underline{C}}(Y) contains XG,C‾(Y)KX_{G,\underline{C}}(Y)_K.

The conjecture predicts arithmetic largeness of relative character varieties with fixed boundary monodromy, generalizing the paper's evidence from the case of SL⁡2\operatorname{SL}_2 and using a reduction to Riemann surfaces together with mapping-class-group dynamics. Its general validity for Chevalley groups and arbitrary smooth quasi-projective varieties remains open.

References

3 papers in this index state this conjecture (2004–2025). The statement above is taken from the most recent of them; the others are arXiv:2404.17185, arXiv:math/0410558.

Source: Simone Coccia and Daniel Litt, “Density of integral points in the Betti moduli of quasi-projective varieties”, arXiv:2507.00167 (2025).

Progress summary

Refreshed
Open · progress reported

A 2025 theorem settles the rank-two cases, while an unverified 2026 submission claims broader progress for special curve cases; the full conjecture remains open.

The conjecture predicts potential Zariski-density of integral points on relative character varieties for arbitrary Chevalley groups, number fields, smooth quasi-projective varieties, and integral boundary data. Coccia and Litt formulated it and established the first general cases in 2025.

Known results

  • G=SL⁡2,ZG=\operatorname{SL}_{2,\mathbb{Z}} and G=PGL⁡2,ZG=\operatorname{PGL}_{2,\mathbb{Z}}, arbitrary smooth quasi-projective YY and prescribed integral boundary data (Coccia--Litt, 2025).
  • In the curve case, a biquadratic extension of the field generated by boundary traces suffices; some extension is necessary in examples such as x2+y2+z2−xyz=3x^2+y^2+z^2-xyz=3 (Coccia--Litt, 2025).

November 2026 claimed curve-rank extension

A submitted manuscript claims that, for closed curves of genus g≥2g\ge 2, every rank r≥2r\ge 2 admits an integral SL⁡r\operatorname{SL}_r representation with Zariski-dense mapping-class-group orbit, yielding potential density after a finite extension. Its relative implications are limited to special boundary data, and the submission is explicitly unverified; arbitrary Chevalley groups and higher-dimensional varieties remain untreated.

Current status (as of November 2026): The conjecture is proved for SL⁡2\operatorname{SL}_2 and PGL⁡2\operatorname{PGL}_2, while the general Chevalley-group and arbitrary-variety statement remains open; the claimed higher-rank curve extension is unverified.

Sources

Solutions 2

Claimed by OpenAI. Potential Zariski density of integral points on SL_r character varieties of smooth complex curves, after a finite number-field extension. The relative case covers prescribed boundary characteristic polynomials with roots of unity. This is a curve and SL_r subcase with quasi-unipotent boundary data; it does not cover arbitrary groups, higher-dimensional varieties or boundary data.See full solutionHide full solution

Claimed by OpenAI.

The manuscript claims potential Zariski density of integral points for SL_r character varieties of smooth connected complex curves, in every rank, after one finite number-field extension. For the relative problem it covers prescribed boundary characteristic polynomials whose roots are roots of unity. Such fixed adjoint-quotient boundary invariants define a closed relative fiber, so the manuscript's ambient integral points satisfying them are integral points of that fiber. This addresses the SL_r curve subcase with quasi-unipotent boundary data; arbitrary Chevalley groups, higher-dimensional varieties and arbitrary integral boundary data are outside the stated claim.

GitHub repository: https://github.com/openai/math

Manuscript: https://github.com/openai/math/blob/adc7f1241b42e322a6451854ab7e4b4c146bf78a/preprints/Integral-points-on-character-varieties-of-curves-September-25-2026/paper.pdf

  • OpenAI-027-01-Integral-points-on-character-varieties-of-curves.pdf466,173 bytesOpen
Conditional partial progress for closed curves, G=SL_r: an integral character with Zariski-dense mapping-class orbit in every rank r≥2, assuming the paper’s pinned source inputs.See full solutionHide full solution

Conditional scope. For a closed oriented surface of genus g≥2 and r≥2, the deduction constructs an SL_r(O_L) representation whose character has Zariski-dense mapping-class orbit in the entire complex character variety. Mapping classes preserve O_L-valued representations; hence that orbit gives Zariski-dense integral points after a finite number-field extension. This treats closed curves and SL_r with no prescribed boundary monodromy. It does not settle the general Chevalley-group or arbitrary quasi-projective case. Rank two is credited to Coccia–Litt and antecedents.

Paper. One integral mapping-class orbit in every rank. Supporting proof and verification archive.

Source assumptions and verification limits. The identified OpenAI source inputs at immutable revision fd4aeeb2ee4fc729c18d98444fed42fd0529eeeb are stipulated true, as explicitly stated in the preprint. This entry preserves that assumption and is labeled partial progress, not a complete unconditional solution. Lean checks scoped supporting results only, not the full theorem. AI-assisted preprint; internal review is not external peer review, MathDB endorsement or certification of worldwide priority.