Neural network solution of non-Markovian quantum state diffusion and operator construction of quantum stochastic...

Area of Science:

• Quantum Physics
• Machine Learning
• Computational Chemistry

Background:

• Open quantum systems require advanced modeling techniques.
• Non-Markovian quantum state diffusion offers a wavefunction-based framework.
• Existing machine learning methods often approximate wavefunctions or expectation values.

Purpose of the Study:

• Introduce a novel machine learning approach for quantum dynamics.
• Develop an operator construction algorithm using neural networks.
• Demonstrate broader applicability beyond wavefunction approximations.

Main Methods:

• Employed a neural network as a universal generator.
• Reconstructed the stochastic time evolution operator from quantum trajectories.
• Utilized an operator construction algorithm.

Main Results:

• Successfully benchmarked the algorithm on the spin-boson model across diverse spectral densities.
• Demonstrated high accuracy in modeling quantum dynamics.
• Showcased utility in calculating absorption spectra and reduced density matrices.

Conclusions:

• Established a new paradigm for machine learning in quantum dynamics.
• The operator-based approach offers broader applicability to stochastic processes.
• The method accurately models open quantum systems over extended timescales.