Transforming generalized Ising models into Boltzmann machines

Area of Science:

• Statistical Mechanics
• Computational Physics
• Condensed Matter Theory

Background:

• Generalized Ising models feature complex many-spin interactions.
• Boltzmann machines offer a simplified framework with only two-spin interactions.
• Efficient simulation of complex models is crucial for understanding physical phenomena.

Purpose of the Study:

• To establish an exact mapping between generalized Ising models and Boltzmann machines.
• To leverage this mapping for enhanced computational efficiency in simulations.
• To investigate the impact of this method on critical slowing down.

Main Methods:

• Utilized algebraic transformations, specifically star-triangle and decoration-iteration transformations.
• Mapped many-spin interactions to equivalent two-spin interactions using auxiliary variables.
• Applied the mapping in conjunction with the Swendsen-Wang algorithm for Monte Carlo simulations.

Main Results:

• An exact mapping was found from generalized Ising models to equivalent Boltzmann machines.
• The transformation reduces the complexity of spin interactions.
• Significant reduction in critical slowing down was observed for a Kagomé lattice model.

Conclusions:

• The developed mapping provides a powerful tool for analyzing complex spin systems.
• This approach offers substantial benefits for Monte Carlo simulations.
• The method effectively mitigates critical slowing down, enabling more efficient studies of phase transitions.