← All problems

Maxwell conjecture for point-charge equilibria

153 years old · traced to

Conjecture 1.3 (Maxwell; [11], see also Section 4). The total number of points of equilibrium, all assumed nondegenerate, of any configuration with ll charges in R3\mathbb{R}^{3} never exceeds (l−1)2(l-1)^{2}.

Equivalent formulations
  1. If nn positive point charges have only nondegenerate electrostatic equilibria, can their potential have more than (n−1)2(n-1)^2 critical points?

References
  1. Andrei Gabrielov, Dmitry Novikov and Boris Shapiro, “Mystery of point charges”, arXiv:math-ph/0409009 (2004).

Progress summary

Refreshed
Open · solution reported

A July 2026 paper gives a verified counterexample, showing that five positive charges can produce more equilibria than Maxwell’s proposed bound allows.

Maxwell’s question, discussed in 1873 and posed as an upper-bound problem by M. Morse and S. S. Cairns in 1969, asks whether nn positive point charges can have more than (n−1)2(n-1)^2 nondegenerate critical points. Gabrielov, Novikov, and Shapiro later formulated this as the Maxwell conjecture.

Known results

  • Three equal-magnitude charges: at most 44 nondegenerate equilibria; isolated-equilibrium counts are 00, 22, 33, or 44 (2015).
  • Three positive charges in Rn\mathbb{R}^n: at most 66 nondegenerate equilibria, improving the prior general bound 1212 but not proving the conjectured bound 44 (2026).
  • General bounds were improved by V. Zolotov (2023) and H. Edelsbrunner, C. Fillmore, and G. Oliveira (2026).

July 2026 counterexample

Philip Arathoon, Gavin Ball, and Matthew D. Kvalheim construct five positive charges with at least 2424 nondegenerate critical points, exceeding (5−1)2=16(5-1)^2=16. Three original equilibria persist while a central one bifurcates into 2121; a small perturbation makes all critical points nondegenerate. The arXiv paper reports that GPT-5.6 Sol suggested the idea, while the authors verified and wrote the mathematics.

Current status (as of July 2026): The Maxwell conjecture is disproved by a published arXiv counterexample for n=5n=5; sharper bounds for smaller cases, including whether 44 is optimal for three unequal charges, remain open.

Sources

Solutions 0

No solutions have been posted yet.