← All problems

Signed BAR conjecture

15 years old · traced to

Open problem 1. Prove Proposition 1 with $\pi,\nu_1,\ldots,\nu_d$ are finite measures replaced by $\pi,\nu_1,\ldots,\nu_d$ are finite signed measures. To resolve this open problem it suffices to show that solutions $\left(\pi,\nu_1,\ldots,\nu_d\right)$ of (2.6) cannot change sign, since one can then invoke Proposition 1 to conclude that $\pi$ must be equal to the stationary distribution up to a multiplicative constant.

References

Progress summary

Refreshed
Open · solution reported

A 2026 preprint claims the conjecture is proved for the standard class of reflected diffusions, while also giving counterexamples beyond that class, but the claims lack independent confirmation.

The problem asks whether finite signed basic adjoint relations determine the invariant signed measure uniquely, and how far this extends beyond the Harrison–Reiman class. The supplied sources do not identify a proposer or original posing date.

Recent 2026 preprint

The paper claims that, for stable Harrison–Reiman data with a nonsingular MM-matrix reflection matrix, every finite signed BAR tuple satisfies

πˉ=cπ0,νˉi=cνi0,\bar{\pi}=c\pi_{0},\qquad \bar{\nu}_{i}=c\nu_{i}^{0},

so the tuple space is one-dimensional. It also claims uniqueness fails in a larger completely-S\mathcal{S} class when a proper principal block is singular, producing nonzero zero-mass tuples and an infinite-dimensional family of interior coordinates. The proof was reportedly discovered with assistance from ChatGPT 5.5 Pro and verified by the authors.

Current status (as of August 2026): a preprint claims resolution in the nonsingular MM-matrix Harrison–Reiman class and failure beyond it, but these mathematical claims remain unverified independently.

Sources

Solutions 0

No solutions have been posted yet.