MULTIVARIATE DATA-ANALYSIS FOR CHROMATOGRAPHY

The combination of a chromatograph with a multichannel detector allows to periodically take a complete spectrum of an eluting sample and to produce a data matric with a high content of information. In order to make optimal use of the qualitative and quantitative information inher ent in the collected data, multivariate data analyse's has to be appli ed. The multivariate methods developed are closely related to multicha nnel signal processing that have been intensively studied in other fie lds like radar and sonar data processing. In chromatographic data anal ysis, there are essentially three estimation problems concerning the r ank of the signal matrix, the individual parameters of the residence t ime distribution and the spectrum identifying each chemical component in a sample mixture. These estimation problems can be solved using alg orithms originating from numerical linear algebra and nonlinear optimi zation. This is demonstrated for the rank estimation problem.