Exponentially reduced circuit depths in Lindbladian simulation

Wenjun Yu, Xiaogang Li, Qi Zhao, Xiao Yuan · 2024

Read Exponentially reduced circuit depths in Lindbladian simulation by Wenjun Yu, Xiaogang Li, Qi Zhao, Xiao Yuan (2024) 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-2412-21062.txt.

Description

Abstract Quantum computers can efficiently simulate Lindbladian dynamics, enabling powerful applications in open system simulation, thermal and ground-state preparation, autonomous quantum error correction, dissipative engineering, and more. Despite the abundance of well-established algorithms for closed-system dynamics, simulating open quantum systems on digital quantum computers remains challenging due to the intrinsic requirement for non-unitary operations.

Catalog record

CREATORWenjun Yu, Xiaogang Li, Qi Zhao, Xiao Yuan
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE33 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-21062

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Exponentially reduced circuit depths in Lindbladian simulation Wenjun Yu, Xiaogang Li, Qi Zhao, Xiao Yuan Abstract Quantum computers can efficiently simulate Lindbladian dynamics, enabling powerful applications in open system simulation, thermal and ground-state preparation, autonomous quantum error correction, dissipative engineering, and more. Despite the abundance of well-established algorithms for closed-system dynamics, simulating open quantum systems on digital quantum computers remains challenging due to the intrinsic requirement for non-unitary operations. Existing methods face a critical trade-off: either relying on resource-intensive multi-qubit operations with experimentally challenging approaches or employing deep quantum circuits to suppress simulation errors using experimentally friendly methods. In this work, we challenge this perceived trade-off by proposing an efficient Lindbladian simulation framework that minimizes circuit depths while remaining experimentally accessible. Based on the incoherent linear combination of superoperators, our method achieves exponential reductions in circuit depth using at most two ancilla qubits and the straightforward Trotter decomposition of the process. Furthermore, our approach extends to simulate time-dependent Lindbladian dynamics, achieving logarithmic dependence on the inverse accuracy for the first time. Rigorous numerical simulations demonstrate clear advantages of our method over existing techniques. This work provides a practical and scalable solution for simulating open quantum systems on quantum devices. Introduction.— Quantum simulation of open quantum systems is one of the most promising applications of quantum computing. To effectively describe these dynamic processes, researchers frequently employ the Lindblad master equation, which models large environments as memoryless baths under the Markovian approximation [ 1 , 2 ] . This framework provides sufficient generality to capture a wide range of open-system phenomena across various fields such as quantum optics [ 3 , 4 ] , quantum biology [ 5 , 6 , 7 , 8 ] , and quantum chemistry [ 9 , 10 ] . Beyond its role in scientific inquiry, Lindbladian simulation has proven transformative for quantum computing tasks, including thermal and ground-state preparation [ 11 , 12 , 13 , 14 ] , dissipative quantum state engineering [ 15 ] , autonomous quantum error correction [ 16 , 17 , 18 ] , and the solution of differential equations [ 19 ] . The potential to simulate these complex dynamics positions Lindbladian simulations at the forefront of efforts to leverage quantum computing for advancing both foundational science and practical quantum technologies. Despite its broad applicability, implementing Lindblad simulation poses significant technical challenges due to the fundamental difficulty of performing non-unitary evolution within the inherently unitary framework of quantum circuits. Current approaches are generally divided into two categories, each with distinct limitations. The first category comprises advanced-resource methods, which utilize linear combinations of unitaries to achieve polylogarithmic circuit depths [ 20 , 21 , 22 , 23 ] . While these methods are theoretically efficient, they rely on sophisticated multi-qubit gates that exceed the capabilities of current hardware. The second category includes methods that prioritize hardware accessibility, leveraging tools such as dilated Hamiltonians [ 24 , 25 , 26 , 27 ] or Kraus expansions [ 28 , 29 ] . These approaches rely on more experimentally feasible quantum operations but are suboptimal in theory, necessitating polynomial circuit depth to suppress simulation errors. This trade-off between implementation accessibility and circuit efficiency plausibly reveals the intrinsic difficulty of simulating system-environment coupling, a challenge that has long been considered fundamental to open quantum system dynamics. Figure 1: Schematic representation of our two-stage simulation framework. (a) The hierarchical approach combines a coarse-grained simulation (Stage 1) with precision-enhancing