Analyzing Raman spectral data without separabiliy assumption

Abstract Raman spectroscopy is a well established tool for the analysis of vibration spectra, which then allow determination individual substances in chemical sample, or their phase transitions. In time-resolved-Raman-sprectroscopy spectra sample are recorded sequentially over time interval, such that conclusions intermediate products (transients) can be drawn within process. The observed data-matrix M from regarded as matrix product two unknown matrices W and H , where first representing contribution latter represents spectra. One approach obtaining non-negative factorization. We propose novel approach, does not need commonly used separability assumption. performance this shown on real world example.