FLAME

Official Resources

• Homepage: https://flame-code.gitlab.io/FLAME/
• Source Repository: https://github.com/flame-code/FLAME
• Documentation: https://flame-code.gitlab.io/FLAME/
• License: GPL-3.0

Overview

FLAME (Fast Library for Atomistic Modeling Environments) is a software package designed for performing a wide range of atomistic simulations to explore the potential energy surfaces (PES) of complex condensed matter systems. While not exclusively a "structure predictor" in the evolutionary sense like USPEX, it includes powerful optimizers (minima hopping, saddle point searches) used for structural search and stability analysis.

Scientific domain: Atomistic modeling, PES exploration, structure optimization
Target user community: Computational physicists, materials scientists

Theoretical Methods

• Minima Hopping: Global optimization method to find low-energy structures.
• Molecular Dynamics: For sampling free energy landscapes.
• Saddle Point Search: Nudged Elastic Band (NEB) and other transition state methods.
• Local Minimization: L-BFGS, FIRE, etc.
• Cell Optimization: Variable cell relaxation.

Capabilities

• Global Optimization: Minima hopping for structure prediction.
• Transition Paths: Finding pathways between stable structures.
• Neural Network Potentials: Interfaces for machine learning potentials (High-Dimensional Neural Network Potentials).
• Calculator Interfaces: Built-in LJ/Morse, interfaces to DFT (BigDFT, etc.) and classic codes.
• Parallelization: MPI/OpenMP hybrid parallelization.

Inputs & Outputs

• Input formats: YAML-based or custom input format, coordinate files (XYZ, GEN).
• Output data types: Trajectories, minimized structures, energy logs.

Interfaces & Ecosystem

• Calculators: Interfaces with BigDFT, GULP, LAMMPS (via library or file).
• Machine Learning: Support for Centrin Neural Network potentials.

Verification & Sources

• Confidence: ✅ VERIFIED
• Primary Source: FLAME GitHub
• Reference: M. Amsler et al., "FLAME: a library of atomistic modeling environments", Comput. Phys. Commun. 256, 107415 (2020).