High-Performance Computational Photochemistry and Spectroscopy

Ziel

The future of modern sustainable technologies lies in the exploitation of solar energy and in harnessing the sun’s practically infinite energy. For fundamental as well as target-oriented research in this direction, computer simulations of the ongoing photochemical processes and electronic spectroscopy of the underlying molecular materials are indispensable. While for smaller molecules, computational photochemistry can nowadays provide highly accurate results and reliable predictions, the limit in applicable molecular system size is quickly reached, and the obtained results often come with an unpredictable error requiring a posteriori validation. Indeed, we are lacking efficient and sufficiently accurate and reliable excited-state ab initio methods reaching out for organic molecular systems with more than 500 second-row atoms.
In HIPERCOPS, we aim at closing this gap by deriving highly efficient and genuinely parallel ab initio methods for the calculation of excited electronic, electron-detached and electron-attached states for execution on modern high-performance computer architectures, whose full potential is impossible to leverage by existing standard quantum chemical program packages. To address this problem, we choose the algebraic-diagrammatic construction (ADC) family of methods, since these schemes offer clear advantages, for instance, numerical stability, easy to use, predictable accuracy. We will exploit novel genuinely parallel concepts and solution strategies for ADC schemes to enable them for HPC architectures. Our developed methods and resulting easy-to-use software will thus push the boundaries of accurate and predictable computational photochemistry to unprecedented molecular system sizes enabling and promoting research in, for example, the areas of functional optoelectronic devices and photovoltaics, molecular solar thermal energy conversion, solar-driven nanomachines, towards efficient molecular harnessing of sun light.