Weak Simplex Conjecture

For every integer n1n\ge 1, energy E>0E>0, and noise standard deviation σ>0\sigma>0, let x0,,xnRnx_0,\ldots,x_n\in\mathbb{R}^n satisfy xi2=E\lVert x_i\rVert^2=E for all ii, and let the signals have equal prior probabilities and be transmitted over an additive white Gaussian noise channel with maximum-likelihood decoding. If Pc(x0,,xn;σ)P_c(x_0,\ldots,x_n;\sigma) denotes the probability of correct decoding, then PcP_c is maximized by the vertices of a regular simplex centered at the origin, characterized by xi,xj=E/n\langle x_i,x_j\rangle=-E/n for all iji\ne j. Moreover, equality holds only when the signal set is a regular simplex, up to a permutation of the vertices and an orthogonal transformation of Rn\mathbb{R}^n.

Progress summary

Solved

A new preprint claims to have proved that the regular simplex is the unique best arrangement, but independent mathematical verification is not yet available.

The conjecture asserts that, for n+1n+1 signals in nn-dimensional Euclidean space under equal energy limits, the regular simplex uniquely minimizes Gaussian decoding error. Its history runs from Shannon’s suspicion through the formulations of Cover (1987) and Massey (1988); the stronger version was disproved by Steiner (1994).

Known results

  • Balakrishnan gave an early partial study.
  • Dunbridge proved low-noise asymptotic optimality and local optimality at every noise level.
  • Landau and Slepian announced a proof in 1966, later found valid only for n=3n=3.
  • Steiner disproved the Strong Simplex Conjecture in 1994, leaving the weak version unresolved.

August 2026 equality-case claim

The preprint posted on August 19, 2026 claims to classify all equality cases, proving uniqueness of the regular simplex, including under an energy budget. It also reports a Lean 4 machine-check of the single-parameter rigidity argument. This is a claimed advance rather than independently verified resolution; a separate preprint had proved optimal values but explicitly left uniqueness open.

Current status (as of August 2026): the optimal-value statement has a preprint proof, while uniqueness is claimed by the August 2026 preprint but remains unverified.

Sources
Sources & referencesView supporting material

Primary source

arXiv

Additional references

Solutions 0

No solutions have been posted yet.