compensation (Stage 2). The first stage constitutes the dominant part of the simulation, making the overall algorithm experimental-friendly. (b) Implementation of non-unitary dynamics in Stage 1, utilizing system-ancilla coupling for dissipative evolution in a dilated Hilbert space. (c) Core circuit architecture for Stage 2, implementing unitary conjugates to achieve high-precision compensation. Here, we demonstrate that this trade-off can be overcome by proposing an efficient simulation protocol that achieves both polylogarithmic depths and minimal resource requirements, using no more than two ancilla qubits. Our approach, depicted in Fig. 1 , employs a two-stage strategy: an accessible coarse-grained simulation followed by systematic compensation. The coarse-grained stage accomplishes the primary part of our simulation, leaving only small residual errors to be addressed. Therefore, the compensation only needs to correct high-order discrepancies, ensuring both computational efficiency and high precision. The key innovation lies in our linear combination of superoperators framework (LCS) to achieve compensation. The LCS framework provides a comprehensive representation through a unitary-conjugate basis for all Hermitian-preserving linear maps, including operations for compensating the simulation errors. Implementation is achieved through the circuit shown in Fig. 1 (c), which realizes the unitary-conjugate LCS formulas with minimal overhead. We further extend our protocol to fulfill time-dependent Lindblad simulations. Our method first discretizes the generic Lindbladians into piecewise constant segments. This new evolution can be approximated by a time-independent simulation according to the preceding protocol. Our LCS framework facilitates the further elimination of the remaining approximation error. Therefore, we construct a time-dependent Lindbladian simulation protocol with exponentially improved accuracy for the first time. Settings.— The dynamics of open quantum systems, which interact with their environment, is described by the Lindblad master equation: \displaystyle\frac{d\rho}{dt}=\underbrace{\vphantom{\left(D_{l}\rho D_{l}^{\dagger}-\frac{1}{2}\{D^{\dagger}_{l}D_{l},\rho\}\right)}-\textup{i}[H,\rho]}_{\mathcal{H}[\rho]}+\sum_{l=1}^{m}\underbrace{\left(D_{l}\rho D_{l}^{\dagger}+\frac{1}{2}\{D^{\dagger}_{l}D_{l},\rho\}\right)}_{\mathcal{D}_{l}[\rho]}, (1) where H represents the system’s internal dynamics, and jump operators \{D_{l}\} characterize how the system interacts with its environment. The evolution can be expressed through a superoperator \mathcal{L}\coloneqq\mathcal{H}+\sum_{l=1}^{m}\mathcal{D}_{l} , which may exhibit continuous time dependence. The challenge of Lindbladian simulation is to compute how an initial state \rho(0) evolves under this dynamics to reach \rho(T)=\mathcal{T}\exp(\int_0^T \mc L dt)\rho(0) . For most practical applications, it suffices to extract measurement outcomes of the final state rather than explicitly constructing it—an approach known as effective simulation . Our work presents an efficient way to effectively simulate the Lindbladian evolution. For practical implementations, we focus on systems where both H and \{D_{l}\} are q -sparse, meaning they can be expressed using at most q non-zero Pauli operators. To quantify the evolution’s strength, we define the Lindbladian norm as \|\mathcal{L}\|_{\textsf{pauli}}\coloneqq 2(\|H\|_{1}+\sum_{l=1}^{m}\|D_{l}\|_{1}^{2}) , where \|\cdot\|_{p} measures the p th order vector norm of the operator’s Pauli-decomposition coefficients. LCS Formula.— Before presenting the algorithm, we first introduce the linear combination of superoperators (LCS) formula, which provides a powerful framework for analyzing and implementing a broad class of superoperators beyond physical operations. Specifically, a LCS formula of a superoperator \mathcal{V} is a convex combination of \mathcal{V}_{i} with positive coefficients c_{i} as \mathcal{V}=\sum_{i=0}^{\Gamma-1}c_{i}\mathcal{V}_{i}=\mu\sum_{i=0}^{\Gamma-1}\Pr(i)\mathcal{V}_{i}, (2) where \mu=\sum_{i=0}^{\Gamma-1}c_{i} , the normalization factor, is defined as the 1-norm and \Pr(i)=c_{i}/\mu . The convex combination structure of the LCS formula in Eq. ( 2 ) naturally lends itself to implementation via Monte Carlo sampling. This approach involves applying superoperator \mathcal{V}_{i} to the quantum state with probability \Pr(i) , followed by a renormalization \mu of the output state. The renormalization \mu introduces additional sample overhead by amplifying statistical fluctuations to \epsilon_{s}=\order{\mu/\sqrt{N}} in contrast to \order{1/\sqrt{N}} with N rounds. For small constant \mu , however, the average of the quantum states still rapidly converges to the result of applying the desired superoperator. For tasks requiring the sequential application of multiple superoperators, we can concatenate their respective LCS formulae. This composition yields a new LCS formula that targets the overall superoperator. While more powerful basis choices {\mathcal{V}_{i}} can enhance performance, we prioritize experimental

Read the complete text