Universal decay of (conditional) mutual information in gapped pure- and mixed-state quantum matter

Basic Information

• Paper ID: 2510.22867
• Title: Universal decay of (conditional) mutual information in gapped pure- and mixed-state quantum matter
• Authors: Jinmin Yi, Kangle Li, Chuan Liu, Zixuan Li, Liujun Zou
• Institutions: Perimeter Institute for Theoretical Physics, University of Waterloo, National University of Singapore
• Classification: cond-mat.str-el, cond-mat.quant-gas, math-ph, math.MP, quant-ph
• Publication Date: arXiv v2, November 5, 2025
• Paper Link: https://arxiv.org/abs/2510.22867

Abstract

This paper proves that superpolynomial decay of mutual information (MI) and conditional mutual information (CMI) is a universal property of gapped pure and mixed quantum phases in arbitrary spatial dimensions for spin and fermionic systems. That is, if a system in a given phase exhibits this property, then all systems in that phase possess it. The authors further prove that a broad class of phases, including chiral phases, indeed exhibit superpolynomial decay of (conditional) mutual information. As a byproduct, the paper also refines the definition of mixed-state phases.

Research Background and Motivation

Core Problems

1. Quantification of long-range correlations: In quantum many-body systems, mutual information (MI) and conditional mutual information (CMI) are fundamental quantities for measuring long-range correlations, but their universal decay behavior has not been rigorously understood.
2. Universality of decay behavior: Although correlation functions decay exponentially in gapped phases, this is insufficient to derive exponential decay of MI and CMI. In particular, how prefactors scale with subsystem size remains unclear.
3. Characterization of mixed-state phases: For mixed-state phases in open systems, there is a lack of rigorous definitions and theoretical foundations, particularly regarding whether CMI decay behavior is preserved within a phase.

Research Significance

1. Foundation of information-theoretic framework: MI and CMI are core tools in the entanglement bootstrap program and characterization of mixed-state phases.
2. Experimental measurability: These quantities can now be measured through random measurements, interferometric protocols, or tomographic experiments in quantum simulators, making their decay behavior a directly testable signature of quantum matter.
3. Connection to quantum error correction: The decay of CMI is directly related to the approximate quantum error correction capability of many-body states, which is crucial for understanding quantum phases.

Limitations of Existing Approaches

1. Empirical assumptions: Previous work assumed exponential decay of CMI in gapped phases but lacked rigorous proof.
2. Unclear prefactors: Even for exponential decay forms like $I \sim fe^{-d/\xi}$, how the prefactor $f$ scales with region size remained unexplained.
3. Incomplete definition of mixed-state phases: Previous definitions required CMI to maintain exponential decay throughout evolution, but whether this requirement is automatically satisfied was unproven.

Core Contributions

1. Universality Theorem (Theorem 1): Proves that for gapped almost-local Hamiltonians, if the ground state satisfies superpolynomial decay of MI or CMI, then all states in the same phase satisfy the same decay behavior.
2. Universality of mixed-state phases (Theorem 2): Proves that MI and CMI decay behavior is automatically preserved for mixed states connected by local invertible finite-depth quantum channels, and this holds even for polynomial or exponential decay.
3. Precise prefactor bounds: Establishes precise decay forms: $I(A:C) = O(\text{poly}(|A|,|B|)\text{dist}(A,C)^{-\infty})$$I(A:C|B) = O(\text{poly}(|A|,|B|)\text{dist}(A,C)^{-\infty})$ Crucially, the prefactor depends only on $|A|$ and $|B|$, independent of $|C|$.
4. Broad applicability: Proves that a wide class of topological phases, including commuting projector models and chiral phases, possess superpolynomial decay of MI and CMI.
5. Refined definition of mixed-state phases: Proposes a precise definition based on local invertible finite-depth channels (Definition 1) and proves automatic preservation of CMI decay without additional assumptions.

