Despite the rising accessibility of quantum hardware, simulation of quantum circuits will remain an important tool for algorithm development, analysis and debugging. Simulators utilising GPUs have shown considerable success in this field, but the ecosystem is supported by only a few accelerator hardware vendors. We propose investigating the performance and scaling of AMD Instinct™ family accelerators for state vector and tensor network circuit simulators.

Hybrid quantum-classical workflows constitute one of the most promising and realistic use cases in the NISQ era. In these, the quantum device is the limiting factor for task scheduling as it is shared among multiple classical resources. We propose studying novel methods of task-based and job-based scheduling for coupled workflows with one or more NISC devices in a large HPC system setting, utilising well-established tools from the HPC ecosystem.

For the foreseeable future, NISQ devices will be predominant. Simulation environments do permit utilising custom noise models, but performance and scaling, as well as parametric behaviour under variable noise models of more ‘realistic’ quantum simulations is less understood compared to ideal circuits. We propose investigating, benchmarking, and improving existing noisy simulation tools with custom models for specific hardware and compare the outcomes with results obtained on actual NISQ devices.