Relative Entropy Decay via BKM coercivity for Quantum Markov Semigroups

AI summary

Overview

Research area: Quantum information theory and mathematical quantum physics, specifically functional inequalities and convergence rates for quantum Markov semigroups.

Technical level: Advanced. The paper is written for readers comfortable with operator algebras, completely positive maps, quantum relative entropy, and Lindbladian dynamics.

Scope: The paper introduces a BKM coercivity constant for finite-dimensional quantum Markov semigroups with a faithful asymptotic conditional expectation and proves that it characterizes modified log-Sobolev inequalities, including their complete matrix-amplified versions, without assuming detailed balance.

What This Paper Is About

Quantum Markov semigroups describe how open quantum systems relax toward equilibrium, and the modified log-Sobolev inequality (MLSI) is the standard tool for proving that the quantum relative entropy decays exponentially fast. Most existing criteria for MLSI require GNS detailed balance, a strong symmetry condition that many physically relevant dynamics, including the Chen-Kastoryano-Gilyén (CKG) Gibbs samplers, do not satisfy. This paper introduces a new quantity called the BKM coercivity constant and shows that its positivity is equivalent to entropy decay, providing a criterion that works under the weaker assumption of a faithful asymptotic conditional expectation with no detailed balance assumed.

Key Contributions

1. Definition of BKM coercivity. The authors define the BKM coercivity constant (\lambda_{\mathrm{BKM}}(\mathcal{L})) using the Bogoliubov-Kubo-Mori metric and prove that the MLSI constant (\alpha(\mathcal{L})) is equal to it up to the Pimsner-Popa index (C(E)).

2. Amplification theorem. They prove that the complete BKM coercivity constant, taken over all matrix amplifications, is already attained with a qubit auxiliary system and equals the GNS spectral gap: (\lambda_{\mathrm{BKM}}^{\mathrm{c}}(\mathcal{L})=\lambda_{\mathrm{BKM}}(\mathrm{id}{M_2}\otimes\mathcal{L})=\lambda{\mathrm{GNS}}(\mathcal{L})).

3. Complete MLSI criterion. Combining the two results yields the bound (\lambda_{\mathrm{GNS}}(\mathcal{L})/(2C_{\mathrm{cb}}(E))\leq\alpha^{\mathrm{c}}(\mathcal{L})\leq\lambda_{\mathrm{GNS}}(\mathcal{L})), so the complete modified log-Sobolev inequality holds if and only if the GNS gap is strictly positive.

4. Application to CKG samplers. They show that every fixed finite-dimensional Chen-Kastoryano-Gilyén Gibbs sampler satisfies the complete modified log-Sobolev inequality relative to its fixed-point algebra.

Main Findings

• BKM coercivity controls MLSI: Under the assumption that (P_t=e^{t\mathcal{L}}) converges in norm to a faithful conditional expectation (E), the paper proves (\lambda_{\mathrm{BKM}}(\mathcal{L})/(2C(E))\leq\alpha(\mathcal{L})\leq\lambda_{\mathrm{BKM}}(\mathcal{L})). Thus MLSI holds exactly when the BKM coercivity constant is positive.

• The upper bound comes from linearization: Expanding entropy production around an invariant state shows that the MLSI constant is bounded above by the BKM coercivity constant. This is Proposition 2.11.

• The lower bound comes from Frenkel's formulas: Using Frenkel's integral representation of relative entropy and its Hessian, the authors compare entropy decay with BKM metric contraction and obtain the lower bound with the factor (2C(E)).

• Qubit amplification suffices: The complete BKM constant equals (\lambda_{\mathrm{BKM}}(\mathrm{id}{M_2}\otimes\mathcal{L})), so a two-dimensional auxiliary system already saturates the matrix amplification. This also equals the GNS spectral gap (\lambda{\mathrm{GNS}}(\mathcal{L})), which is defined without any symmetry assumption.

• Complete criterion: The complete MLSI constant satisfies (\lambda_{\mathrm{GNS}}(\mathcal{L})/(2C_{\mathrm{cb}}(E))\leq\alpha^{\mathrm{c}}(\mathcal{L})\leq\lambda_{\mathrm{GNS}}(\mathcal{L})). Consequently, (\alpha^{\mathrm{c}}(\mathcal{L})>0) if and only if (\lambda_{\mathrm{GNS}}(\mathcal{L})>0).

• No detailed balance required: The entire framework applies to semigroups that admit a faithful asymptotic conditional expectation, with no GNS or KMS detailed balance assumed. Under GNS symmetry, the BKM constants and the GNS and KMS gaps all coincide with the spectral gap of (-\mathcal{L}), recovering the earlier index bound of Gao and Rouzé.

• CKG samplers satisfy CMLSI: For CKG samplers, the GNS Dirichlet form is an integral of squared commutator norms involving filtered jump operators. The authors show that its kernel is exactly the fixed-point algebra, which in finite dimensions implies the complete modified log-Sobolev inequality.

• Examples separate the constants: Section 6 provides examples distinguishing spectral gaps from entropy decay constants. The paper does not report numerical benchmark values for these examples in the truncated content.

