Application of differential evolution to fitting OSL decay curves

• Published in: Radiation measurements Vol. 41; no. 7; pp. 886 - 891
• Main Authors: Bluszcz, A., Adamiec, G.
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Oxford Elsevier Ltd 01.08.2006; Elsevier
• Subjects: Deconvolution of OSL curve; Differential evolution; Earth sciences; Earth, ocean, space; Exact sciences and technology; Geochronology; Isotope geochemistry. Geochronology; Non-linear regression; Differential evolution; Non-linear regression; Deconvolution of OSL curve; algorithms; deconvolution; efficiency; luminescence; regression; OSL
• ISSN: 1350-4487; 1879-0925
• DOI: 10.1016/j.radmeas.2006.05.016

The differential evolution (DE) algorithm and its application to deconvolution of OSL decay curves into first order components is presented. The algorithm is very fast and robust in finding a global minimum in the problem under consideration. Falling into local minima is a typical problem with deterministic fitting methods, like the Levenberg–Marquardt (LM) algorithm. The algorithm was adapted to form a hybrid evolutionary-linear algorithm, HELA. The evolutionary part of the algorithm was applied only to the decay constants (optical detrapping cross-sections) of the exponential components. For given decay constants, the intensities were calculated using a generalised linear regression method and further improvements were introduced to enhance the efficiency of the algorithm. After finding a solution using the HELA algorithm the parameters obtained were fed into the LM algorithm, providing a nearly automatic procedure for fitting OSL decay curves with a sum of exponential components which in turn may prove very useful in estimating absorbed doses for different components.