Speaker
Description
Three-axes spectrometers (TAS) are considered one of the most flexible and versatile instruments for inelastic neutron scattering measuring fundamental excitations (phonons, magnons, and crystalline electric field) in materials with sub-meV energy resolution. However, the limitation of a three-axes spectrometer usually comes from the restriction of a point-by-point measurement strategy in the reciprocal space which does not allow investigations in short time scales. To increase the data collection efficiency in terms of (Q,ω)-coverage of a conventional TAS instrument, multiplexing techniques have been developed (such as CAMEA at PSI, UFO-IN12 at ILL, BAMBUS-PANDA and Multiplexing-PUMA at MLZ). In these techniques for an incoming neutron energy E$_i$ in the primary side of the spectrometer, several (Q, E$_f$)-channels are employed to cover a larger range of (Q,ω) in the secondary side of the spectrometer. In parallel, there is also a growing interest in optimizing and enhancing the measurement strategy at a conventional TAS by employing artificial intelligence. Here we will be presenting a concept to generalize an active learning-assisted TAS measurement [1] for multiplexing setups and show its potential for improving the experimental strategy for users.
Reference: [1] M. T. Parente, G. Brandl, C. Franz, U. Stuhr, M. Ganeva, and A. Schneidewind. Active learning-assisted neutron spectroscopy with log-Gaussian processes. Nat. Comm. 14, 2246 (2023).
