Anthropic-Harvard paper: LLM-evolved quantum circuits beat known designs on 7 of 8 tasks
dair_ai · x · 2026-10-17
A paper from Anthropic, Harvard, and colleagues uses LLM-driven evolutionary search to discover quantum circuit designs.
Method (built on OpenEvolve):
- An LLM (Claude Mythos 5) proposes circuit-generating programs and architecture rewrites
- An independent evaluator verifies each candidate circuit's correctness and scores cost (ranking by Toffoli count × peak qubit count)
- Best designs seed the next evolutionary round
Results:
- On 7 of 8 tasks, discovered circuits significantly beat the state of the art, reducing both Toffoli gates and qubits
- Tasks span numerical computation, chemistry, optimization, quantum simulation, and number theory
- On the eighth task, the search found a new reversible architecture for prime-field modular inversion, making divsteps — a classical algorithm impractical for reversible arithmetic — viable and improving the Toffoli-qubit trade-off
- The evaluator verifies every circuit before scoring, so each reported gain is checked
The authors say this opens the door to evaluator-guided LLM co-design of new algorithmic structures for quantum computation.
Related event: Anthropic: Claude Designs Quantum Circuits Beating Best Published Results(14 posts)→
More from Research
- AstraZeneca strikes $40 million AI drug discovery deal with insitro — HealthcareLdr · 2026-10-17
- Fizgig LoRA Isolation: 0/50 face bleed-through vs 48/50, likeness kept — shootthesound · 2026-10-17
- NVIDIA's open midtraining recipe lifts SWE-bench Multilingual from 55.8 to 62.8 with 500B tokens — dair_ai · 2026-10-17
- Audit of 160,860 citations: nearly 40% of links shown by Chinese AI search go unused — Sanbi_Ai · 2026-10-17
- Dogs show the Bouba–Kiki effect, study finds — abenitezburraco · 2026-10-17
- rasbt's Reasoning From Scratch Finale: Distilling Reasoning Models Hands-On — rasbt · 2026-10-17