Reducing acquisition time and radiation damage: data-driven subsampling for spectro-microscopy

What is it about?

X-ray absorption spectroscopy (XAS) is a technique for investigating the chemical state within a material. Spectro-microscopy combines XAS and microscopy to also understand the spatial variations of these properties in materials. In a typical spectro-microscopy experiment, the sample is rastered through the beam and the absorption for each pixel is measured. This is repeated for beams with varied energies. The long acquisition time and radiation damage to the sample pose the two main limitations to this powerful technique. In this work, we propose a novel measurement strategy to significantly reduce the number of measurements necessary, without sacrificing spatial or spectral resolution. The subsampling strategy has theoretical foundations in the CUR decomposition, a low-rank approximation tool for matrices for (randomized) numerical linear algebra.

Featured Image