arXiv:2609.25242
From Effective Temperature to Non-Boltzmann State Selection in Driven-Dissipative Quantum Criticality
Tiago Jorge, Jens Paaske, Pedro Ribeiro
Abstract
Whether a nonequilibrium quantum-critical steady state can be described by an effective temperature remains an open question. We address it in a voltage-biased electronic Lipkin--Meshkov--Glick model, a minimal driven-dissipative model of a collective spin coupled to metallic leads. A controlled large-$N$ treatment reveals an overdamped open quantum-critical regime distinct from the closed model. The surrounding quantum critical fan, organized by temperature and voltage, is explored and fluctuations are found to be governed by an effective temperature $T_{\rm eff}$. At strong bias, the transition becomes first-order through a tricritical point. Remarkably, the same $T_{\rm eff}$ governs the strongly driven regime, but now varies across the entire order-parameter landscape. The steady state is therefore selected by a non-Boltzmann rule, shifting the first-order transition away from the equal-depth point of a deterministic potential. Driven criticality thus remains organized by an effective temperature while revealing non-thermal state selection.
AI Summary
The paper asks when an effective temperature describes driven quantum-critical fluctuations and whether it also determines phase selection at a first-order transition. It studies an electronic Lipkin–Meshkov–Glick dot with infinite-range Ising exchange, coupled to two voltage-biased metallic leads. Integrating out the fermions gives a Keldysh action with a controlled large-$N$ saddle point and Gaussian fluctuations. At zero temperature and bias, lead coupling changes the closed-model fluctuation divergence from $|h-h_c|^{-1/2}$ to logarithmic. For weak hybridization $\Gamma$, the overdamped critical regime crosses over to closed-like behavior at $|h-h_c|\sim\Gamma^4$, with $J=1$. At finite temperature or bias, the low-frequency fluctuation–dissipation ratio defines $T_{\rm eff}\simeq T/2+(V/4)\coth[V/(2T)]$ for small bias, approaching $V/4$ at $T=0$. This organizes the Gaussian quantum-critical fan. At $T=h=0$, the bias-driven transition becomes first order below a tricritical hybridization, with $(\Gamma_{\rm tc},V_{\rm tc})=(3,\sqrt3)/(2\pi)$. A slow-field expansion yields multiplicative-noise Langevin dynamics. Its stationary large-deviation function is $\mathcal W(\phi)=\int_0^\phi d\phi'\,\mathcal F'(\phi')/T_{\rm eff}(\phi')$. Consequently, coexistence follows equal depths of $\mathcal W$, not the deterministic potential $\mathcal F$: effective temperature varies along the order-parameter landscape. The setting assumes degenerate levels, screened charging effects and collective exchange. The state-selection calculation additionally requires collective relaxation slower than both electronic escape and the effective-temperature scale. These conditions are best satisfied near tricriticality; non-Markovian corrections are expected toward the bistable region’s small-$\Gamma$ or low-$T_{\rm eff}$ edges. The time-local noise description fails at the equilibrium quantum critical point.
Central claim may not hold: The non-Boltzmann selection formula requires slow, effectively classical collective dynamics across the relevant order-parameter landscape, whereas the abstract states the strong-drive result without this restriction. Large N alone does not justify discarding higher quantum-field vertices when computing rare interwell switching, so the proposed quasipotential is established for the approximate Gaussian-noise dynamics rather than generally for the full electronic model.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.