Release 2026.8.27

Release 2026.8.27

Release 2026.8.27 introduces multiple fixes and new capabilities. We highlight (1) a new AFIR + MACE reaction path discovery capability; (2) a new Jupyter notebook demonstrating defect formation energy calculations; and (3) a step-by-step tutorial reproducing band structure calculations for twisted MoS2 bilayers.

Feature

• Implement AFIR + MACE reaction path discovery
• Create NB+WF for Defect Formation Energy Calculation
• Create NB for HOMO/LUMO Calculation + Frequency
• Create CLI job example of NEB calculation
• Relaxation SWF/WF for NWChem
• Files endpoint on the external REST API - create and delete
• Add documentation tutorials for the newly added notebooks

Improvement

• [Wode] Save k-grid to persisted context
• [Standata] Make name search deterministic
• Plan + execute the documentation updates

Bugfix

• Fix material import from Bank
• [Made] Fix metadata behavior
• [WA] Avoid storing ALL platform account references in bankWorkflowHistory and other bank collections

Developers

• Make InMemoryEntity generic to eliminate per-subclass _json / toJSON declarations
• Remove the need for dist in JS Development
• [WA] Reuse getIcon() for Material

Highlight 1: AFIR + MACE Reaction Path Discovery

A new notebook discovers reaction paths by driving the reaction with the artificial force induced reaction (AFIR) method on top of the MACE machine-learning force field. With the bias removed, the maximum along the path gives the transition-state guess, and the forming and breaking bond distances show the mechanism directly:

The reactant, transition state and product are then inspected side by side in the notebook - here for a Claisen rearrangement of allyl vinyl ether into 4-pentenal:

Highlight 2: Defect Formation Energy Calculations

A new notebook and workflow compute the neutral defect formation energy of a defective supercell through a multi-material DFT workflow, with elemental chemical potentials taken from Standata reference materials:

The workflow reports the formation energy directly on the platform, alongside the usual total-energy breakdown, atomic forces and stress tensor:

Highlight 3: Band Structure of Twisted MoS2 Bilayers

A step-by-step tutorial reproduces the interlayer-coupling result for twisted molybdenum disulfide bilayers, following the stacking configurations and band gaps reported in the literature:

Figure from Kaihui Liu, Liming Zhang, Ting Cao, Chenhao Jin, Diana Qiu, Qin Zhou, Alex Zettl, Peidong Yang, Steve G. Louie and Feng Wang, "Evolution of interlayer coupling in twisted molybdenum disulfide bilayers", Nature Communications 5, 4966 (2014). DOI: 10.1038/ncomms5966

Running the notebook across structures that differ only in interlayer distance reproduces the published trend: the indirect gap shifts by 0.199 eV over 0.4 A, against 0.20 eV in the manuscript, while the K-valley gap moves only 0.012 eV. The indirect gap tracks the interlayer distance and the K-valley gap does not:

‍

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

‍