Dissipation-Enabled Operator-Norm Locality Bounds for Bose-Hubbard Hamiltonians

Technical Summary: Dissipation-Enabled Operator-Norm Locality Bounds for Bose-Hubbard Hamiltonians

Problem Statement
The principle of locality, formalized by Lieb–Robinson bounds, is a cornerstone of quantum many-body physics, asserting that information propagation in lattice systems is confined to a linear light cone. While this holds for quantum spin systems with finite-dimensional local Hilbert spaces, it fails for closed bosonic lattice systems (such as the Bose-Hubbard model). In these systems, the local Hilbert space is infinite-dimensional, and the generator involves unbounded operators (creation, annihilation, and number operators). Consequently, the information propagation velocity is not uniformly bounded by the operator norm of local interactions; instead, it can grow with the local boson occupancy, which may be macroscopically large.

Existing literature has established state-dependent Lieb–Robinson bounds for "well-behaved" initial states (e.g., bounded density or particle-free regions). However, even for such states, bosonic propagation can be super-ballistic (e.g., accelerating exponentially or propagating distances scaling as $t^D$). There is no general operator-norm bound that holds uniformly over all initial states for closed Bose-Hubbard dynamics.

Methodology
The authors investigate how on-site dissipation can restore operator-norm locality. They consider the dissipative Bose-Hubbard model governed by a Lindblad master equation with on-site $\ell$-photon loss ($\ell > 2$). The core methodology relies on three interconnected ingredients:

1. Sobolev Regularization via Dissipation: The authors utilize the fact that local photon loss acts as a dynamical regulator. They employ quantum Sobolev spaces (weighted by local particle moments) to show that the dissipative dynamics instantaneously regularizes the state. Specifically, the predual dynamics maps trace-class operators into weighted Sobolev spaces, providing uniform bounds on local particle moments even for initial states with arbitrarily high occupancy.
2. Adaptive Particle-Number Truncation: To handle the unbounded hopping terms, the authors introduce a time-dependent particle-number cutoff $M(t)$. They replace the unbounded hopping operators with truncated versions $H^{(M)}_e$ that act only within a finite subspace. The cutoff $M(t)$ is chosen adaptively based on the moment bounds derived from the dissipation, ensuring the truncation error remains controlled.
3. Bounded Interaction Lieb–Robinson Bounds: Once the hopping terms are truncated, the system effectively has bounded interactions (though unbounded on-site terms remain). The authors apply existing Lieb–Robinson bounds for systems with bounded interactions, noting that the velocity is controlled by the bounded interaction strength and is insensitive to the unbounded on-site terms (which are handled by contractive, support-preserving semigroups).

The proof strategy involves a Duhamel expansion comparing the full evolution to a truncated evolution on a finite buffer region. The error is split into:

• Truncation Error: Controlled by the Sobolev regularization (moment bounds) supplied by the dissipation.
• Boundary Propagation Error: Controlled by the standard Lieb–Robinson bound for the truncated (bounded) dynamics.

Key Contributions and Results

1. Uniform Operator-Norm Lieb–Robinson Bound (Theorem 3.1):
   For the dissipative Bose-Hubbard model with $\ell > 2$, the authors establish a Lieb–Robinson bound that holds uniformly in the initial state and the finite volume. The bound compares the full Heisenberg evolution $\tau_t(A)$ with the evolution restricted to a finite region $R$, $\tau^R_t(A)$.

   • The bound features a time-dependent occupation cutoff that follows the moment regularization.
   • On a $D$-dimensional lattice ($\mathbb{Z}^D$), the error decays super-polynomially in the buffer radius $r$ at fixed time $t$: $\sim \exp(-c [\log(2+r)]^2)$.
   • The bound is valid for all $p > 2$ (where $p$ relates to the moment order), allowing for arbitrary polynomial decay rates.

2. Fixed Cutoff Approximation (Theorem 3.2):
   The authors provide a complementary bound for a fixed occupation cutoff $M$. This separates the hopping truncation error (scaling as $M^{-(p-2)/2}$) from the propagation error across the boundary. This result allows for explicit control of the error by choosing a sufficiently large $M$ relative to the distance and time.

3. Extension to Cat-Code Dissipation (Proposition 3.5):
   The framework is extended to polynomial interactions (up to quartic degree) with shifted four-photon loss, relevant for bosonic cat codes. A similar uniform Lieb–Robinson bound is established, showing super-polynomial decay for initial states in the code space.

4. Consequences for Quantum Information:

   • Thermodynamic Limit (Corollary 5.2): The uniform bounds imply the existence of a unique thermodynamic limit (quasi-local algebra) for the dissipative dynamics, converging uniformly on finite time intervals.
   • Local Adiabatic Approximation (Theorem 5.3): For initial states in the cat-code space, the authors prove a local adiabatic approximation uniform in the total volume, with an error scaling as $O(\gamma^{-1+\eta})$.
   • Hybrid Simulation (Theorem 5.5): For $\ell > 3$, the authors propose a hybrid bosonic-qubit simulation scheme. Physical multiphoton loss regularizes the input, followed by an occupation test and encoding into qubits. This yields a diamond-norm error bound that is uniform over all input states, enabling efficient digital quantum simulation without assumptions on the initial state's moments.

Significance and Claims
The paper claims that local dissipation fundamentally alters the locality structure of bosonic lattice systems. By "regularizing" the state on a Sobolev-type scale of local particle moments, dissipation suppresses the macroscopic occupancies that typically break Lieb–Robinson bounds.

• Uniformity: The primary significance is that these locality bounds are uniform in the input state. Unlike closed Bose-Hubbard systems where bounds depend on specific initial state properties (e.g., bounded density), the dissipative model admits operator-norm bounds similar to those in quantum spin systems.
• Almost-Ballistic Propagation: The results establish "almost-ballistic" propagation, where the effective light cone radius grows as $t^{1+\eta}$ for any $\eta > 0$. While a strict finite-velocity exponential light cone (linear in $t$) is not proven (due to constants growing with moment order), the super-polynomial decay is a significant improvement over the super-ballistic behavior of closed systems.
• Practical Implications: The work provides a rigorous foundation for the thermodynamic limit of open bosonic systems and enables efficient simulation protocols that do not require prior knowledge of the initial state's moment structure.

The authors note that extending these results to two-photon loss ($\ell=2$) remains an open challenge due to a non-integrable singularity in the adaptive cutoff near $t=0$, and establishing a strict linear light cone is an open direction for future work.