A Route to Quantum-Coherent Thermodynamics

The gist The authors develop a framework in which energy fluctuations can retain genuinely quantum-coherent contributions by foliation state space into “minimum-variance leaves,” leading to a “leaf typicality” hypothesis that extends eigenstate thermalization beyond equilibrium.

Minimum-Variance Foliation

The state space is foliated into “minimum-variance leaves,” defined by minimizing the average energy variance over all pure-state decompositions, with the minimum set by the quantum Fisher information on page 1, and each leaf is associated with a fixed family of pure states on page 3. On each leaf, a “leaf-canonical ensemble” is constructed by defining the least-biased state compatible with normalization and mean energy on page 5. The Gibbs ensemble is recovered on the distinguished commuting leaf, while generic states are organized by their leaf label on page 5.

Leaf Typicality Hypothesis

The framework provides a natural setting to extend eigenstate thermalization beyond equilibrium via a “leaf typicality” hypothesis on page 5. According to this hypothesis, local observables depend only on the leaf and energy and are reproduced by a representative pure state drawn from the optimal ensemble, whose minimized energy spread reduces the complexity of time evolution on page 5. This is formalized by asking whether typical states within a given leaf are locally indistinguishable from the corresponding leaf-canonical ensemble (or its microcanonical counterpart) on page 5. The refinement nearstrong when the number of outliers grows more slowly than any power of the Hilbert-space dimension on page 5.

Quantifying Energy Incoherence

The foliation provides a concrete handle on quantum coherence with respect to H, i.e., coherence in the energy eigenbasis on page 5. The variance-minimizing optimal family encodes the coherent structure, while varying the populations amounts to classical mixing within that structure on page 5. A natural leaflevel incoherence indicator is provided by the barycenter ρ0(L) and its von Neumann entropy, I(L) on page 5. This measure satisfies 0 ≤ I(L) ≤ log d, and it is maximal, I(L) = log d, if and only if ρ0(L) = I/d on page 5.

Spectral Compression

The analysis demonstrates a spectral compression where the effective number of energy eigenstates participating in a pure state is reduced on page 8. This is quantified by comparing the standard spectral decomposition with the min-variance decomposition, where the entropy-density gain ∆sdiag(L) corresponds to an exponential reduction in the effective number of relevant states on page 9. The data are consistent with an exponential gain on page 14.

Dynamical Content of Leaf Typicality

For local observables, the exact evolutions are reproduced by evolving a representative (pure) state drawn from the optimal ensemble, even when the evolving density matrix is highly mixed on page 6. This reduction in complexity for time evolution is achieved because typical representatives have an energy-variance density controlled by the QFI density, yielding a spectral compression that may also be advantageous for dynamical simulations on page 8. The cost of the calculation is governed by the spread of energy eigenstates of H in Φ¯ on page 6.

Role of SLD

Remarkably, all density matrices on a leaf share the same symmetric logarithmic derivative (SLD) on page 8. Under mild physical assumptions, this SLD has a quasilocal density, and states belonging to thermodynamically distinct leaves cannot be locally equivalent on page 8. Leaf typicality then implies that typical representatives have an energy-variance density controlled by the QFI density, yielding a spectral compression that may also be advantageous for dynamical simulations on page 8.

Conclusion

The framework extends the notion of thermodynamic ensemble beyond equilibrium, providing a comparably effective distillation of microscopic complexity into macroscopic data on page 1. Leaves replace individual states as the fundamental thermodynamic objects of description on page 5. The foliation also naturally induces a quantitative notion of energy (in)coherence on page 5.

How it works

1. State space is foliated into “minimum-variance leaves,” defined by minimizing the average energy variance over all pure-state decompositions, with the minimum set by the quantum Fisher information on page 3.

2. On each leaf, a “leaf-canonical ensemble” is constructed by defining the least-biased state compatible with normalization and mean energy on page 5.

3. The Gibbs ensemble is recovered on the distinguished commuting leaf, while generic states are organized by their leaf label on page 5.

4. The “leaf typicality” hypothesis posits that local observables depend only on the leaf and energy and are reproduced by a representative pure state drawn from the optimal ensemble on page 5.

