Ozaki, Taisuke

Biography

In accordance with development of recent massively parallel computers, first-principles calculations based on density functional theories (DFT) have been playing a very important role in understanding and designing properties of a wide variety of materials. We have been developing efficient and accurate methods and software packages to extend applicability of DFT to more realistic systems as discussed in industry. Although the computational cost of the conventional DFT method scales as the third power of number of atoms, we have developed an O(N) Krylov subspace method, of which computational cost scales only linearly, based on nearsightedness of electron. The O(N) method enables us to simulate Li ion battery, structural materials, and graphene nanoribbon based devices which cannot be easily treated by the conventional method, and to directly compare simulations with experiments. In addition to this, we are aiming at realization of materials design from first-principles. As a first step towards the materials design, we have been trying to develop a method to predict complicated crystal structures based on machine learning techniques. Our continuous methodological developments have been all implemented in OpenMX (Open source package for Material eXplorer), which has been released to public under GNU-GPL, and widely used around world for studies of a wide variety of materials.

Research Interest

Development of efficient methods and algorithms for first-principles electronic structure calculations Development of first-principles electronic transport calculations First-principles calculations of two-dimensional Si structures Development of first-principles methods of core-level binding energies in solids Development of the OpenMX software package