Accelerating Quantum Materials Characterization: Hybrid Active Learning for Autonomous Spin Wave Spectroscopy

Published

April 26, 2026

Author(s)

William Ratcliff

Abstract

Autonomous neutron spectroscopy must solve three distinct tasks: detection (where is the signal?), inference (which Hamiltonian governs it?), and refinement (what are the parameters?). No single controller solves all three equally well. We present TAS-AI, a hybrid agnostic→physics-informed framework for autonomous triple-axis spin-wave spectroscopy that separates these tasks explicitly. In blind reconstruction benchmarks, agnostic baselines reach a global relative-error threshold more reliably and with fewer measurements, establishing model-agnostic discovery as the correct front end for unknown spectra. Once signal structure is localized, the physics-informed stage performs in-loop Hamiltonian discrimination and parameter refinement: in a controlled NN-vs-1-2 test, TAS-AI reaches decisive AIC-derived evidence ratio >100 in fewer than 10 measurements, and motion-aware scheduling reduces wall-clock overhead on the same measurement budget. We further identify a failure mode of posterior-weighted experimental design—algorithmic myopia—in which the planner over-refines the current leading model while under-sampling low-intensity falsification probes,and show that a constrained falsification channel sharply reduces wrong-leader dwell and accelerates decisive recovery. In controlled ablations, a simple deterministic max-disagreement rule and an LLM committee both achieve this gain under identical constraints, confirming that the active ingredient is the falsification principle; the LLM offers additional generality across diverse problem descriptions without per-problem engineer. We demonstrate the full workflow in silico using a high-fidelity digital twin and provide an open-source Python implementation.

Citation

arvix

Pub Weblink

https://arxiv.org/abs/2604.23821

Pub Type

Websites

Keywords

autonomous experiments, neutron scattering, spin waves

Citation

Ratcliff, W. (2026), Accelerating Quantum Materials Characterization: Hybrid Active Learning for Autonomous Spin Wave Spectroscopy, arvix, [online], https://arxiv.org/abs/2604.23821 (Accessed July 2, 2026)

Additional citation formats