OPTIMADE Customary Revolutionizes Entry to Materials Databases for Sustainable Expertise Growth

At SINTEF, we work to create it supplies for brand new sustainable know-how wanted by the business, akin to batteries, photo voltaic cells, LED lighting and bio-materials. The merchandise will need to have exact properties, be sustainable and include the least quantity of drugs which are harmful to the atmosphere and your well being.

Designing new merchandise is now sooner because of highly effective computer systems and superior calculations. A number of totally different materials databases have been developed that present info on the properties of the varied supplies. These datasets can be utilized to coach AI fashions to optimize merchandise.

– To create new, correct AI fashions, we’d like a whole lot of information from all related databases. Nevertheless, the totally different databases have totally different interfaces and methods of representing the info. This has made it troublesome to make use of them collectively, says Casper Welzel Andersen, researcher at SINTEF.

Simpler entry to materials databases

Over the previous eight years, a brand new customary has been developed for sharing digital information about objects; OPTIMADE (Integration of open databases for materials design). The usual permits customers to speak with totally different materials databases in the identical means. It’s based mostly on web-based protocols and developed in collaboration with massive materials databases in Europe and the USA. There’s a massive worldwide community with greater than 30 establishments worldwide, together with SINTEF, behind it.

– With OPTIMADE, researchers at universities or in business can simply talk with all these databases and perceive the knowledge they obtain, says Gian-Marco Rignanese, professor on the Division of Condensed Issues and Nanoscience at UCLouvain in Belgium.

Giant quantities of disparate information

Highly effective computer systems are used at this time to make superior simulations of how electrons transfer in supplies, which offers details about the varied properties of supplies. These calculations present a considerable amount of information that can be utilized to coach machine studying fashions. The AI ​​fashions can then predict the properties of recent supplies based mostly on new calculations.

Nevertheless, a whole lot of information is required to coach the fashions. Over time, many of those databases have been collected from numerous analysis teams and tasks. Every of those databases has its personal methodology of study, its personal means of describing the info, and its personal means of describing what the info represents. With OPTIMADE, information from massive simulations, and normal information about supplies, can be found by way of the identical protocol. On this means, analysis and information illustration turn out to be synonymous.

– The newest model of OPTIMADE gives a a lot improved potential to precisely describe the properties of varied supplies utilizing clear machine-readable definitions. With these, we are able to perceive information from the totally different databases and use them collectively, says Casper Welzel Andersen.

Giving researchers extra time for scientific work

The OPTIMADE customary is a extremely valued device for researchers in supplies science at SINTEF.

– In considered one of our AI tasks, we confronted challenges in acquiring experimental outcomes and simulation outcomes to agree some vital metallurgical information. It turned out that though these two communities have been speaking about the identical issues for a number of years, it was troublesome to shut the hole between them. Options that OPTIMADE brings to the desk may help us spend extra time on scientific and technical challenges and fewer time coping with issues which are “misplaced” in translation, says researcher Daniel Marchand .

The OPTIMADE customary has already been used within the EU-funded OntoTrans venture. An Open Translation Surroundings has been developed right here, which is a toolbox for understanding and fixing enterprise issues associated to totally different supplies. The toolbox makes use of, amongst different issues, OPTIMADE to provide the consumer entry to a variety of calculation outcomes. OntoTrans is now in its ultimate stage, however the toolbox has already been pursued within the Semantic Supplies enterprise consortium, the place SINTEF is likely one of the important companions.