Simulation numérique de systèmes quantiques ouverts en grande dimension
Roussel, Julien
Abstract
Solving exactely the Lindblad equations, that model the evolution of open quantum systems, is only possible for small sizes. This is why we wish to have a range of methods enabling us to compute approximate solutions. We presente a deterministic low-rank method that is based upon a reduced approximate form of the density matrix, then a stochastic method of Monte-Carlo that gives the solution up to a statistical noise. We will first introduce an extension of the low-rank method, allowing us to use an adaptive rank while enforcing an upper bound on the error, and then a new variance reduction method. This new method speeds a lot the Monte-Carlo simulations by using the approximate solution given by the low-rank method as a control variate.