Key Concepts

**)minimum-variance leaves is defined by minimizing the average energy variance over all pure-state decompositions on page 3. **

**)leaf typicality suggests that local observables are determined, in generic systems, by the leaf and the energy on page 5. **

**)energy (in)coherence is quantified by the barycenter ρ0(L) and its von Neumann entropy, I(L) on page 5. **

**)spectral compression is demonstrated by comparing the standard spectral decomposition with the min-variance decomposition, where the entropy-density gain ∆sdiag(L) corresponds to an exponential reduction in the effective number of relevant states on page 9. **

**)SLD all density matrices on a leaf share the same symmetric logarithmic derivative (SLD) on page 8. **

**)leaf canonical ensemble is defined by maximizing the Shannon entropy of the populations, S(H)th [ρ] on page 5. **

**)near-strong leaf typicality is a refinement of leaf typicality where limL→∞ logd N∆ = 0 on page 5. **

**)spectral compression is shown to be consistent with an exponential gain on page 14. **

How it works

The analysis shows that the cost of the calculation is governed by the spread of energy eigenstates of H in Φ¯ on page 6. The reduction in complexity for time evolution is achieved because typical representatives have an energy-variance density controlled by the QFI density, yielding a spectral compression that may also be advantageous for dynamical simulations on page 8.

The paper is significant because it introduces a method to organize nonequilibrium states based on local properties and identifies typical versus exceptional behavior through the concept of leaf typicality, which extends eigenstate thermalization beyond equilibrium on page 8. This framework provides a statistical description of local physics that retains large equivalence classes of locally indistinguishable states on page 5. The resulting spectral compression suggests an exponential reduction in the effective number of relevant energy eigenstates, which is relevant for dynamical simulations on page 9. This work connects the finite-dimensional foliation introduced above with the thermodynamic regime, where physically relevant information is encoded in local observables on page 5.

Key Concepts

**)minimum-variance leaves is defined by minimizing the average energy variance over all pure-state decompositions on page 3. **

**)leaf typicality suggests that local observables are determined, in generic systems, by the leaf and the energy on page 5. **

**)energy (in)coherence is quantified by the barycenter ρ0(L) and its von Neumann entropy, I(L) on page 5.

1. Leaf-Canonical Ensemble for Non-Equilibrium States: Develop an ensemble description beyond equilibrium by defining a leaf as a set of pure states that minimize average energy variance over all pure-state decompositions. This allows for a statistical description of local physics away from equilibrium, where the text states that local observables depend only on the leaf and energy.

2. Leaf Typicality Hypothesis: Implement an effective replacement for individual quantum states in time evolution by using a representative state drawn from the optimal ensemble. This is achieved by asserting that local observables depend only on the leaf and energy and are reproduced by a representative pure state drawn from the optimal ensemble, whose minimized energy spread reduces the complexity of time evolution.

3. Quantifying Energy Incoherence: Introduce a quantitative measure of quantum coherence in the energy eigenbasis via leaf level incoherence indicator, defined as the von Neumann entropy of the barycenter [ρ0(L)] or I(L) = SvN [ρ0(L)], which is maximal, I(L) = log d, if and only if ρ0(L) = I/d.

4. Spectral Compression for Efficient Simulations: Utilize the concept of spectral compression to reduce the effective number of energy eigenstates involved in calculations. This is quantified by comparing the energy-spectral compression where the min-variance ensemble provides a spectral compression, leading to an exponential reduction in the effective number of relevant states.

5. Adaptive Simulation Strategy: Employ a diagnostic based on "N∆ as the total number of eigenstates for which ⟨φiOφi⟩ − fO,L(Eρ,i) > ∆" to dynamically select and focus computational resources on the most typical eigenstates for a given local observable O and leaf L.

Equilibrium thermodynamics emerges because a vast number of microscopic states become indistinguishable at the level of local observables. We start from the observation that a nonequilibrium quantum state can also be represented by an ensemble of pure states that minimizes the average energy variance. We show that varying the statistical weights within this optimal family organizes state space into ``minimum-variance leaves'', which generalize the space of stationary states underlying equilibrium. We observe thermodynamic compression on the leaves, which in the simplest scenario manifests itself in the indistinguishability of states with the same energy density. We call this ``leaf typicality''; at equilibrium it reduces to eigenstate thermalization or its generalizations. Numerical tests support leaf typicality on some leaves and indicate a breakdown of its stronger forms on others, suggesting a rich thermodynamic organization beyond equilibrium. These results open a route to a unified statistical mechanics of equilibrium and nonequilibrium quantum states.