Machine-learned correction to ensemble-averaged wave packet dynamics.

Area of Science:

• Quantum mechanics
• Computational chemistry
• Theoretical physics

Background:

• The theory of open quantum systems is crucial for understanding natural processes like energy and electron transfer.
• Accurately simulating quantum dynamics for large systems is computationally challenging.
• Existing machine learning methods often require extensive training data from similarly large systems, limiting their scalability.

Purpose of the Study:

• To develop a novel machine learning approach to reduce the computational cost of simulating dissipative quantum dynamics.
• To enable accurate quantum dynamics calculations for larger systems that are currently intractable.
• To introduce a scalable and efficient method for studying complex quantum systems.

Main Methods:

• A machine-learned correction term was developed and integrated into the Numerical Integration of Schrödinger Equation (NISE) approach.
• The correction term was trained on data from smaller quantum systems where accurate calculations are feasible.
• The enhanced NISE scheme was then applied to simulate dissipative quantum dynamics for larger systems.

Main Results:

• The proposed machine-learned correction significantly improves the accuracy of dissipative quantum dynamics simulations.
• The method demonstrates feasibility for larger systems by training on data from smaller, computationally manageable systems.
• The new machine-learned correction outperforms a previously developed handcrafted correction term.

Conclusions:

• The novel machine-learned correction offers a computationally efficient and accurate method for studying dissipative quantum dynamics in large systems.
• This approach overcomes the limitations of previous machine learning techniques by enabling training on smaller systems.
• The findings pave the way for more accessible and scalable simulations in quantum chemistry and physics.