Claude Designs Smaller Quantum Circuits Across Eight Tasks

Anthropic said in a company research post that Claude designed more efficient quantum circuits across eight tasks in chemistry, optimization, physics simulation, number theory and numerical computing. Anthropic says the designs improve on the best published circuits, in some cases by an order of magnitude or more; The Quantum Insider reports that seven of the eight cut both gates and qubits against the study's baselines.

Claude Mythos 5 generated circuit implementations in Python and received automated feedback on correctness, qubit use and gate counts. It then revised its designs without human input between rounds. Work on an elliptic curve arithmetic design ran for 53 hours, with human involvement limited to initial instructions and occasional prompts to explore more approaches.

According to The Quantum Insider's reading of the paper, for a decoder used in decoded quantum interferometry, Claude replaced the extended Euclidean algorithm with Berlekamp-Massey decoding and reconstructed temporary information instead of storing it. The resulting circuit used about 440,000 Toffoli gates, down from 5.1 million, while peak qubit use fell from 1,938 to 1,837.

Claude also produced a reversible modular inversion method that recovered earlier computational decisions from the values remaining in the circuit. One configuration used 12% more qubits than the smallest previous design but required 17 times fewer gates. The results are based on software evaluation and resource estimates, not tests on quantum hardware, and the paper, posted on arXiv, has not been peer reviewed.