GradNav: Accelerated Exploration of Potential Energy Surfaces with Gradient-Based Navigation

Area of Science:

• Computational Chemistry
• Molecular Dynamics
• Biophysics

Background:

• Understanding molecular system behavior requires exploring potential energy surfaces (PES) to identify metastable states.
• Transitions between these states often involve high energy barriers, necessitating extensive computational simulations.

Purpose of the Study:

• To introduce and evaluate the gradient-based navigation (GradNav) algorithm for accelerated PES exploration.
• To enhance the efficiency of molecular simulations and improve PES reconstruction.

Main Methods:

• Developed the GradNav algorithm, utilizing short simulation runs from updated starting points to navigate energy barriers.
• Introduced two performance metrics: deepest well escape frame (DWEF) and search success initialization ratio (SSIR).
• Applied GradNav to Langevin dynamics on Müller-type PESs and molecular dynamics of the Fs-peptide protein.

Main Results:

• GradNav demonstrated enhanced ability to escape deep energy wells, indicated by reduced DWEF values.
• The algorithm showed reduced reliance on initial conditions, evidenced by increased SSIR values.
• Improved exploration capability led to more precise energy estimations from simulation trajectories.

Conclusions:

• GradNav significantly accelerates the exploration of potential energy surfaces.
• The algorithm offers a more efficient approach to molecular simulations, reducing computational costs.
• GradNav facilitates more accurate characterization of molecular systems and their dynamics.