Kaitlin Smith
University of Chicago
kns@uchicago.edu
Bio
Kate is a Chicago Quantum Exchange IBM Postdoctoral Scholar at the University of Chicago under the mentorship of Prof. Fred Chong. She is a member of EPiQC: Enabling Practical-scale Quantum Computation an NSF Expedition in Computing. Her current research involves technology-aware programming computer architecture networks and high-dimensional quantum processing with a focus on quantum computing. She graduated with a PhD in Electrical Engineering from Southern Methodist University in December 2019.
Decoherence Mitigation in Quantum Computers through Instruction Scheduling
Decoherence Mitigation in Quantum Computers through Instruction Scheduling
Current quantum devices suffer from the rapid accumulation of error that prevents the storage of quantum information over extended periods. The unintentional coupling of qubits to their environment and each other adds significant noise to computation and improved methods to combat decoherence are required to boost the performance of quantum algorithms on real machines. While many existing techniques for mitigating error rely on adding extra gates to the circuit or calibrating new gates this paper’s primary contribution leverages the gates already present in a quantum program and does not extend circuit runtime duration. We exploit circuit slack for single-qubit gates that occur in idle windows scheduling the gates such that their timing can counteract some errors. TimeStitch is a novel framework that pinpoints the optimum execution schedules for single-qubit gates within quantum circuits. TimeStitch implemented as a compilation pass leverages the reversible nature of quantum computation to improve the success of quantum circuits on real quantum machines. Unlike past approaches that apply reversibility properties to improve quantum circuit execution TimeStitch boosts fidelity without violating critical path frontiers in either the slack tuning procedures or the final rescheduled circuit. On average compared to a state-of-the-art baseline a practically constrained TimeStitch achieves a mean 34% relative improvement in success rates with a maximum of 106% while observing bounds on circuit depth. When unconstrained by depth criteria TimeStitch produces a mean relative fidelity increase of 45% with a maximum of 256%. Finally when TimeStitch intelligently leverages periodic dynamical decoupling within its scheduling framework a mean 57% improvement is observed over the baseline (relatively outperforming standalone dynamical decoupling by 20%) with a maximum of 287%.