arXiv:2610.09373
Particle-Number-Preserving Quantum Signal Processing for Thermal Properties: Canonical-Ensemble Estimation and Grand-Canonical Reconstruction
Taehee Ko, Sungbin Lim, Sangkook Choi
Abstract
We develop an end-to-end framework for estimating canonical thermal properties of particle-number-conserving second-quantized Hamiltonians using quantum simulations. The framework integrates initial state preparation with mixed Dicke states, particle-number-preserving Trotterized simulation, and generalized quantum signal processing to estimate canonical partition functions and observables. Building on this canonical scheme, we introduce a grand-canonical reconstruction scheme in which the same quantum measurement data can be reused for arbitrary chemical potentials through classical reweighting, thereby avoiding repeated quantum circuit executions as the chemical potential is varied. We prove the existence of system-size-independent temperature regimes in which both the canonical and reconstructed grand-canonical schemes have lower asymptotic time complexity than classical Lanczos methods, up to polynomial factors, for physically relevant Hamiltonians. We numerically validate our theoretical results for the two-dimensional Hubbard model in both weak- and intermediate-coupling regimes.
AI Summary
The paper addresses canonical thermal estimation for particle-number-conserving fermionic Hamiltonians and reconstruction of grand-canonical properties across chemical potentials without repeating quantum measurements. It combines maximally mixed fixed-number states, prepared through Dicke-state purification, with number-preserving first-order Trotter simulation and generalized quantum signal processing approximating $e^{-\beta H/2}$. For sector dimension $d_s$, partition function $Z_s$, and normalization $\|H_s/\lambda_s\|\leq1-\delta$, partition-function estimation to relative error $\epsilon$ and failure probability $\eta$ requires $O(d_s e^{\beta\lambda_s}Z_s^{-1}\epsilon^{-2}\log(1/\eta))$ circuit samples. Observable estimation carries an additional squared Pauli-coefficient-sum factor. Per-sample gate complexity is polynomial in system size and the stated parameters. An effective-Hamiltonian analysis controls thermal errors without requiring Trotter precision proportional to the potentially exponentially small postselection probability. Estimating all number sectors enables classical reweighting by $e^{\beta\mu n_s}$ for arbitrary $\mu$. Uniform sector-relative errors yield the same relative partition-function error and $O(\epsilon\|O\|)$ observable error. Jensen’s inequality establishes system-size-independent temperature regimes with cost $\widetilde O(d^\gamma)$, $\gamma<0.99$, versus a dimension-linear Lanczos baseline when spectral norms grow linearly and the relevant sector dimensions grow exponentially. Numerical tests use Hubbard systems on periodic $2\times L_y$ geometries up to $L_y=8$, with $U=4,8$ and $t'=-0.2$. Estimated sampling exponents remain below roughly $0.9$ at $T=0.05,0.1,0.5$, but exceed one at $T=0.01$. These comparisons omit polynomial overheads and use estimated spectral bounds. The paper explicitly distinguishes outperforming Lanczos from establishing classical hardness; its general sample bounds can remain exponential.
Error found: Theorem 1's absolute observable-error guarantee lacks the observable-dependent Trotter accuracy needed for its allowed polynomial Pauli sums. The printed numerical formula for gamma_2 contains L instead of L log 4, incorrectly giving approximately -0.279 at infinite temperature. The numerical low-temperature advantage estimates depend on spectral bounds whose validity is not certified by the imposed margins around Lanczos estimates. The Hubbard simulations increase only the length of width-two ladders and therefore do not establish the reported scaling in a genuinely two-dimensional thermodynamic limit.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.