For example,Бобцов

STUDY OF ERRORS OF SOME METHODS FOR SEPARATING OVERLAPPED SPECTRAL LINES UNDER NOISE EFFECT

Annotation

Subject of Research. We consider one of the actual problems of spectroscopy, that is, the separation of close spectral lines. Method. The problem is solved by a mathematical (computer) way, namely, by minimizing a functional of discrepancy between the meas­ured and calculated spectra. In this case, the lines (components) are modeled by Gaussians and the problem is reduced to localization of their parameters. Main Results. To minimize the functional we propose the coordinate descent method modification with the use of decremental constraintstechnique. To smooth and differentiate noisy experimental spectral data, we suggest using splines. The software on MatLab is developed and a number of spectra are processed. Practical Relevance. The developed technique can be used to restore the fine structure of spectra and, thereby, to increase the resolving power of spectrometers.

Keywords

Articles in current issue