White nights of materials science: From physics and chemistry to data analysis, and back
Saint Petersburg, Russia
In recent years, the rapid development of computer hardware and electronic-structure software has made first-principles materials design not only feasible, but also a powerful and cost-effective alternative to experimental screening approaches. It has also opened up new possibilities and with them new challenges, some of which will be addressed at this conference (http://th.fhi-berlin.mpg.de/sitesub/meetings/wnms2014/).
There are several efforts worldwide using database construction and analysis to assist in the design of new materials. Recently, the Materials Genome Initiative  has been launched in the USA. Ceder et al. [2,3] have started and are continuing to develop the related Materials Project, a database aimed at accelerating materials discovery in all application areas that use computed data. Norskov et al.  are developing a database of heterogeneous catalytic materials and reactions, CatApp, which is the first element of the planned quantum materials informatics project . The software platform AiiDA ("Automated interactive infrastructure and Database for Atomistic simulations")  is being developed by Kozinsky and Marzari, to automate submission of computational tasks, and storage and validation of the results. Within the international collaborative research program CoSMIC , Rajan et al. [8,9] are applying multivariate analysis and information-theory concepts (e.g., Shannon entropy) to find descriptors and design new functional materials. Neugebauer et al.  are developing SAPIENS, a DFT- and experiment-based thermophysical database for pure elements. In their present form, the above efforts are not sufficient for several reasons: (i) Typically only a limited amount of data is stored; thereby valuable information is lost, which greatly hinders the discovery and understanding of correlations between properties of different materials. (ii) The existing databases do not allow for complex queries of the data, and are not designed to store very large amounts of data. (iii) The first important steps towards data analysis in the materials sector [8,9] are so far limited to only a few examples and these initiatives are decoupled from large databases. The constructive extraction of hidden information from the enormous amount of data that can be generated is an essential aspect that has so far been largely ignored.
The planned workshop will address these issues. It is intended to bring together researchers who are experts in electronic-structure theory or database structure/analysis, interested in both fields. We intend to create an environment where communication between people with different expertise and backgrounds is strongly encouraged, in order to understand and overcome barriers on all sides. Advances in electronic-structure method development, multiscale modelling, materials data analysis, and machine learning will be discussed in close connection with technologically important applications, such as heterogeneous catalysis, thermoelectric materials, optoelectronics, hybrid organic-inorganic interfaces, and others. The goals of the workshop are to identify: both common and application-specific features of a comprehensive materials data structure, specific areas where data analysis would be especially helpful, and the availability or absence ("white spots") of data.
Evgeny Blokhin (Tilde Materials Informatics) - Organiser
Luca Ghiringhelli (NOMAD Laboratory at the Fritz Haber Institute of the Max Planck Society and Humboldt University, Berlin) - Organiser
Matthias Scheffler (Fritz-Haber-Institut der Max-Planck-Gesellschaft) - Organiser
Sergey Levchenko (Skolkovo Institute of Science and Technology, Moscow, Russia) - Organiser