Adaptive partitioning molecular dynamics using an extended Hamiltonian approach.

Area of Science:

• Computational chemistry
• Molecular dynamics simulations
• Statistical mechanics

Background:

• Switching interaction potentials is crucial for various molecular simulations.
• Existing extended Hamiltonian methods require careful parameterization for adaptive partitioning.
• Efficient and stable methods are needed for complex simulations.

Purpose of the Study:

• To generalize an extended Hamiltonian approach for adaptive partitioning molecular dynamics (APMD).
• To develop an adaptive fictitious mass strategy for consistent switching time scales.
• To ensure stability and accuracy of APMD simulations over extended periods.

Main Methods:

• Generalization of an extended Hamiltonian approach for switching potentials.
• Introduction of a fictitious classical degree of freedom for potential mixing.
• Adaptive determination of the fictitious mass to maintain a constant switching time scale.
• Application to model systems (harmonic oscillator, Lennard-Jones fluid) in microcanonical and canonical ensembles.
• Integration with Nosé-Hoover chain thermostat for canonical ensemble simulations.

Main Results:

• The proposed adaptive partitioning method ensures a constant time scale for switching processes.
• The extended Hamiltonian is conserved for thousands of consecutive switching events.
• Simulations in both microcanonical and canonical ensembles demonstrate stability.
• A modified Hamiltonian prevents the accumulation of numerical errors.
• Constant temperature is maintained throughout canonical ensemble simulations.

Conclusions:

• The generalized extended Hamiltonian approach enables stable and accurate adaptive partitioning molecular dynamics.
• Adaptive fictitious mass is key to maintaining consistent switching dynamics.
• The method offers a robust framework for complex simulations requiring potential switching.