Accurate multiple time step in biased molecular simulations.

Area of Science:

• Computational Chemistry
• Molecular Dynamics Simulations
• Enhanced Sampling Techniques

Background:

• Enhanced sampling methods often rely on biasing collective variables (CVs).
• Calculating these CVs can be computationally expensive, limiting simulation speed.
• This bottleneck hinders the efficiency of molecular dynamics (MD) simulations.

Purpose of the Study:

• To develop an algorithm for treating smooth biasing forces within a multiple time step (MTS) framework.
• To accelerate MD simulations that utilize computationally demanding CVs.
• To provide a theoretical basis for assessing sampling accuracy in biased simulations.

Main Methods:

• Implementation of an algorithm for smooth biasing forces within an MTS framework.
• Integration of the algorithm into massively parallel and GPU-based MD software.
• Development of a theoretical framework to evaluate sampling accuracy.

Main Results:

• The algorithm significantly speeds up MD simulations with expensive CVs.
• Substantial performance gains are observed on parallel and GPU architectures.
• The theoretical framework allows for informed selection of integration time steps.

Conclusions:

• The proposed MTS algorithm offers a simple yet effective solution to accelerate biased MD simulations.
• This approach is particularly beneficial for large-scale computations.
• Accurate assessment of sampling accuracy is crucial for reliable simulation outcomes.