arXiv:2609.18841
A first introduction to Matrix Product State algorithms for the integration of Lindblad equation
Christophe Chatelain
Abstract
In this introductory review, we present and compare four algorithms for the numerical integration of the Lindblad equation for one-dimensional quantum lattice systems. All four methods are based on Matrix Product State representations and can be viewed as extensions of the Time-Evolving Block Decimation (TEBD) algorithm to open quantum systems. Two approaches directly integrate the vectorized Lindblad equation, one of them explicitly enforcing the positivity of the density matrix. The other two rely on stochastic unravelings of the Lindblad equation, namely the quantum trajectory and quantum state diffusion approaches. We discuss the principles, numerical implementation, accuracy, and computational efficiency of the different methods, and benchmark them against an exactly solvable free fermion model.
AI Summary
This introductory review addresses how to integrate the Lindblad equation for one-dimensional quantum lattices using matrix-product-state methods. It develops four TEBD-based implementations: direct evolution of a vectorized density matrix as an MPDO, a locally purified representation enforcing positivity, quantum-jump trajectories, and quantum-state diffusion. The presentation includes implementation details, computational scaling, and distinctions between time-discretization, tensor-truncation, and sampling errors. The methods are benchmarked on an initially empty, boundary-driven free-fermion chain with $N=24$ sites, particle injection at one end, and removal at the other. Density and current profiles are compared with reference results from closed equations for two-point correlations, integrated with a tenfold smaller time step. Tests include MPDO bond dimensions up to $\chi=64$, time steps $\Delta t=0.01,0.003,0.001$, and stochastic calculations illustrated with $\chi=16$ and 3,000 trajectories. Direct MPDO evolution is the most computationally efficient method in this benchmark. The simplified locally purified implementation is costlier and less accurate. State diffusion has smaller sampling fluctuations than quantum jumps and lower deviations at early times, but its systematic errors grow; their deviations become comparable near $t\simeq15$. The conclusions are explicitly model-dependent. Only simple, lowest-order implementations are compared; higher-order or more elaborate algorithms may change the outcome. Positivity violations in direct MPDO evolution are not measured, and entanglement-growth comparisons are left for further study. Increasing trajectory counts cannot remove the late-time discretization and truncation errors observed in state diffusion.
Error found: The printed vectorized Liouvillian is inconsistent with the paper's explicit vectorization map, most importantly replacing the required recycling factor L_k⊗L_k* with L_k⊗L_k†. The locally purified recycling update in equation MatrixR omits γ_i Δt, so the printed update does not approach the identity as the time step tends to zero. The displayed effective Hamiltonian for the correlation benchmark omits right-boundary damping at N=24 because it contains δ_{N,1} where δ_{i,N} is required.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.