Following the establishment of a collaborative partnership between CASUS and the Sandia National Laboratories (SNL) to create a machine learning framework – the Materials Learning Algorithms (MALA), the MALA team is now hosting its second hackathon.

MALA is a software suite leveraging artificial intelligence to significantly speed up density functional theory (DFT) calculations, which are indispensable in modern computational chemistry and materials science. Routinely employed across global academic high-performance computing systems, these DFT calculations play a pivotal role in elucidating the properties of molecules and solids. By harnessing the power of artificial intelligence, the MALA team has succeeded in accelerating the standard DFT algorithm significantly.

The current hackathon aims to refine MALA into a production-grade code, enabling efficient computations on both central processing units (CPUs) and graphical processing units (GPUs). The productive collaboration with MALA partners, including Sandia Labs and Nvidia Corp., is propelling this project forward. During the hackathon, ten participants will address programming challenges such as native support for GPUs, scalable data streaming, the incorporation of advanced neural networks like graph neural networks, ingenious data clustering methods, advanced multivariate interpolation techniques, as well as enhancements to the front-end and documentation.

The MALA team expects releasing the outcomes of this hackathon in their forthcoming MALA update this summer.