Awesome
Table of Contents
- Introduction
- Basics
- ES tutorial
- FF/MC related tutorial
- JARVIS-School
- AI tutorial
- QC tutorial
- NanoHub tutorial
- References
- How to contribute
- Correspondence
- Funding support
- License
<a name="intro"></a>
JARVIS-Tools Notebooks (Introduction)
The JARVIS-Tools Notebooks is a collection of Jupyter/ Google-Colab notebooks to provide tutorials on various methods for materials design. It consists of several types of applications such as for electronic structure (ES), force-field (FF), Monte Carlo (MC), artificial intelligence (AI), quantum computation (QC) and experiments (EXP). This project is a part of the NIST-JARVIS infrastructure. A few more detailed tutorial are also available at: JARVIS-Tools.
<a name="basics"></a>
Basics
A few preliminary notebooks before diving into materials design
- Python beginners notebook
- For absolute Beginners in ML using Python
- Silicon atomic structure and analysis example
<a name="es"></a>
Electronic structure
Density functional theory, Tight-binding and Beyond-DFT methods using various software
- Analyzing_data_in_the_JARVIS_DFT_dataset
- Analyzing_data_in_the_JARVIS_Leaderboard
- Basic quantum espresso run
- JARVIS_Optoelectronics_Computational_screening_of_high_performance_optoelectronic_materials_using_OptB88vdW_and_TB_mBJ_formalisms
- JARVIS_TopologicalSpillage_High_throughput_Discovery_of_Topologically_Non_trivial_Materials_using_Spin_orbit_Spillage
- JARVIS_Solar_Accelerated_Discovery_of_Efficient_Solar_Cell_Materials_Using_Quantum_and_Machine_Learning_Methods
- Si_bandstructure&densityof_states
- JARVIS_Wannier90Example
- BoltztrapExample
- Making 2D heterostructures
- JARVIS_DFT_FormationEnergiesAccuracyCheck
- Downloading raw analysis data and input/output files
- JARVIS_DFT_2D_High_throughput_Identification_and_Characterization_of_Two_dimensional_Materials_using_Density_functional_theory
- JARVIS_CONVERG_Convergence_and_machine_learning_predictions_of_Monkhorst_Pack_k_points_and_plane_wave_cut_off_in_high_throughput_DFT_calculations
- JARVIS_DFPT_High_throughput_Density_Functional_Perturbation_Theory_and_Machine_Learning_Predictions_of_Infrared,_Piezoelectric_and_Dielectric_Responses
- JARVIS_TE_Data_driven_discovery_of_3D_and_2D_thermoelectric_materials
- JARVIS_ELAST_Elastic_properties_of_bulk_and_low_dimensional_materials_using_van_der_Waals_density_functional
- Get JARVIS-DFT final structures in ASE or Pymatgen format
- JARVIS_Solar_Accelerated_Discovery_of_Efficient_Solar_Cell_Materials_Using_Quantum_and_Machine_Learning_Methods
- JARVIS_TopologicalSpillage_High_throughput_Discovery_of_Topologically_Non_trivial_Materials_using_Spin_orbit_Spillage
- JARVIS_Optoelectronics_Computational_screening_of_high_performance_optoelectronic_materials_using_OptB88vdW_and_TB_mBJ_formalisms
- JARVIS_QuantumEspressoColab_Designing_High_Tc_Superconductors_with_BCS_inspired_Screening,_Density_Functional_Theory_and_Deep_learning
- JARVIS_WTBH_Database_of_Wannier_tight_binding_Hamiltonians_using_high_throughput_density_functional_theory
- Comapre_MP_JV
- ParsingWebpages(JARVIS_DFT)
- ConvexHull
- DimensionalityAndExfoliationEnergy
- Element_filter_for_JARVIS_DFT_dataset
- Run GPAW on Google-colab and calculate interface energy with jarvis-tools
- ParsingWebpages(JARVIS_DFT)
- WTBH_MagneticMats.ipynb
- QMCPACK_Basic_Example
<a name="ff"></a>
Force-field
- JARVIS_LAMMPS
- MLFF SNAP training
- ALIGNN-FF for energy and forces
- ALLEGRO training for Silicon
- Analyzing_MOF_datasets
<a name="ai"></a>
Artificial intelligence/Machine learning
AI models for chemical formula, atomic structures, image and text for both forward and inverse design. Some of the methods include descriptor/feature based, graph based and transformers based designs.
