{"id":"arxiv/2604.11314","url":"https://arxiv.org/abs/2604.11314","tags":["quant-ph"],"score":0,"title":"Risk-aware neural pulse compilation of continuously parameterized single-qubit gates with optimal-control benchmarking and NMR hardware validation","domain":"quant-ph","source":"arxiv","authors":["Arash Fath Lipaei","Ebrahim Khaleghian","Gani Göral","Zidong Lin","Selin Aslan","Özgür E. Müstecaplıoğlu"],"reviews":0,"subject":"quant-ph","summary":"Pulse-level compilation can reduce the depth and calibration overhead associated with decomposing continuously parameterized quantum operations into fixed gate libraries, but conventional optimal control requires a new optimization for each target and can be sensitive to hardware uncertainty. We introduce a physics-informed neural compilation framework that learns a reusable map from continuously parameterized single-qubit gate targets directly to experimentally admissible control waveforms. The compiler is trained end-to-end through differentiable quantum dynamics without precomputed…","category":"research","comments":0,"disputed":false,"division":"physics","mentions":0,"discussed":0,"published":"2026-04-13","reactions":{}}