Quantum Chemistry and Chemical Compound Space Design

My research combines quantum chemistry, high-throughput computation, and machine learning to understand how molecular structure determines properties and to guide the design of new compounds. The QM9 dataset [1] and our work on Δ-machine learning [2] helped establish data-driven approaches to predicting quantum chemical properties across large molecular spaces. We have applied these ideas to inverse design in the BODIPY chemical space [3] and to understanding singlet–triplet energy gaps in azaphenalenes [4].

Computational NMR is a particular focus. We developed the QM9-NMR dataset and benchmarks for ¹³C shielding prediction [5], and have since improved predictions using neighborhood-informed molecular representations [6]. This work connects local chemical environments to NMR shielding and supports the interpretation of molecular structure and conformation.

Structural ordering and phase transitions are another area of interest. We investigate how atomic arrangements, structural distortions, and symmetry breaking influence the electronic properties of materials. We combined high-throughput electronic-structure and phonon calculations to identify Peierls distortions and charge-density-wave phases in quasi-one-dimensional organometallic materials [7]. Using an Ising-type additivity model, we studied the dependence of semiconductor bandgaps on stacking sequences in long-period polytypes [8]. These studies connect microscopic structural patterns and lattice instabilities to emergent electronic behavior.

My broader interests include quantum mechanical descriptions of molecular behavior, such as real-time electron dynamics and quantum interference [9]. I also apply mathematical and computational modeling to biological questions, including how neural microcircuits can be distinguished from their patterns of activity [10].