arXiv:2611.04009
Tight Quasiprobabilistic Entropic Uncertainty Relations for Discrete Quantum Systems
Jonathan J. Thio, Stephan De Bièvre, Crispin H. W. Barnes, David R. M. Arvidsson-Shukur
Abstract
It was conjectured by Wehrl and proven by Lieb that the Shannon differential entropy of the Husimi function of a continuous-variable quantum state is lower bounded by the value it takes on coherent states. This corroborates the intuition, familiar from Heisenberg's uncertainty relations, that coherent states are minimal uncertainty states. The analogous statement for the Wigner function would be that the Wigner differential entropy of states with a positive Wigner function is bounded below by the entropy of pure Gaussian states. This was conjectured a decade ago, but recently disproven. Here, we prove the analogous statement for Kirkwood-Dirac distributions with mutually unbiased bases and for the Gross-Wigner distributions. For these discrete systems, the entropy-minimizing role is played by the pure states with positive quasiprobability distributions. Further, we prove that Kirkwood-Dirac distributions with large nonpositivity have large uncertainty.
AI Summary
Does positivity of a discrete quasiprobability distribution make pure positive states the minimum-uncertainty states, despite the existence of mixed extremal positive states? The paper answers this for Kirkwood–Dirac (KD) distributions defined by mutually unbiased bases and Gross–Wigner (GW) distributions for odd-dimensional qudit systems, contrasting with the recently disproven continuous-variable Wigner–Shannon conjecture. Writing $d$ for the Hilbert-space dimension, the main result is $H(A)+H(B)\geq H^{\mathrm{MH}}(\rho)\geq\ln d-\ln\operatorname{Tr}(\rho^2)\geq\ln d$, whenever the real part of the quasiprobability distribution—the Margenau–Hill distribution—is nonnegative. The minimum $H^{\mathrm{MH}}=\ln d$ occurs exactly at pure quasiprobability-positive states. All pure KD-positive states also satisfy $H(A)+H(B)=\ln d$; pure GW-positive states need not, although their minimum marginal-entropy sum is $\ln d$. The proof combines Shannon-versus-collision entropy, a self-dual-frame identity relating squared quasiprobabilities to purity, and known characterizations of pure positive states as uniform distributions on supports of size $d$. It avoids characterizing mixed extremal states. A complementary bound connects KD mana to measurement uncertainty: $\mathcal M(\rho)\leq[H_{1/2}(A)+H_{1/2}(B)-\ln d]/2\leq\tfrac12\ln d$, with the first inequality saturated for pure states. The conclusions depend on the representation. A non-MUB KD example has a rank-two positive state with entropy approximately $1.033$, below the pure-positive minimum of approximately $1.054$. Saturation at fixed purity is characterized only partially, and the continuous-variable analogue does not share the discrete minimizer result.
Error found: The Discussion incorrectly says Eq. (graal2) generically fails for non-MUB KD representations, whereas the appendix establishes that it holds for the probability-one class whose KD-positive states form the convex hull of the basis states. The continuous-variable appendix incorrectly claims that only coherent states saturate the position–momentum Shannon uncertainty bound; axis-aligned squeezed pure Gaussian states also saturate it.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 5 + (4) pages