Premium
A non‐linear least squares fitting routine for optimizing empirical XRF matrix correction models
Author(s) -
Schreiner W. N.,
Jenkins R.
Publication year - 1979
Publication title -
x‐ray spectrometry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.447
H-Index - 45
eISSN - 1097-4539
pISSN - 0049-8246
DOI - 10.1002/xrs.1300080111
Subject(s) - minification , computer science , algorithm , least squares function approximation , gradient descent , matrix (chemical analysis) , process (computing) , line (geometry) , computer program , mathematical optimization , mathematics , statistics , artificial intelligence , materials science , geometry , estimator , artificial neural network , composite material , operating system
An off‐line non‐linear least squares fitting procedure is described as part of a program package ‘RUNFIT’. This program supports most of the intensity/correction algorithms currently employed in X‐ray fluorescence spectrometry. The program runs under the Data General DOS operating system and requires a minimum hardware configuration of 32K 16 bit words and two floppy disks. The minimization technique is based on the Marquardt scheme which has the main advantage that the step direction taken in the minimization process always lies in the direction of steepest descent. The RUNFIT program includes various statistical tests on the fitted data which enable the user to make accurate judgements as to the quality of the fitted data. Several examples are given of the use of this approach to typical analytical situations.