Detailed Methodology

Problem Setup

Consider spin or fermionic systems on a $D$-dimensional lattice, partitioned into three regions $A, B, C$ (as shown in Figure 1):

• $A$ is a contractible region
• $B$ separates $A$ from $C$
• $ABC$ constitutes the entire lattice

Mutual Information (MI) is defined as: $I(A:C) = S(A) + S(C) - S(AC)$

Conditional Mutual Information (CMI) is defined as: $I(A:C|B) = S(AB) + S(BC) - S(B) - S(ABC)$

where $S(R)$ is the von Neumann entropy of the reduced density matrix on region $R$.

Core Technical Framework

1. Quasi-adiabatic Evolution Decomposition

Almost-local Hamiltonians: A Hamiltonian $H = \sum_j H_j$ where each interaction term $H_j$ is $f$-local, meaning: $H_j = \sum_n H_j^{(n)}, \quad \|H_j^{(n)}\| < 2Cf(n-1)$ where $f(n) = O(n^{-\infty})$ is a superpolynomially decaying function.

Quasi-adiabatic evolution: Two states $\rho$ and $\rho'$ in the same phase are connected by quasi-adiabatic evolution operator $U_t^H$: $\rho' = U_t^H \rho (U_t^H)^\dagger$

Key Lemma (Lemma 1): For evolution $U_t^H$ generated by any almost-local Hamiltonian, there exists an approximate decomposition: $U_t^H \approx \tilde{U}_t^H := U_t^{H_B}(U_t^{H_{A^+}+H_{C^+}})^\dagger U_t^{H_{CC^+}+H_{A^+A}}$

where:

• $A^+ := \{j \in B | \text{dist}(j,A) < \text{dist}(A,C)/3\}$
• $C^+ := \{j \in B | \text{dist}(j,C) < \text{dist}(A,C)/3\}$
• Approximation error: $\|U_t^H - \tilde{U}_t^H\| < \epsilon = O(\text{poly}(|B|)\text{dist}(A,C)^{-\infty})$

2. Proof Strategy for MI Decay

Step 1: Construct reference state Define $\tilde{\rho} = \tilde{U}_t^H \rho (\tilde{U}_t^H)^\dagger$, utilizing the decomposition structure: $\tilde{\rho}_{AC} = \text{tr}_{A^+C^+}(U_t^{H_{CC^+}+H_{A^+A}} \rho_{AA^+CC^+} (U_t^{H_{CC^+}+H_{A^+A}})^\dagger)$

Step 2: Exploit relative entropy monotonicity Since relative entropy is monotone under quantum channels: $I_{\tilde{\rho}}(A:C) = S(\tilde{\rho}_{AC}\|\tilde{\rho}_A \otimes \tilde{\rho}_C) \leq S(\rho_{AA^+CC^+}\|\rho_{AA^+} \otimes \rho_{CC^+}) = I_\rho(AA^+:CC^+)$

Step 3: Continuity argument Using the Fannes-Audenaert inequality, from $\|\rho' - \tilde{\rho}\|_1 < 2\epsilon$ we obtain: $I_{\rho'}(A:C) \leq |I_{\rho'}(A:C) - I_{\tilde{\rho}}(A:C)| + I_{\tilde{\rho}}(A:C)$$< 3\epsilon \log d \cdot |A| + 3\epsilon H_2(\epsilon) + I_\rho(AA^+:CC^+)$

3. Proof Strategy for CMI Decay

Core idea: Small CMI is equivalent to the existence of an approximate recovery map.

Step 1: Utilize Petz recovery map For $\rho$, if $I_\rho(AA^+:CC^+|B\backslash A^+C^+)$ is small, there exists a Petz recovery map $\mathcal{E}_B^P$ supported on $B^- := B\backslash(A^+C^+)$ such that: $\rho \approx \mathcal{E}_{B^-}^P(\rho_{AB\backslash C^+})$

Step 2: Construct recovery map for $\rho'$ Using the decomposition structure, construct a recovery map (as shown in Figure 3): $\mathcal{E}_B' := \text{Ad}_{U_t^{H_B}(U_t^{H_{C^+}})^\dagger U_t^{H_{CC^+}}} \circ \mathcal{E}_{B^-}^P \circ \text{tr}_{C^+} \circ \text{Ad}_{(U_t^{H_B})^\dagger}$

The key is that through the trace structure in the decomposition, the evolution operators cancel, making $\mathcal{E}_B'$ completely supported on region $B$.

