Using Reinforcement-Learning for Optimization of Ansätze for Hybrid Quantum-Classical Workflow Study of Periodic Material Systems

Nevertheless, the study of binding to periodic material systems is crucial for real industry applications. We take corrosion inhibition as a first proof of concept for such systems as it plays an important role in automotive and aerospace industries. We apply our workflow - combining classical Density Functional Theory (DFT) calculations with quantum algorithms through CP2K code and Qiskit integration – to calculate the binding energies.

As we aim to get results not only running on simulators but also on real quantum devices - like IQM Garnet and IonQ Aria1 – by using error mitigation techniques. We need also to optimize the underlying ansätze with theory and with applying machine learning techniques, like reinforcement learning. These optimizations will give us the opportunity to enlarge the active space and thus, will result in more accurate calculations.

These results then will be benchmarked against the gold standard using UCCSD as ansatz.