scientific big data
Tilde: scientific research and consulting in materials informatics
Materials informatics is a junction of computational materials science and computer science, focusing on data processing and analytics for the new materials. This comprehends software development, high-throughput (database-driven) scientific simulations, and screening of scientific data.
- computational research for perspective materials,
- high-throughput database-driven ab initio simulations,
- full-stack software development for the scientific data repositories.
Managing the big amounts of scientific data, sharing them online, automation of computational research — all these topics are the applications of materials informatics. Do not hesitate to contact for details.
Materials platform for data science
Materials platform for data science (MPDS) is an online version on the well-known PAULING FILE materials database. More than a million of data entries, containing phase diagrams, crystalline structures, and physical property values, are available for download. The source of these entries are about 300,000 peer-reviewed publications in materials science, processed during the last 16 years by an international team of PhD editors. The results are presented online with a quick search interface. The basic access is provided for free.
Online simulated materials database
The present online simulated materials database contains: (a) selected data prepared by Evgeny Blokhin, (b) open-access data prepared by Zhongnan Xu, Jan Rossmeisl, and John Kitchin, and (c) tutorial datasets for simulation packages Quantum ESPRESSO and CRYSTAL. In total there are more than 5000 ab initio calculations done using CRYSTAL, VASP, and Quantum ESPRESSO packages. The database serves mainly demonstration purposes. It is powered by fully open-source database software, and can be created automatically at any Unix PC for arbitrary simulation data of the above mentioned packages.
Encyclopedia of the perspective energy materials
In this project nearly one hundred materials, related to the up-to-date commercial applications (according to the public specifications), are considered at the ab initio level. The state-of-the-art theoretical approximations are adopted. To reduce computational costs, only the ideal bulk prototype systems are modeled. All the results, including raw and intermediate data, and all the detailed software workflows, will be published as the open-access online interactive encyclopedia for educational purposes, allowing anyone to represent and enhance the results. When possible, the open-source modeling packages are employed. The two projects below are relevant.
Server-free scientific web-applications for the browser
With the ubiquitous penetration of Internet, browsers became powerful and comfortable web-application platform. However the server is normally required, implying such drawbacks as software complexity, traffic overhead, and privacy issues. These drawbacks are avoided in the presented web-applications. They visualize crystalline structures in common formats (see also other solutions) and identify the space group. The only browser and no plugins are required. The web-applications are open-source and ready to be used as the parts of more complex software.
Artificial intelligence techniques for materials data mining
In a narrow sense, a special-purpose artificial intelligence is the number of computer science techniques, copying the certain cognitive aspects of the human brain. Among them the deep learning and logic reasoning are of the special interest. They were already applied in materials design and chemoinformatics. The present study aims to review the existing experience and recognize the most promising use cases. Using these two artificial intelligence techniques, this study will offer the open-source solutions for some practical materials science problems, concerned with the big amount of data. As an introduction, the tutorial to logic reasoning and the toy model of interlocutor are available.
NoMaD: Novel Materials Discovery Repository
This project was the collaboration of Fritz Haber Institute of the Max Planck Society and Humboldt University of Berlin (Germany) with the aim to create an international ab initio materials science data repository called NoMaD. The database structure was designed, and the repository program core was implemented. In 2014 the first version of NoMaD user interface was launched. In 2015 NoMaD project was successfully funded by European Union's infrastructure call for Centers of Excellence (CoE) in computational sciences.
Defect thermodynamics of mixed conductors
Water adsorption on perovskite surfaces
The Quantum Chemistry Chair of St. Petersburg State University (Russia) had worked on several international projects granted by NATO, DFG, INTAS, and CRDF in fundamental research. One of the projects was to combine classical force-field and ab initio modeling techniques to describe water adsorption on the surfaces of SrXO3 perovskites (X = Ti, Zr, Hf), owing to its high technological importance. During this project, the number of utility software tools for data processing were created. For details please refer to Evarestov, Bandura, Blokhin, J. Phys. Conf. Ser. (2007) and Evarestov, Bandura, Blokhin, Surf. Sci. (2008).
Research and consulting
We explore the screening strategies for materials design using the artificial intelligence techniques, such as deep learning and semantic technologies. Let the advanced data analytics serve for materials.
You manage big amounts of scientific data or want to share data online? Or process your data heavily (e.g. with Python)? It is time to invest in your own data laboratory.