Step 3: CMI upper bound The existence of the recovery map gives an upper bound on CMI: $I_{\rho'}(A:C|B) \leq 7\log d \cdot |A|\sqrt{\|\rho' - \mathcal{E}_B'(\rho'_{AB})\|_1/2}$

Technical Innovations

1. Light-cone decomposition technique: Precisely decomposes quasi-adiabatic evolution into a product of evolution operators with clear light-cone structure. This is a key technical innovation, superior to traditional finite-depth circuit approximations which have larger errors while this decomposition maintains sharp light-cone structure with controllable error.
2. Precise prefactor control: By carefully tracking size dependence of each region, proves that the prefactor is only $\text{poly}(|A|,|B|)$, independent of $|C|$.
3. Rigorous treatment of mixed-state phases: Through local invertibility conditions, proves that recovery maps can be reduced from global to local without additional assumptions.
4. Unified framework: The same set of techniques applies to both pure-state phases (via quasi-adiabatic evolution) and mixed-state phases (via finite-depth channels).

Experimental Setup

Theoretical Verification Objects

This is theoretical work, with main verification objects including:

1. Commuting projector models: Hamiltonians describing topological order $H = \sum_j P_j$, where $[P_i, P_j] = 0$.
2. Chiral states: 2D chiral topological phases that cannot be described by commuting projector models.
3. General topological order: Constructed through pairing with time reversal.

Verification Methods

Commuting Projector Models

For ground state $\rho$, utilizing local indistinguishability:

• When $\text{dist}(A,C)$ exceeds some $O(1)$ constant, $I_\rho(A:C) = 0$
• CMI vanishes through error correction properties: $I_\rho(A:C|B) = 0$

Chiral States

Through time-reversal pairing technique:

• Chiral state $\rho$ stacked with its time-reversed partner $\rho^t$ gives $\rho \otimes \rho^t$ with modular tensor category $\mathcal{Z}(\mathcal{F})$ (Drinfeld center of $\mathcal{F}$)
• Such phases always have Levin-Wen commuting projector model representatives
• Since entropy of $\rho \otimes \rho^t$ is twice that of $\rho$, the state $\rho$ itself satisfies equations (3) and (4)

Experimental Results

Main Theoretical Results

Theorem 1 (Universality in Pure-State Phases)

Statement: Let $H_0$ be a gapped almost-local Hamiltonian. If any (possibly mixed) ground state $\rho$ satisfies: $I(A:C) = O(\text{poly}(|A|,|B|)\text{dist}(A,C)^{-\infty})$ or $I(A:C|B) = O(\text{poly}(|A|,|B|)\text{dist}(A,C)^{-\infty})$

then any state $\rho'$ in the same phase also satisfies the same equations.

Proof highlights:

• Error bound: $\epsilon = O(\text{poly}(|B|)\text{dist}(A,C)^{-\infty})$
• MI bound: $I_{\rho'}(A:C) < 3\epsilon \log d \cdot |A| + I_\rho(AA^+:CC^+)$
• CMI via recovery map fidelity bound: $F(\rho', \mathcal{E}_B'(\rho'_{AB})) \geq 2^{-I_\rho(AA^+:CC^+|B^-)/2}$

Theorem 2 (Universality in Mixed-State Phases)

Statement: Let $\rho$ and $\rho'$ be two mixed states in the same phase (connected by local invertible finite-depth channels). If $\rho$ satisfies equation (3) (or (4)), then $\rho'$ also satisfies it. This conclusion holds for superpolynomial, exponential, and even polynomial decay.

