EXTRACTING EXPERIMENTAL INFORMATION FROM LARGE MATRICES .1. A NEW ALGORITHM FOR THE APPLICATION OF MATRIX RANK ANALYSIS

For many, especially complex, systems, modern spectroscopic measuremen ts can be generated as large experimental data sets in matrix form. We report a new algorithm for the application of matrix rank analysis to extract significant experimental information from these large matrixe s. The algorithm may be used to detect and remove erroneous rows and/o r columns from the matrixes and to monitor the most significant experi mental information along the rows and/or columns of the data sets. A n ew method for determining the number of absorbing species and a new co ncept for the treatment of experimental errors are presented. The algo rithm is illustrated on real experimental examples.