Release 2024.2.15

Overview

Release 2024.2.15 contains multiple new features as well as some UX improvements. We highlight (1) the JupyterLite Environment for running arbitrary Python code, including Pymatgen, ASE, JARVIS-tools, the numpy/scipy, plotly stack, and no need to install/setup anything; (2) a JupyterLite notebook demonstrating how to construct interfaces with strain matching per Zur and McGill, DOI:10.1063/1.333084; (3) the addition of DeepMD application with corresponding executables + a default DeePMD MLFF Bank workflow using Quantum ESPRESSO cp.x to produce training data; (4) a default bank workflow for Quantum ESPRESSO with cp.x.

Feature

Generic

• Addition of a JupyterLite Environment for running arbitrary Python code, including Pymatgen, ASE, JARVIS-tools, the numpy/scipy/pandas/plotly stack, with no need to install anything

Specific

• Add a JupyterLite notebook Zur-McGill interface builder with strain minimization
• Create a default bank workflow for DFT+U with hp.x example calculation
• Add support for Quantum ESPRESSO cp.x Car-Parinello molecular dynamics calculations
• Add DeepMD application with corresponding executables + add DeePMD MLFF default Bank workflow
• Add Intel Oneapi 2023.1.0
• Add QE version 7.3
• Add QE 7.2 compiled with intel oneapi 2023.1.0
• Add libxc v6.2.2

‍

Improvements

• Swap all RJSF usages to use MUI
• Migrate jQuery-QueryBuilder to React Query Builder

Bug

• Fix list_files on AWS
• Fix fast queues scaling logic
• Fix user creation

For Developers

• Build JARVIS-to-ESSE converter in Express-Py
• Make Jarvis-tools available in Pyodide
• Adjust property endpoints in api-client
• Publish tests definitions library (@mat3ra/tede) using Cypress
• ESSE: Build resolved schemas and make available at  
• Build ESSE JavaScript and Python modules with JS runtime

Highlight: JupyterLite Environment in Materials Designer

We highlight the ability to run arbitrary Python code, including Pymatgen, ASE, JARVIS-tools, the numpy/scipy, plotly stack, and no need to install/setup an environment for it. Below is a quick demonstration.

‍

Highlight: ZSL strain matching in JupyterLite

We highlight the ability to construct interfaces with strain matching inside the Materials Designer per Zur and McGill, DOI:10.1063/1.333084. In the example below we create an interface by placing Graphene on top of a surface of Ni (111).

Highlight: DeepMD machine-learned force field (MLFF) workflow

We introduced support for neural-network-based machine-learned forcefields (MLFF) construction workflows combining ab-initio calculations with QE cp.x, ML with DeepMD, and molecular dynamics with LAMMPS. A step-by-step tutorial including a voiceover with a video screenshare are available in https://youtu.be/daTwJyMPIvE and https://docs.mat3ra.com/tutorials/ml/deepmd-mlff-with-espresso-cp-and-lammps/

‍

Highlight: Quantum ESPRESSO cp.x

Here's the new workflow flavor allowing one to use cp.x to run Car-Parinello molecular dynamics with Quantum ESPRESSO.

‍

Try the new functionality online at https://platform.mat3ra.com/

‍