Proof highlights:

• MI: Direct application of relative entropy monotonicity
• CMI: Through local invertibility, global recovery map reduces to local recovery map: $\tilde{\mathcal{C}} \circ \mathcal{E}^P \circ \text{tr}_{CC^+}(\mathcal{C}(\rho'_{AB} \otimes \pi_C)) = \mathcal{E}'(\rho'_{AB})$ where $\mathcal{E}'$ is supported on $B$

Verification of Specific Phases

1. Commuting Projector Models

Result: MI and CMI are exactly zero when $\text{dist}(A,C)$ exceeds an $O(1)$ constant.

Proof:

• MI: Local indistinguishability directly gives $\rho_{AC} = \rho_A \otimes \rho_C$
• CMI: Proven through error correction properties and purification techniques: $I_\rho(A:C|B) = I_\psi(A:C) = 0$

2. Chiral Phases

Result: All bosonic 2D chiral states satisfy equations (3) and (4).

Proof path:

• Stack chiral state $\rho$ with time-reversed partner $\rho^t$ to get $\rho \otimes \rho^t$
• The latter has Levin-Wen model representative, thus satisfies (3) and (4)
• Since stacking only doubles entropy, $\rho$ itself satisfies (3) and (4)

3. General Topological Order

Conjecture: Assuming any topological order stacked with its time-reversed partner has a commuting projector representative (widely believed but unproven), all topological orders satisfy equations (3) and (4).

Discussion of Counterexamples

Ising model counterexample: $H = -J\sum_{\langle i,j\rangle} Z_i Z_j$ Ground state $(|\uparrow\uparrow\cdots\uparrow\rangle + |\downarrow\downarrow\cdots\downarrow\rangle)/\sqrt{2}$ violates equations (3) and (4).

Reason analysis: This ground state has long-range correlations and is unstable to perturbations. For gapped Hamiltonians with robust ground state subspaces, the authors have not found counterexamples.

Related Work

Theoretical Foundations of Entanglement Measurement

1. MI and correlation bounds: Wolf et al. 1 proved MI universally bounds correlations between observables.
2. Characterization of topological order: Kitaev-Preskill 2 and Levin-Wen 3 use topological entanglement entropy to characterize topological order.
3. CMI and topological phases: Appropriately partitioned CMI can distinguish different topological phases.

Entanglement Bootstrap Program

• Shi et al. 4-6: Derive fusion rules from entanglement, establish entanglement bootstrap methods
• Yang et al. 7: Axiomatic approach to topological mixed states
• This paper provides rigorous theoretical foundations for these programs

Mixed-State Phases

1. Recent developments: Lessa et al. 8, Sang-Hsieh 9 and others proposed characterizations of mixed-state phases
2. Definition evolution:
   • Coser-Pérez-García 26: Via fast dissipative evolution
   • Barthel-Zhang 27,28: Phase classification in open systems
   • Sang et al. 25: Definition via local invertible finite-depth channels
3. This paper's contribution: Proves automatic preservation of CMI decay, elevating conjecture to theorem

Quasi-adiabatic Evolution

• Hastings-Wen 22: First introduced quasi-adiabatic continuation concept
• Bravyi et al. 23: Stability of topological quantum order
• Bachmann et al. 24: Automorphism equivalence in gapped phases
• Kapustin-Sopenko 50: Local Noether theorem

Lieb-Robinson Bounds

• Lieb-Robinson 52: Finite group velocity in quantum spin systems
• Nachtergaele et al. 53: Correlation propagation
• This paper extends to almost-local Hamiltonians

Conclusions and Discussion

Main Conclusions

1. Established universality: Superpolynomial decay of MI and CMI is a universal property of gapped pure and mixed-state phases.
2. Precise prefactors: Prefactor form is $O(\text{poly}(|A|,|B|))$, independent of $|C|$ size.
3. Broad applicability: A wide class of phases including commuting projector models and chiral phases possess this property.
4. Refined definition of mixed-state phases: Definition based on local invertible finite-depth channels, with automatic preservation of CMI decay.

