Projects in the Business Area Virtual Material Design

Our projects include research, development, and applications in:

  • Multiscale modeling and numerical simulation for material science, chemistry and nanotechnology.
  • High performance computing in quantum mechanics, molecular dynamics and continuum mechanics.

© Fraunhofer SCAI

MaGriDo – Mathematics for Machine Learning Methods for Gaph-Based Data with Integrated Domain Knowledge

MaGriDo's goal is to (further) develop and analyze deep neural networks (NNs) for industrial problems, which allow existing domain knowledge to be incorporated into the architecture of the networks. Such a hybrid approach can make use of the complementary strengths of "end-to-end" learning approaches and "a-priori models/rules". This approach promises more efficient solutions for many fields of application. For example, the amount of data required is reduced, or the predictions of the ML model are consistent with existing knowledge.

 

The focus of research and development in MaGriDo is on so-called graph networks, since complex systems can usually be represented very well as compositions of entities and their interactions. These contain various special cases such as conventional fully-connected NN, convolution NN and recurrent NN, can be applied to relational structures, and make a hierarchical processing of input data possible.

Project duration: 04/2020 - 03/2023

More Info

SONAR – Better Batteries for Electricity from Renewable Energy Sources

Organic redox flow batteries (RFBs) are considered to be a promising approach for the storage of temporary surpluses of renewable energies. The SONAR project aims at capturing the entire development process with all relevant aspects in a digital manner in order to accelerate the screening for suitable substances and the optimization of a battery system's design for specific operating conditions. The project partners develop tools and workflows for investigating electroactive materials up to whole battery systems. To this end, they combine simulation methods on different physical scales – ranging from the electronic/quantum mechanical level to the visible, macroscopic behavior. Factors such as cost, lifetime and performance are taken into account in order tto compare competing energy storage technologies comprehensively.
Project duration: 01/2020 - 12/2023

More Info

Deep Learning for Virtual Material Design

Empirical analysis potentials and ab-initio methods such as density function theory have been the pillars of computer-aided materials science. With theoretical advances in machine learning and the rapid increase in computing power, data-based approaches have emerged a new class of models with the goal of combining the predictive power of ab-initio methods and the computational efficiency of empirical potentials.

Standard machine learning techniques such as kernel learning (e.g. for the Gaussian approximation potential), deep neural networks (e.g. neural network potentials by Behler et al.), and generalized linear models (e.g. for momentum tensor potentials) have been employed to develop fast and accurate force fields from data without the need for human knowledge about the underlying chemistry.

In this project we develop high-quality, easy-to-use implementations of such machine learning potentials and investigate possibilities to improve the existing approaches by utilizing modern tools from the constantly growing toolbox of data science.

Human Brain Project

In the »Human Brain Project« (HBP), which is funded by the European Commission, leading research institutions work together to better understand the human brain. For that purpose the project partners will develop new simulation methods, for example on high performance computers. The project aims to develop new therapeutic approaches for the treatment of brain diseases and new methods of High Performance Computing. Fraunhofer SCAI participates in a HBP sub-project through researchers of its Department of Virtual Material Design (VMD). In particular, they develop software, numerical algorithms and methods to realize neuroscientific simulations on high performance computers. SCAI’s researchers provide their expertise in multiscale simulation and numerical simulation for Molecular Dynamics.

More Info