English

Gradient-based reconstruction of molecular Hamiltonians and density matrices from time-dependent quantum observables

Quantum Physics 2022-08-10 v1 Computational Physics

Abstract

We consider a quantum system with a time-independent Hamiltonian parametrized by a set of unknown parameters α\alpha. The system is prepared in a general quantum state by an evolution operator that depends on a set of unknown parameters PP. After the preparation, the system evolves in time, and it is characterized by a time-dependent observable O(t){\cal O}(t). We show that it is possible to obtain closed-form expressions for the gradients of the distance between O(t){\cal O}(t) and a calculated observable with respect to α\alpha, PP and all elements of the system density matrix, whether for pure or mixed states. These gradients can be used in projected gradient descent to infer α\alpha, PP and the relevant density matrix from dynamical observables. We combine this approach with random phase wave function approximation to obtain closed-form expressions for gradients that can be used to infer population distributions from averaged time-dependent observables in problems with a large number of quantum states participating in dynamics. The approach is illustrated by determining the temperature of molecular gas (initially, in thermal equilibrium at room temperature) from the laser-induced time-dependent molecular alignment.

Keywords

Cite

@article{arxiv.2205.05251,
  title  = {Gradient-based reconstruction of molecular Hamiltonians and density matrices from time-dependent quantum observables},
  author = {Wucheng Zhang and Ilia Tutunnikov and Ilya Sh. Averbukh and Roman V. Krems},
  journal= {arXiv preprint arXiv:2205.05251},
  year   = {2022}
}