Limitations

1. Almost-locality assumption: Requires Hamiltonians to be almost-local (superpolynomially decaying interactions). While this naturally appears in quasi-adiabatic continuation, it limits applicability.
2. Existence of counterexamples: For certain special ground states (like symmetry-broken Ising model), equations (3) and (4) do not hold. However, the authors argue these states are unstable.
3. Completeness of topological order: The claim that all topological orders satisfy (3) and (4) depends on an unproven conjecture (having commuting projector representatives after pairing with time reversal).
4. Lindbladian evolution: The definition of mixed-state phases uses finite-depth channels, while actual systems should involve finite-time Lindbladian evolution. This extension is left for future work.

Future Directions

1. Lindbladian evolution: Extend the definition of mixed-state phases to finite-time Lindbladian evolution, develop quasi-adiabatic continuation concept for mixed-state phases.
2. Role of local invertibility: Understand the exact role and necessity of local invertibility in mixed-state phases.
3. Precise phase classification: Rigorously understand which phases have superpolynomial decay of MI and CMI, completely characterize such phases.
4. Finite temperature generalization: Extend results to finite temperature systems (see recent work 48).
5. Experimental verification: Measure MI and CMI decay behavior in quantum simulators using random measurements and other techniques.

In-Depth Evaluation

Strengths

1. Theoretical rigor:
   • Provides complete mathematical proofs from Lieb-Robinson bounds to quasi-adiabatic evolution decomposition with tight logical chains
   • Precisely tracks all error terms and prefactor size dependencies
   • Comprehensive supplementary material (17 pages) containing all technical details
2. Methodological innovation:
   • Light-cone decomposition technique: Decomposes quasi-adiabatic evolution into evolution operator products with clear light-cone structure, key innovation superior to traditional finite-depth circuit approximations
   • Recovery map construction: Cleverly uses decomposition structure to reduce global recovery maps to local ones
   • Unified framework: Single set of techniques handles both pure and mixed-state phases
3. Result importance:
   • Elevates previous conjectures (CMI decay preservation within phases) to rigorous theorems
   • Provides solid theoretical foundation for entanglement bootstrap programs and mixed-state phase characterization
   • Precise prefactor bounds (independent of $|C|$) have practical application value
4. Broad applicability:
   • Applies to arbitrary spatial dimensions
   • Covers both spin and fermionic systems
   • Includes difficult topological phases like chiral phases
5. Academic impact:
   • Refines definition of mixed-state phases (Definition 1), important contribution to this emerging field
   • Connects multiple fields: topological order, quantum information, quantum error correction

Weaknesses

1. Technical complexity:
   • Proofs involve extensive technical details (e.g., Lemma III.2 proof spans 5 pages)
   • Error bound expressions are extremely complex (equations III.6-III.11), difficult for intuitive understanding
   • High barrier to entry for non-specialists
2. Limited applicability:
   • Requires almost-local Hamiltonians (superpolynomially decaying), excludes long-range interaction systems
   • Does not apply to systems with spontaneous symmetry breaking (e.g., symmetric ground state of Ising model)
   • Mixed-state phase definition requires local invertibility, may not hold in actual systems
3. Theoretical completeness:
   • Statement about all topological orders satisfying (3) and (4) depends on unproven conjecture
   • No sufficient and necessary conditions given for determining which phases have superpolynomial MI/CMI decay
   • For systems with robust ground state subspaces, no counterexamples found but necessity not proven
4. Practical considerations:
   • Constants in error bounds (such as $M_f, C_h$) may be large, actual decay in physical systems may be slower
   • Superpolynomial decay, though weaker than exponential, may show little difference in finite systems
   • Lacks numerical verification or quantitative analysis of specific examples
5. Missing Lindbladian evolution:
   • Mixed-state phase definition uses finite-depth channels rather than physical Lindbladian evolution
   • Technical difficulties of this extension not fully discussed

