Towards Efficient Quantum Thermal State Preparation via Local Driving: Lindbladian Simulation with Provable Guarantees
Dominik Hahn, S. A. Parameswaran, Benedikt Placke · 2025
Read Towards Efficient Quantum Thermal State Preparation via Local Driving: Lindbladian Simulation with Provable Guarantees by Dominik Hahn, S. A. Parameswaran, Benedikt Placke (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-22816.txt.
Introduction
Quantum computers need ways to settle a system into a heat-based state for study. Dominik Hahn, S A Parameswaran, and Benedikt Placke tackle that task in a 2025 paper. Exact methods exist that reach the right state, but they need large machines that are not yet at hand. The authors seek a close match that near-term devices can run.
Their scheme repeats three steps that small devices can do. It runs the system itself, links it briefly to extra bits at set spots, and resets those bits. They prove bounds on how far the end state sits from the exact heat state. The gap rests on the plan's own settings and its mixing time. For local systems, the cost grows gently with size.
Catalog record
| CREATOR | Dominik Hahn, S. A. Parameswaran, Benedikt Placke |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 121 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-22816
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Towards Efficient Quantum Thermal State Preparation via Local Driving: Lindbladian Simulation with Provable Guarantees Dominik Hahn, S. A. Parameswaran, Benedikt Placke Abstract Preparing the thermal density matrix ρ β ∝ e − β H \rho_{\beta}\propto e^{-\beta H} corresponding to a given Hamiltonian H H is a task of central interest across quantum many-body physics, and is particularly salient when attempting to study it with quantum computers. Although solved in principle by recent constructions of efficiently simulable Lindblad master equations — that provably have ρ β \rho_{\beta} as a steady state [C.-F. Chen et al , Nature 646, pp. 561–566 (2025)] — the implementation of these “exact Gibbs samplers” requires large-scale quantum computational resources and is hence challenging in practice on current or even near-term quantum devices. Here, we propose a scheme for approximately simulating an exact Gibbs sampler that only requires the [repeated] implementation of three readily available ingredients: (a) analog simulation of H H ; (b) strictly local but time-dependent couplings to ancilla qubits; and (c) reset of the ancillas. We give rigorous guarantees on the difference between the fixed point reached by our protocol and the exact thermal state, which only depend on parameters of the protocol and its mixing time . The procedure is efficiently implementable on near-term devices if H H is local, and the mixing time scales mildly with both system size and protocol parameters. While guaranteeing the latter for Hamiltonians of interest remains an important problem for future work, here we lay the groundwork for developing fully efficient thermal state preparation protocols on quantum simulators. I Introduction Preparing thermal states of many-body systems is a key goal for a wide range of quantum devices, since it enables the exploration of such systems through “quantum numerical experiments”. However, the ability of analog quantum simulators to access low-temperature regimes of target systems such as the Hubbard model is often limited by the absence of scalable, problem-agnostic techniques for removing entropy in a controlled manner so as to reach the target energy density. Similarly, while a proposed near-term use-case for digital quantum computers is to attack problems in quantum chemistry, this goal is often obstructed by the absence of efficient ways to prepare a thermal density matrix. While various physically-motivated approaches have been proposed to meet this challenge [ 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 ] , these are typically of a heuristic and case-by-case nature. As such, their systematic error is poorly understood: in other words, it is often unclear how exactly one has to scale the available resources to obtain a close approximation to a given target state. In counterpoint to this are recent rigorous results in the quantum computer science literature [ 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 ] . A subset of particular interest to this work concerns quantum Gibbs sampling [ 24 , 25 , 26 , 27 , 28 , 29 ] : the problem of engineering a dissipative quantum dynamics whose steady state corresponds to a specified Gibbs density matrix. Although a formal solution has long been available in terms of so-called Davies generators [ 30 , 31 , 32 ] , the corresponding dynamics involves a purely dissipative Lindblad evolution under highly nonlocal and hence unphysical ‘jump operators’, making it unfeasible in practice. The recent work upends this conventional wisdom by demonstrating that Gibbs sampling is possible with controllable accuracy (in the sense above) while inducing dissipation using only quasi-local jump operators, but at the cost of introducing a specially tailored coherent evolution. The steady state of the resulting Lindblad evolution is exactly given by \rho_{\beta} , which follows from the fact that the Lindbladian satisfies a certain [quantum] detailed balance property [ 33 , 34 , 35 , 36 ] . The sole remaining unknown is the mixing time \tau_{\rm mix} of the Lindbladian, which controls the approach to the steady state and can be long for physically meaningful reasons much as in the classical case. Thus, although Gibbs samplers are unlikely to speed up problems that are classically hard because of glassy landscapes [ 37 , 38 , 39 , 40 , 41 , 42 , 43 ] (such as generic optimization problems), they are not limited by the sign problem or entanglement growth which are the usual “intrinsically quantum” obstacles to simulating many-body physics. Figure 1: Schematic of one time step of the local driving sampler. In each step, the “bath” qubits are initialized in the all-zero state, followed by an entangling step consisting of time evolution with the system Hamiltonian H together with a dynamically driven system-bath coupling V(t)=\sum_{i}V_{i}(t) [cf. Sec. III ]. Finally, the bath qubits are measured and the outcomes discarded. The final unitary “rewinding” e^{iHT} can be dropped with only a modest increase to the error bound, and can be ignored with no penalty if a quasiparticle picture applies. Despite this remarkable breakthrough, convenient protocols to implement these exact Gibbs samplers (or their close approximations) in physically realistic settings relevant to current and near-term quantum hardware remain largely unexplored (but see e.g. Refs. 44 , 45 for concurrent work). Proposed approaches rely on block-encoding Lindbladians [ 27 , 29 ] ; doing so at appreciable scale is likely out of the question for the near term. It is this gap between formal theory and practical implementation that we bridge here, while bolstering a physically motivated picture with mathematical rigor. Specifically, we devise a protocol which approximately prepares thermal states of an arbitrary Hamiltonian, with controlled error \epsilon , using a total Hamiltonian simulation time scaling as \tilde{O}(\beta\tau_{\rm mix}^{3}\epsilon^{-2}) . Crucially, we only require three relatively standard features of the current generation of quantum devices: (a) “analog” simulation of the n_{\rm S} -qubit Hamiltonian; (b) a dynamically tunable coupling to an (arbitrary) number of n_{\rm B} ancilla qubits [usually, we will take n_{\rm B}=\order{n_{\rm S}} ]; and (c) the ability to reset the ancillas (see Fig. 1 ). The proposed protocol implements a controlled discrete-time approximation of a certain Lindblad evolution. The corresponding Lindbladian has exactly the same dissipative part as the exact sampler, but the “wrong” coherent part: a Lamb shift correction rather than the coherent dynamics necessary for exact thermal state preparation. However, using the so-called secular approximation [ 46 , 47 , 5 ] , we show that this difference in coherent evolution only incurs a controlled, arbitrary small error per Lindbladian simulation time step. Assuming that the mixing time, and hence the simulation time necessary to reach the steady state, is short, this implies also that the fixed point of our protocol is close to the thermal state. In practice, the mixing time generally depends on parameters of the protocol which are also used to control the error per time step. This limits the minimum achievable fixed-point error. Developing a better understanding of , and eventually overcoming, these limitations—which also impede the practicality of the exact samplers—remains an important open question raised by our work. Aficionados of near-term quantum algorithms will recognize a resemblance between our