AI material prediction platform launched
The platform, called Xaedra, was created to address today’s materials discovery process, which, says the company, is tedious, expensive, and slow. Current approaches rely on sophisticated trial and error and some level of simulation, resulting in only a small number of all known compounds having been characterized, says the company, while most materials remain “undiscovered” for potential uses.
Xaedra is offered as the first AI platform to successfully predict – not simulate – a wide variety of properties for over 50,000 known compounds.
“Xaedra first creates atomistic “fingerprints” of the 50,000 known compounds in its database,” says Dr. Pawel Pisarski, the creator of Xaedra. “The user then can define a property of interest – mechanical, chemical, thermodynamic, electrical or other. Like any machine learning system, some level of measured data must be loaded into the database, then using the “fingerprints” the neural network is trained and Xaedra makes a prediction of the property for the remaining compounds.”
Charlie Baker P.E., leader of Xaedra Business Development adds, “We believe that Xaedra will both enable and disrupt a wide range of technical fields, including energy storage, electronics, solar cells, catalysts, lightweight structures, and many others. Prospecting for breakthrough materials can be done at the computer, and lab resources redeployed to confirmation and validation. We think this can both level the playing field between entrepreneurs and established companies as well as spark a wave of innovation in material science.”
A limited number of Beta user opportunities are currently available.
Related articles:
Open materials database advances battery design
New cathode material could triple lithium battery capacity
Material found that harnesses multiple forms of energy simultaneously
Nanostructures of “impossible” materials open up perspectives for optoelectronics
New material class promises zero-ohm conductivity
New materials promise thinner, more energy efficient circuits
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
