arXiv:2609.33389
Temperature--Hamiltonian Ambiguity in Strong-Coupling Quantum Equilibrium
Jinshuang Jin
Abstract
Strong coupling generally drives a subsystem away from the Gibbs state of its bare Hamiltonian at the bath temperature. We formulate a temperature--Hamiltonian ambiguity: the reduced equilibrium state fixes the dimensionless Gibbs generator, but does not by itself uniquely determine the local Hamiltonian and temperature entering a Gibbs representation. We propose an operational resolution in which the local Hamiltonian is identified from physical information independent of the stationary Gibbs parametrization. Whenever the exact reduced state is compatible with the Gibbs family generated by that independently identified Hamiltonian, it then determines a unique local temperature. An exactly solvable two-level-system--oscillator model makes the ambiguity explicit through state-equivalent descriptions based on temperature renormalization relative to a specified local Hamiltonian and Hamiltonian-of-mean-force level renormalization at the bath temperature. An exact bosonic model provides the complementary case in which retarded dynamics independently fixes a renormalized local excitation scale and thereby specifies the local Hamiltonian before the stationary populations determine the temperature. These results distinguish state representation from thermodynamic assignment and provide an operational framework for local thermodynamics beyond weak coupling.
AI Summary
The paper addresses whether a subsystem’s reduced equilibrium state at strong coupling determines a local temperature. A full-rank state fixes the dimensionless Gibbs generator $-\ln\rho_S$ up to an additive constant, but not its factorization into inverse temperature and Hamiltonian. The proposed prescription is to identify a local Hamiltonian independently of the stationary Gibbs representation, then test whether $-\ln\rho_S=\beta^*H_S^{\mathrm{loc}}+cI$. If compatible, the populations determine the corresponding temperature; otherwise no scalar temperature is assigned relative to that Hamiltonian. Two exactly solvable models illustrate this prescription. For a QND-coupled two-level system and oscillator, with stability condition $|\chi|<\omega$, exact partition functions yield $\beta^*\varepsilon=\ln(p_-/p_+)$. Choosing the bare TLS Hamiltonian gives a temperature-renormalized description equivalent, at the state level, to mean-force level renormalization at the bath temperature. Population inversion produces negative $T^*$; the intervening infinite-temperature point is a regular maximally mixed state. For a number-conserving bosonic mode coupled to a continuum, retarded dynamics supplies a renormalized excitation scale, while the exact stationary occupation determines temperature through $\beta^*\widetilde\omega_S=\ln[(1+\bar n_S)/\bar n_S]$. Canonical local heat capacities at fixed identified Hamiltonian are nonnegative, unlike mean-force excess heat capacities, which can be negative. Hamiltonian identification remains model- or measurement-specific. The bosonic construction assumes an existing asymptotic frequency and a continuum regime without additional localized-mode contributions. Fixed-Hamiltonian canonical derivatives are explicitly distinguished from responses along trajectories where the Hamiltonian changes.
Central claim may not hold: For the stated Lorentz–Drude continuum below the bound-state threshold, the strict long-time retarded frequency tends to zero rather than the generally nonzero resonance frequency, so the bosonic Hamiltonian calibration as defined cannot generate the claimed stationary Gibbs state. The claimed uniqueness of local temperature needs a nonconstant local Hamiltonian, since a Hamiltonian proportional to the identity generates the maximally mixed state at every inverse temperature.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 5 pages, 3 figures