- Analyzing data in JARVIS-Leaderboard
- ML_Chem_Formula_Descriptors
- Basic_Machine_learning_training_example_with_CFID_descriptors
- JARVIS_ML_TrainingGPU
- JARVIS_CFID_LightGBM_GPUvsCPU
- JARVIS_ML_TensorFlowExample
- JARVIS_Leaderboard_MatMiner
- CIF To Graph_example
- JARVIS_ALIGNN_Basic_Training_example
- ALIGNN-FF pretrained model/calculator usage
- JARVIS_Leaderboard_contribution_ALIGNN
- JARVIS_Leaderboard_MLFF/ALIGNN-FF for Silicon
- JARVIS_Leaderboard_KGCNN
- ALIGNN Superconductor training
- Train ALLEGRO-FF for Silicon
- Train NEQUIP-FF for Silicon
- Train CHGNet-FF for Silicon
- MatGL-FF_Mlearn for Silicon
- SNAP-FF_Mlearn for Silicon
- Pretrained CHGNet Prediction
- Pretrained OpenCatalystProject Model
- ALIGNN-Pretrained property-predictor models
- ALIGNN-PhononDos
- Inverse design of superconductors with CDVAE
- JARVIS_STEM_2D
- AtomVision_Leaderboard_Example
- AtomVision_Example
- ChemNLP example
- ChemNLP HuggingFace example
- AtomGPT training example
- AtomGPT HuggingFace inference example
- Open catalyst project load model
- Vacancy formation ML
- Interface Materials Design/InterMat example
- ALIGNN-FF Unified force-field structure relaxation
- Basic external tutorial on linear models
<a name="qc"></a>
Quantum computation
- With new qiskit package version: Quantum computation and Qiskit based electronic bandstructure
- With old qiskit package version: Quantum computation and Qiskit based electronic bandstructure
<a name="nanohub"></a>
NanoHub FAIR workflow workshop/JARVIS-School @ Purdue university
https://nanohub.org/FAIR_workshop_2024
- Learn a basic DFT calculation
- Once you run a lot of these, you can make a database, and analyze trends (AKA Exploratory Data Analysis)
Analyzing_data_in_the_JARVIS_DFT_dataset
- These datasets can also be used to develop fast surrogate machine learning models
JARVIS_Leaderboard_contribution_ALIGNN
- Beyond single property prediction models, they can be used to train machine-learning force-fields as well
ALIGNN-FF for energy and forces
- While the above MLFF was trained for single element system, a more generalized model was developed with JARVIS-DFT diverse dataset, and the developed model can be used for fast atomic structure optimization and phonon etc. property predictions
ALIGNN-FF Unified force-field structure relaxation
- While the above ML models were for forward design, we can use AtomGPT for inverse design as well
AtomGPT HuggingFace inference example
- Other optional notebooks for the tutorial session
Quantum computation and Qiskit based electronic bandstructure
AIMS2024 tutorial and presentation slides
https://github.com/usnistgov/aims2024_workshop
<a name="school"></a>
JARVIS-School
<a name="aps"></a>
APS2023 tutorial
<a name="aims"></a>
AIMS2022 tutorial
<a name="refs"></a> References
The joint automated repository for various integrated simulations (JARVIS) for data-driven materials design & various other publications
<a name="contrib"></a> How to contribute
For detailed instructions, please see Contribution instructions
<a name="corres"></a> Correspondence
Please report bugs as Github issues(preferred) or email to kamal.choudhary@nist.gov.
<a name="fund"></a> Funding support
NIST-MGI (https://www.nist.gov/mgi).
Code of conduct
Please see Code of conduct