Methodology in Plain English

The authors build a bridge between two ways of measuring convergence. The first is relative entropy, which measures how far a state is from equilibrium. The second is the BKM metric, which comes from the curvature of relative entropy and is defined through the map (\Gamma_\sigma(X)=\int_0^1 \sigma^s X \sigma^{1-s},ds).

To get an upper bound on the MLSI constant, they look at states very close to equilibrium. In that regime, both entropy production and relative entropy can be expanded to second order, and their ratio is exactly the BKM coercivity ratio.

To get a lower bound, they use an integral formula due to Frenkel that writes relative entropy as an integral over a positive measure. The same formula applied to a state and to its evolution under the semigroup lets them compare the entropy decrease directly with the decrease in the BKM metric. The comparison costs a factor of (2C(E)), where (C(E)) is the Pimsner-Popa index measuring how far the conditional expectation is from being trace preserving in the relevant sense.

For the complete version, they tensor the generator with matrix algebras and use the explicit structure of finite-dimensional conditional expectations. The key surprise is that the infimum over all matrix sizes is reached at (2\times 2) matrices, so the complete constant equals the GNS spectral gap.

Finally, they apply the criterion to CKG samplers by identifying the kernel of the relevant Dirichlet form with the fixed-point algebra.

Why This Matters

Impact on research: The paper removes a long-standing technical obstruction. Earlier quantitative MLSI bounds relied on GNS symmetry, a quantum analog of detailed balance. The new BKM coercivity approach works for semigroups that only preserve the Gibbs state through KMS symmetry, which is the setting of the CKG Gibbs samplers. It also connects three previously separate quantities: the BKM metric from quantum information geometry, the Pimsner-Popa index from operator algebras, and the GNS spectral gap from quantum Markov semigroup theory. The result that a qubit auxiliary system saturates the complete constant is a structural insight likely to simplify future arguments about complete inequalities.

Potential application areas:

• Quantum Gibbs state preparation: The CKG sampler is designed for efficient quantum simulation of thermal states. A positive CMLSI constant gives an exponential relative entropy decay guarantee for these dynamics.

• Quantum thermal simulation: The paper directly addresses whether the CKG construction, which is efficiently simulable, also mixes rapidly in the relative entropy sense.

• Open quantum system dynamics: The framework applies to any finite-dimensional quantum Markov semigroup with a faithful asymptotic conditional expectation, including non-primitive semigroups with nontrivial fixed-point algebras.

• Many-body thermalization: The results connect to the broader program of proving rapid thermalization for noncommuting Hamiltonians, where GNS symmetry is typically unavailable.

Industry relevance: The work is theoretical and does not report implementations or hardware benchmarks. Its relevance is to groups developing quantum algorithms for thermal state preparation and quantum simulation, where mixing-time guarantees are needed to justify resource estimates. Because the paper establishes positivity of the CMLSI constant rather than explicit system-size scaling, its immediate industrial impact is indirect.

Future Directions

• Quantitative system-size bounds: The CMLSI criterion gives positivity, but the lower bound depends on (C_{\mathrm{cb}}(E)). Deriving explicit bounds for CKG samplers that are uniform in system size, analogous to existing one-dimensional results, is a natural next step.

• Beyond faithful asymptotic conditional expectations: The paper assumes norm convergence to a faithful conditional expectation. Extending the BKM coercivity framework to non-faithful limits or to settings where the limit is only in a weaker topology is not addressed.

• Infinite-dimensional and continuous systems: The analysis is finite-dimensional throughout. Whether the amplification theorem and the qubit saturation phenomenon persist for infinite-dimensional systems is an open direction.

• Sharpness and the factor (2C_{\mathrm{cb}}(E)): The paper provides examples separating spectral gaps from entropy decay constants, but the optimality of the constants in the complete criterion, and whether the Pimsner-Popa index dependence can be improved, remain to be clarified.

• Algorithmic consequences: The criterion establishes exponential entropy decay but does not directly translate into a complexity statement for the CKG sampler. Connecting the GNS gap positivity to explicit mixing-time and gate-complexity estimates is a logical follow-up.

Target Audience

This paper is intended for researchers and graduate students in quantum information theory, mathematical physics, and operator algebras who work on quantum Markov semigroups, functional inequalities, or thermalization. Readers should be comfortable with completely positive maps, conditional expectations, quantum relative entropy, and Lindbladian generators. It will be particularly valuable to those studying entropy decay beyond detailed balance and to researchers analyzing the mixing properties of quantum Gibbs samplers. It is not suitable as an introductory text.

Authors’ abstract

We introduce the notion of the BKM coercivity constant and show that the positivity of this constant governs the exponential relative entropy decay of the semigroup. Indeed, we prove that the modified log-Sobolev constant is equivalent to the BKM coercivity constant up to a constant depending on the asymptotic conditional expectation. We also discover that the BKM coercivity constants for matrix amplifications are already attained with a qubit auxiliary system, which also coincides with the GNS spectral gap of the generating Lindbladian. As a corollary, we establish a criterion that the complete modified log-Sobolev inequality holds if and only if the GNS gap of the semigroup is strictly positive. All the discussion above applies to quantum Markov semigroups that admit a faithful asymptotic conditional expectation as long-time equilibration, with no detailed balance condition assumed. As an application, we show the finite-dimensional Chen-Kastoryano-Gilyén Gibbs samplers always satisfy the complete modified log-Sobolev inequality.