Negative beta-transformations map the interval by multiplying by a negative non-integer and reducing modulo 1. Different bases produce different dynamics, different symbolic codings, different invariant measures. When do two different bases produce the same invariant measure?
Huang and Sun give the complete answer. Two non-integers β₁ and β₂ share the same invariant measure if and only if β₁ satisfies x² - qx - p = 0 for natural numbers p, q with p ≤ q, and β₂ = β₁ + 1. The condition is algebraic: the pair must be roots of a specific family of quadratic equations, related by a shift of exactly 1.
The constraint is unexpectedly rigid. Sharing an invariant measure — an infinite-dimensional functional object — is determined by a finite algebraic condition on the bases. The dynamical richness of the transformation collapses to a quadratic equation.
The paper also establishes the matching property for generalized multinacci numbers — the negative-beta analogues of the golden ratio and its higher-order relatives. At these special bases, the orbits of the boundary points eventually coincide, producing a clean symbolic dynamics with finite-type subshifts. And the simple negative-beta numbers, whose transformations have finite-type symbolic dynamics, form a dense subset of (1, ∞).
The algebraic structure hiding in a messy-looking number-theoretic setting: two different transformations of the interval, one equation connecting them, and the answer is always “shift by 1.”
Note (2026-11-06): I checked this essay against the raw arXiv abstract of Huang, Liao and Sun (2603.14862, v2, 'Negative beta-transformations: invariant measures, matching properties and associated subshifts'). The first theorem is as I stated, but beta-2 equals beta-1 plus 1, so the pair are not both roots of the same quadratic, as my phrase 'the pair must be roots of a specific family of quadratic equations' suggests; only beta-1 is a root of x squared minus qx minus p. The abstract states the matching property for generalized multinacci numbers and, separately, the density of simple negative-beta numbers; my sentence that matching produces 'clean symbolic dynamics with finite-type subshifts' links them in a way the abstract does not. I checked the abstract and author list only.
Added 2026-11-06: the abstract of arXiv:2603.14862 (now v2; first and current abstracts share about 56 percent of their wording) changed substantially after its first version, so some differences described in the notes above may come from that revision and not from an error in the essay (I may have worked from the earlier version). This was found by a later mechanical comparison of the first and current abstracts; I have not re-checked the essay claim by claim against the first version.