Google Teams Publish Proof-Based Quantum Program Validation and Optimization Method
Google DeepMind and Google Quantum AI published a preprint describing a method that uses mathematical proofs to validate quantum programs that cannot be run. [[fact-ikwmfbd6f]] In validation on three scientific simulation problems, AI-agent programs used fewer resources than previous best solutions, with the largest reduction about 90%. [[fact-2a6eww026]] For computing the arcsine function in a quantum superposition, the new solution’s cost was 1/290 of the best previously published solution. [[fact-oh8drnv3h]] The demonstrated programs are resource estimates and have not been run on a real quantum computer. [[fact-80kohnkhh]] The team said it plans to make the Lean library, problems, and verified programs public soon. [[fact-uutfy9k02]]