Many-body theory of antimatter interactions with molecules

Antimatter particles have the same mass but opposite electrical charges to their ordinary matter counterparts. Antimatter quickly annihilates in contact with ordinary matter - that is particle/antiparticle pairs are destroyed and an equivalent amount of energy is emitted. This property is used in current applications of positrons (anti-electrons) such as PET scans in hospitals and technology for characterisation of new materials. This eCSE project focusses on profiling, optimisation and enhancement of “EXCITON+”, a new and currently unrivalled code developed by project PI Dermot Green’s group since 2020 for ab initio many-body theory calculations of antimatter interactions with atoms and molecules. EXCITON+ is built on the underlying all-electron EXCITON code (authored by Co-I Charles Patterson). EXCITON+ can calculate positron binding energies, annihilation rates and Doppler-broadened annihilation gamma spectra with first principles ab initio inclusion of important positron-molecule correlations including polarization and virtual-positronium formation. Its first application resulted in quantitative understanding of a long-standing question of positron binding in polyatomic molecules [Hofierka et al, 2022, Nature 606, 688-693]. Here, we report on progress made on profiling and targeted optimization of EXCITON+ coding, with particular emphasis on opportunities for nested MPI/OpenMP parallelism and passive remote memory access to distribute parallelized tasks. Ongoing performance-enhancing work to introduce point group symmetry-aware calculation of molecular leptonic (positron + electrons) integrals is described. The overall aim is to ensure that EXCITON+ modules are computationally as efficient and future proof as the scientific content of the code is ground-breaking and comprehensive.