Impact

1. Contribution to field:
   • Foundational contribution: Provides rigorous mathematical basis for entanglement bootstrap programs and mixed-state phase theory
   • Methodological contribution: Light-cone decomposition technique may apply to other problems
   • Conceptual contribution: Refines mixed-state phase definition, promotes standardization in the field
2. Practical value:
   • Experimental guidance: Precise decay forms can guide quantum simulator experiment design
   • Numerical computation: Prefactor bounds useful for finite-size extrapolation
   • Quantum error correction: Connection to approximate quantum error correction codes (via CMI) has practical applications
3. Reproducibility:
   • Theoretical work with complete proofs provided
   • Detailed supplementary material containing all technical details
   • Rigorous mathematical exposition, strong verifiability

Applicable Scenarios

1. Theoretical research:
   • Classification and characterization of topological phases
   • Applications of entanglement bootstrap programs
   • Study of mixed-state phases
   • Information-theoretic properties of quantum many-body systems
2. Numerical simulation:
   • Truncation error estimation in tensor network methods
   • Finite-size extrapolation to thermodynamic limit
   • Numerical identification of phase boundaries
3. Experimental physics:
   • Measurement of MI and CMI in quantum simulators
   • Experimental characterization of topological phases
   • Performance evaluation of quantum error correction codes
4. Inapplicable scenarios:
   • Long-range interaction systems (e.g., dipolar interactions)
   • Critical systems (gapless)
   • Degenerate ground states with spontaneous symmetry breaking

Technical Depth Analysis

Key Technical Breakthroughs

Elegance of light-cone decomposition: Traditional methods (e.g., Haah et al. 54) decompose evolution into shallow quantum circuits, but depth grows linearly with time, making error control difficult. This paper's innovation:

1. Decomposes into only $O(1)$ evolution operators (rather than depth-$O(t)$ circuits)
2. Through four applications of basic decomposition (Figure 2a), constructs decomposition with clear light-cone (Figure 2b)
3. Error is only $O(\text{poly}(|B|)w(B)^{-\infty})$, where $w(B)$ is width of region $B$

Essence of prefactor independence: Prefactor $O(\text{poly}(|A|,|B|))$ independent of $|C|$ is non-trivial, stemming from:

1. Decomposition makes $\tilde{\rho}_{AC}$ depend only on $\rho_{AA^+CC^+}$
2. Both $|A^+|, |C^+|$ are $O(|B|)$
3. In continuity argument, Fannes bound on entropy only introduces $\log d^{|A|}$ factor

Application of Mathematical Tools

1. Relative entropy monotonicity: $S(\mathcal{E}(\rho)\|\mathcal{E}(\sigma)) \leq S(\rho\|\sigma)$ is core to proving MI preservation.
2. Fannes-Audenaert inequality: Controls entropy continuity, key to converting 1-norm distance to entropy difference.
3. Petz recovery map: Deep connection between CMI and error correction; small CMI equivalent to good recovery map.
4. Lieb-Robinson bounds: Controls information propagation speed, foundation for decomposition error estimation.

Selected References

1 Wolf et al., "Area Laws in Quantum Systems: Mutual Information and Correlations," PRL 100, 070502 (2008)

2 Kitaev & Preskill, "Topological Entanglement Entropy," PRL 96, 110404 (2006)

10 Fawzi & Renner, "Quantum Conditional Mutual Information and Approximate Markov Chains," CMP 340, 575 (2015)

22 Hastings & Wen, "Quasiadiabatic continuation of quantum states," PRB 72, 045141 (2005)

25 Sang et al., "Mixed-state phases from local reversibility," arXiv:2507.02292 (2025)

52 Lieb & Robinson, "The finite group velocity of quantum spin systems," CMP 28, 251 (1972)

Overall Assessment: This is a high-quality theoretical physics paper that achieves important progress in information-theoretic properties of quantum many-body systems. Despite high technical complexity, its mathematical rigor and universal applicability make it a significant contribution to the field. Particularly for the emerging theory of mixed-state phases, this paper provides urgently needed rigorous mathematical foundations. The paper's main value lies in elevating empirical conjectures to rigorous theorems, establishing solid foundations for future research.