Non-markovian neural quantum propagator and its application to the simulation of ultrafast nonlinear spectra

Jiaji Zhang,Lipeng Chen

2024-10-08

Abstract:The accurate solution of dissipative quantum dynamics plays an important role on the simulation of open quantum systems. Here we propose a machine-learning-based universal solver for the hierarchical equations of motion, one of the most widely used approaches which takes into account non-markovian effects and nonperturbative system-environment interactions in a numerically exact manner. We develop a neural quantum propagator model by utilizing the neural network architecture, which avoids time-consuming iterations and can be used to evolve any initial quantum state for arbitrarily long times. To demonstrate the efficacy of our model, we apply it to the simulation of population dynamics and linear and two-dimensional spectra of the Fenna-Matthews-Olson complex.

Chemical Physics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to develop an efficient and general - purpose solver for simulating non - Markovian dynamics in open quantum systems. In particular, it aims to build a Neural Quantum Propagator (NQP) model based on neural networks to solve the Hierarchical Equations of Motion (HEOM). Although existing iterative methods can accurately describe the non - Markovian effects and non - perturbative interactions between the system and the environment, they have problems of high computational cost and long - term numerical instability. Therefore, the authors propose an NQP model using machine - learning techniques, aiming to avoid the time consumption of traditional iterative methods and be able to evolve any initial quantum state to any long time, thereby improving computational efficiency and stability.

Specifically, the goals of the paper include:
1. **Develop an efficient NQP model**: By using neural network architectures, especially the Fourier Neural Operator (FNO), build a model that can quickly and accurately simulate the time evolution of quantum systems.
2. **Reduce the computational cost in the data preparation stage**: Introduce a super - resolution algorithm to build a high - resolution NQP model relying only on low - resolution data, and use a Physics - Informed Loss Function (PILF) to systematically improve the inherent errors in the data set.
3. **Verify the effectiveness of the model**: By applying the NQP model to the population dynamics of the Fenna - Matthews - Olson (FMO) complex and the simulation of linear and two - dimensional spectra, demonstrate its performance and accuracy in different computational scenarios.

The achievement of these goals can not only significantly improve the computational efficiency of non - Markovian quantum dynamics, but also provide new tools and methods for future quantum dynamics research.