Presentation
Compiler Optimization for Quantum Computing Using Reinforcement Learning
DescriptionSimilar to classical compilation, quantum compilation is a sequential process with many compilation steps and possible optimization passes.
In this work, we propose a framework which models the compilation flow as a Markov Decision Process and applies Reinforcement Learning to try to learn the best sequence of compilation passes.
By defining a unifying interface for each compilation step, the framework supports the combination of techniques from different compilers and optimization tools in a single compilation flow.
Experimental evaluations demonstrate an average increase in expected fidelity.
Furthermore, the compilation flow can be optimized with regard to different criteria.
In this work, we propose a framework which models the compilation flow as a Markov Decision Process and applies Reinforcement Learning to try to learn the best sequence of compilation passes.
By defining a unifying interface for each compilation step, the framework supports the combination of techniques from different compilers and optimization tools in a single compilation flow.
Experimental evaluations demonstrate an average increase in expected fidelity.
Furthermore, the compilation flow can be optimized with regard to different criteria.
Event Type
Research Manuscript
TimeTuesday, July 11th10:55am - 11:10am PDT
Location3003, 3rd Floor
Design
Quantum Computing