BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:MatterGen: a generative model for inorganic materials design - Cla
 udio Zeni\, Microsoft Research
DTSTART:20240219T140000Z
DTEND:20240219T143000Z
UID:TALK211141@talks.cam.ac.uk
CONTACT:Eszter Varga-Umbrich
DESCRIPTION:The design of functional materials with desired properties is 
 essential in driving technological advances in areas like energy storage\,
  catalysis\, and carbon capture. \nGenerative models provide a new paradig
 m for materials design by directly generating entirely novel materials giv
 en desired property constraints. \nDespite recent progress\, current gener
 ative models have low success rate in proposing stable crystals\, or can o
 nly satisfy a very limited set of property constraints. \nHere\, we presen
 t MatterGen\, a model that generates stable\, diverse inorganic materials 
 across the periodic table and can further be fine-tuned to steer the gener
 ation towards a broad range of property constraints. \nTo enable this\, we
  introduce a new diffusion-based generative process that produces crystall
 ine structures by gradually refining atom types\, coordinates\, and the pe
 riodic lattice. \nWe further introduce adapter modules to enable fine-tuni
 ng towards any given property constraints with a labelled dataset. \nCompa
 red to prior generative models\, structures produced by MatterGen are more
  than twice as likely to be novel and stable\, and more than 15 times clos
 er to the local energy minimum. \nAfter fine-tuning\, MatterGen successful
 ly generates stable\, novel materials with desired chemistry\, symmetry\, 
 as well as mechanical\, electronic and magnetic properties. \nFinally\, we
  demonstrate multi-property materials design capabilities by proposing str
 uctures that have both high magnetic density and a chemical composition wi
 th low supply-chain risk.
LOCATION:Wolfson Lecture Theatre Yusuf Hamied Department of Chemistry
END:VEVENT
END:VCALENDAR