flare
development
Installation
Tutorials
Code Documentation
Frequently Asked Questions
Applications
How To Contribute
How to Cite
flare
Docs
»
FLARE: Fast Learning of Atomistic Rare Events
Edit on GitHub
FLARE: Fast Learning of Atomistic Rare Events
¶
Contents
¶
Installation
Installation of FLARE
Requirements
Installation using pip
Manual Installation with Git
Acceleration with multiprocessing and MKL
Environment variables (optional)
Compile LAMMPS with MGP Pair Style
Download
MPI
Compiling
Running
MPI+OpenMP through Kokkos
Compiling
Running
MPI+CUDA through Kokkos
Compiling
Running
Notes on Newton (only relevant with Kokkos)
Tutorials
Introduction to FLARE: Fast Learning of Atomistic Rare Events
Installation
Machine learned force fields
Training data
Training a GP model.
Task: Find hyperparameters that give a positive log marginal likelihood.
Solution
Optimizing the hyperparameters rigorously
Calculating the learning curve
Learning a force field on the fly
The Lennard Jones Potential
Perform an on-the-fly training simulation.
Step 1: Set up the initial structure.
Step 2: Set up a GP model.
Step 3: Set up an OTF training object.
Step 4: Perform the simulation!
Parsing the OTF output file.
Comparing the learned model against ground truth.
Prepare your data
VASP data
Data from Quantum Espresso, LAMMPS, etc.
Try building GP from data
Training a Gaussian Process from an AIMD Run
Roadmap Figure
Step 1: Setting up a Gaussian Process Object
Step 2 (Optional): Extracting the Frames from a previous AIMD Run
Step 3: Training your Gaussian Process
Pre-Training arguments
On-the-fly training using ASE
Step 1: Set up supercell with ASE
Step 2: Set up FLARE calculator
Optional
Step 3: Set up DFT calculator
Optional: alternatively, set up Quantum Espresso calculator
Step 4: Set up On-The-Fly MD engine
Step 5 (Optional): Resume Interrupted Training
After Training
Parse OTF log file
Construct GP model from log file
Map the GP force field & Dump LAMMPS coefficient file
Run LAMMPS with MGP pair style
Code Documentation
Bayesian Force Fields
Gaussian Process Force Fields
Predict
Helper functions for GP
FLARE ASE Calculator
Mapped Gaussian Process
Splines Methods
Formulation of Mapped Gaussian Process
Sparse Gaussian Process Force Fields
Bayesian Active Learning
On-the-Fly Training
GP From AIMD
Seed frames
Conditions to add training data
ASE MD Engine
Fake MD
LAMMPS Calculator and MD
NoseHoover (NVT Ensemble)
Descriptors
Atomic Environments
Kernels
Single-element Kernels
Multi-element Kernels (simple)
Multi-element Kernels (Separate Parameters)
Cutoff Functions
Helper Functions
File Input and Output
Output
OTF Parser
Utility
Advanced Hyperparameters Set Up
Construct Atomic Environment
Frequently Asked Questions
Frequently Asked Questions
Installation and Packages
Gaussian Processes
OTF (On-the-fly) Training
GPFA
MGP
Applications
How To Contribute
Git Workflow
General workflow
Master, development, and topic branches
Pushing changes to the MIR repo directly
Pushing changes from a forked repo
Code Standards
PEP 8
Docstrings
Tests
How to Cite
Read the Docs
v: development
Versions
latest
jon-cpp
development
Downloads
pdf
html
epub
On Read the Docs
Project Home
Builds
Free document hosting provided by
Read the Docs
.