Learning from Physics Experiments with Quantum Computers: Applications in Muon Spectroscopy

Computational physics is an important tool for analyzing, verifying, and at times-replacing physical experiments. Nevertheless, simulating quantum systems and analyzing quantum data has so far resisted an efficient classical treatment in full generality. While programmable quantum systems have been developed to address this challenge, the resources required for classically intractable problems still lie beyond our reach. In this work, we consider a new target for quantum-simulation algorithms; analyzing the data arising from physics experiments specifically, muon-spectroscopy experiments. These experiments can be used to probe the quantum interactions present in condensed-matter systems. However, fully analyzing their results can require classical computational resources scaling exponentially with the simulated system size, which can limit our understanding of the studied system. We show that this task may be a natural fit for the coming generations of quantum computers. We use classical emulations of our quantum algorithm on systems of up to 29 qubits to analyze real experimental data, and to estimate both the near-term and error-corrected resources required for our proposal. We find that our algorithm exhibits good noise resilience, stemming from our desire to extract global parameters from a fitted curve, rather than targeting any individual data point. In some respects, our fault tolerant resource estimates go further than some prior work in quantum simulation, by estimating the resources required to solve a complete task, rather than just to run a given circuit. Taking the overhead of observable measurement and calculating multiple datapoints into account, we find that significant challenges still remain if our algorithm is to become practical for analyzing muon-spectroscopy data.