Compressed Sensing of 2D IR Using Spectroscopic Models

Published in International Conference on Coherent Multidimensional Spectroscopy, 2022

Recommended citation: Chi-Jui Ho, Mason L Valentine*, Wei Xiong, and Nick Antipa, "Compressed Sensing of 2D IR Using Spectroscopic Models," in The 10th International Conference on Coherent Multidimensional Spectroscopy, 2022



Abstract: For systems with weak chromophores, coherent multidimensional spectroscopy acquisition times can be on the order of hours to days, which can make experiments challenging, particularly for systems that evolve over time, such as metastable structures and biological samples. Standard compressed sensing techniques have previously been applied to multidimensional spectroscopy to reduce the necessary amount of data and speed up experiments, generally under the assumption that 2D spectroscopic data is low rank, sparse, or both.1–3 In a separate line of investigation, Kubo lineshape fitting has demonstrated utility for robustly characterizing 2D spectra even in the presence of a low signal-to-noise ratio.4 We have combined the approaches above into a technique that utilizes Kubo lineshape models to faithfully reconstruct 2D spectra using a combination of nonuniformly subsampled time domain data and initial parameters from linear and pump-probe spectra. Compared to conventional methods, our method reduces the necessary number of t1 and t2 samples, and hence the total data collection workload, by close to an order of magnitude.