AFLOW School for Materials Discovery

Description

Autonomous computational frameworks such as aflow++ are generating large databases that power materials discovery workflows. The aflow.org repository is one of the largest of its kind, containing more than 3.5 million compounds each characterized by 200+ different properties. The data has been employed for the discovery of new magnets (the first designed by computational approaches), superalloys, high-entropy carbides, and phase change memory compositions.

This half-day, hands-on workshop on AFLOW covers topics that include database organization and interfaces, structure prototypes and crystal symmetry, thermodynamic stability analysis, and integration of machine learning models for property prediction and descriptor development. Please bring your laptop with an internet connection.

The workshop content is broadly accessible and offers a crash course on some of our most popular web applications. All interested researchers are welcome to attend: faculty, postdocs, and graduate and undergraduate students. No prior experience in coding or computational materials science is required.

Please register as early as possible so that we have a good count for coffee and pastries. In person is highly recommended, but to attend the event virtually, please use the Zoom link.

Session 1: Machine learning

• Models: PLMF (electronic + vibrational), MFD (vibrational), and ASC (superconductivity)
• AFLOW-ML web interface
• AFLOW-ML API

Session 2: Crystallographic tools

• AFLOW-SYM
• AFLOW prototype encyclopedia
• AFLOW-XtalFinder

Session 3: aflow.org database and APIs

• Database organization
• aflow.org web portal
• AFLUX search-API

Session 4: Thermodynamic stability

• Convex hull analysis
• AFLOW-CHULL online tool

This event is hosted by Advanced Research Computing at Hopkins.