relentless: Transparent, reproducible molecular dynamics simulations for optimization

Area of Science:

• Computational chemistry
• Materials science
• Software development

Background:

• Molecular dynamics (MD) simulations are crucial for understanding materials at the atomic level.
• Optimizing objective functions within MD requires specialized tools for parameterization and analysis.
• Existing workflows can be complex, hindering transparency and reproducibility.

Purpose of the Study:

• To introduce Relentless, an open-source Python package designed to optimize objective functions computed via MD simulations.
• To provide a high-level, extensible interface for seamless integration with established MD software.
• To facilitate gradient-based optimization for computational materials design.

Main Methods:

• Development of the Relentless Python package with a focus on extensibility.
• Implementation of interfaces for native simulation setup, execution, and analysis.
• Application of Relentless within the framework of relative entropy minimization.
• Demonstration of designing pairwise interactions for targeted structure formation.

Main Results:

• Relentless successfully enables gradient-based optimization of MD simulations.
• The package facilitates the design of specific inter-particle interactions to achieve desired material structures.
• Demonstrated ability to streamline complex computational workflows.

Conclusions:

• Relentless offers a powerful and flexible platform for advancing computational materials design.
• The package enhances transparency and reproducibility in MD-driven research.
• It simplifies the optimization process, accelerating the development of new materials methodologies.