February 1, 2019
Special Symposium: Materials Informatics for Society 5.0
Prediction of higher-order structure and properties of polymeric materials by combining machine learning and computational simulation
National Institute of Advanced Industrial Science and Technology (AIST)
Research Center for Computational Design of Advanced Functional Materials
Principal Research Manager
Dr. Takeshi Aoyagi
I had been involved in computational modeling of mainly polymeric materials in chemical industry for 29 years since Apr.1987. After moving to AIST in May 2016, I'm continuing a research of polymer modeling as well as playing a roll of bridging between industry and national institute.
Higher-order structures of polymeric materials such as phase separated structures and filler filled structures are critical for functions and properties of the materials. We will introduce a combining approach of computational simulation and machine learning taking such a higher-order structure into consideration. We will introduce new functions and applications of extended OCTA system which is under development in a NEDO project "Ultra High-Throughput Design and Prototyping Technology for Ultra Advanced Materials Development Project (U2M)".
Mathematics and Materials Science
Advanced Institute for Materials Research
Prof. Yasumasa Nishiura
Y. Nishiura is currently Project Professor, Advanced Institute for Materials Research (AIMR), Tohoku University and Research Advisor of MathAM-OIL.
He also serves as associate editors of Physica D, SIADS, SIAM Interdisciplinary Journal "Multiscale Modeling and Simulation", and a series editor for Mathematics of Planet Earth, Springer. https://www.wpi-aimr.tohoku.ac.jp//nishiura_labo/index-e.html
Mathematics and materials science coevolve for many centuries starting from the symmetry of crystals to more advanced mathematics in nanotechnology and spintronics.
In this talk we focus on two cases. One is a characterization of amorphous structures or heterogeneities arising in polymers that lies in between perfect crystals and random structure. The question here is what is an adequate mathematical descriptor for this. Machine learning combined with topological data analysis clarifies some aspects of it. The other is self-organization process of nanoparticles consisting of mixture of copolymers and homopolymers. To find an appropriate free energy is a key to understand the whole process to a variety of final morphologies. Several Cahn-Hilliard system of equations will be presented.
Materials Informatics: State-of-the-Art and Future Perspectives
The Institute of Statistical Mathematics, Research Organization of Information and Systems
Data Science Center for Creative Design and Manufacturing
Prof. Ryo YOSHIDA
In general, material spaces are considerably high-dimensional. For instance, the chemical space of small organic molecules is known to contain as many as 10^60 candidates, whereas the total number of currently identified chemical structures is at most 10^8. The problem entails a complicated multi-objective optimization where it is impractical to fully explore the vast landscape of structure-property relationships. The emergence of machine intelligence trained on massive amounts of data has been expected to accelerate the pace of expanding the frontier in the vast universe of materials. In this talk, I describe state-of-the-art and future perspectives of materials informatics.
"Polymer Informatics" in view of Industrial Materials Development
Mitsubishi Chemical Corporation
R&D Planning Department
Dr. Hisao Takeuchi
After joining Mitsubishi Chemical Industry, he studied computational materials science of polymer and composite materials in universities. His specialty is chemical physics and computational materials science of polymer materials. He has been engaged in this field since the early days of utilization of these technologies in industry. The current scope is utilization of MI in products development.
In recent years, Materials Informatics (MI) has been highlighted as a new approach to materials research and development. It can be said, however, that reports on "Polymer MI" have been very limited in view of development of polymeric materials in industry. Provided that "Polymer MI" is considered as "application of data science to polymer development", it has a broad spectrum according to a targeted product. This is because a property of a polymer material is dependent on the chemical structure of its monomer in some cases, but on the other hands, other properties may be governed by the morphology dependent on its composition and/or processing conditions. In this talk, "Polymer MI" will be discussed in view of this hierarchical nature of polymer materials.