Well-tempered metadynamics: a smoothly converging and tunable free-energy method

Area of Science:

• Computational chemistry
• Statistical mechanics
• Molecular dynamics

Background:

• Accurately determining free-energy landscapes is crucial for understanding molecular processes.
• Existing methods like metadynamics and canonical sampling have limitations in efficiency and control.
• Adaptive bias methods offer potential for improved sampling and free-energy calculations.

Purpose of the Study:

• To present a novel adaptive bias method for calculating free-energy dependence on collective variables.
• To provide a unified theoretical framework encompassing metadynamics and canonical sampling.
• To enable rigorous control over convergence and errors in simulations.

Main Methods:

• Development of a new adaptive bias formalism.
• Implementation of tunable simulation parameters to focus computational effort.
• Application of the algorithm to reconstruct the free-energy landscape of an alanine dipeptide.

Main Results:

• The proposed method successfully determined the free-energy landscape.
• The formalism demonstrated a unified description of existing methods.
• Convergence and error control were shown to be rigorous and manageable.
• Computational effort was effectively focused on relevant order parameter space regions.

Conclusions:

• The adaptive bias method offers a powerful and flexible approach for free-energy calculations.
• This formalism enhances the efficiency and reliability of molecular simulations.
• The method provides a unified framework with improved control over simulation parameters.