Vectorization of the density matrix and quantum simulation of the von Neumann equation of time-dependent Hamiltonians (2306.08775v4)

Abstract: Based oh the properties of Lie algebras, in this work we develop a general framework to linearize the von-Neumann equation rendering it in a suitable form for quantum simulations. We show that one of these linearizations of the von-Neumann equation corresponds to the standard case in which the state vector becomes the column stacked elements of the density matrix and the Hamiltonian superoperator takes the form $I\otimes H-H^\top \otimes I$ where $I$ is the identity matrix and $H$ is the standard Hamiltonian. It is proven that this particular form belongs to a wider class of ways of linearizing the von Neumann equation that can be categorized by the algebra from which they originated. Particular attention is payed to Hermitian algebras that yield real density matrix coefficients substantially simplifying the quantum tomography of the state vector. Based on this ideas, a quantum algorithm to simulate the dynamics of the density matrix is proposed. It is shown that this method, along with the unique properties of the algebra formed by Pauli strings allows to avoid the use of Trotterization hence considerably reducing the circuit depth. Even though we have used the special case of the algebra formed by the Pauli strings, the algorithm can be readily adapted to other algebras. The algorithm is demonstrated for two toy Hamiltonians using the IBM noisy quantum circuit simulator.