A Periodic Table of Molecules
Dr. Sandip De from the Laboratory of Computational Science and Modelling wins the best poster award in materials science at PASC 2016 conference, co-sponsored by the Association for Computing Machinery (ACM) and the Swiss National Supercomputing Centre (CSCS), held from June 8 to 10, 2016 at EPFL in Lausanne, Switzerland.
Funded by the NCCR MARVEL, and in collaboration with scientists in Cambridge, UK, Dr. Sandip De is working on building intuitive and meaningful "maps" of materials and molecular systems, representing the relations between thousands of configurations. This approach provides a unique way of visualizing the whole database at once with both structural similarity and materials property correlation information. The mathematical formulations behind building these maps, which based on measuring similarity between two given structures have also been used in machine learning algorithms for predicting materials properties with a ground braking accuracy of 0.5 kcal/mol. The results have been published as cover article in PCCP. Fully interactive version of these maps are also available at www.sketchmap.org