Research

We aim to develop our own AI methods to explain chemical phenomena and solve optimization problems in chemistry. Using graph-based machine learning methods, HWK can predict molecular properties and contribute to research projects related to the improved method. Our AI method will be helpful in drug discovery (optimization of molecular structures and properties) as we previously applied AI to optimize reaction conditions. We are currently developing AI methods that can be applied to quantum simulators.

We develop MD simulation methods capable of efficiently handling quantum effects and applying them to explain interesting chemical phenomena. We are particularly interested in the mixed quantum-classical (MQC) approach. Participating in research projects related to the MQC approach to explain photosynthetic energy transfer in an atomistic model, HWK has become an expert in an approximate MQC approach. In the future, we will develop a practical MQC method to simulate electrocatalytic reactions in the presence of heterogeneous catalysts.

Existing quantum chemistry software can be used to describe chemical phenomena based on electronic structure calculations. We select systems and perform calculations based on the interests of group members and our collaborators. Currently, we are interested in computationally characterizing heterogeneous catalysts with composite structures. AI can be applied to these computational data as well if a sufficient amount of data is accumulated.