Hybrid HPC and quantum computing framework for multiscale modelling of chemical systems

We present a multiscale workflow that couples quantum computing algorithms for quantum chemical calculations with conventional High-Performance Computing (HPC) software. The resulting embedding framework, implemented in the multiscale computational chemistry environment ChemShell, combines a generic N-layer subtractive Quantum Mechanics/Molecular Mechanics (QM/MM) scheme and active orbital space embedding for the quantum computing region. We have benchmarked the embedding models for gas phase and solvated molecules to understand the performance, benefits and current limitations of embedding quantum computing algorithms within a QM/MM model. We compare the 2 performance of simulated Quantum Phase Estimation (QPE) calculations on an HPC cluster using Graphical Processing Units (GPUs) and Central Processing Units (CPUs). Using GPUs to simulate Quantum Phase Estimation requires only 16% of the runtime when compared to CPU.