Calculation of maximum and minimum band boundaries of feasible solutions for species profiles obtained by multivariate curve resolution

A method for the calculation of maximum and minimum band boundaries of feasible solutions corresponding to the species profiles estimated by multivariate curve resolution is presented. The method is based on a non-linear constrained optimization of an objective function defined by the ratio of the signal contribution of a particular species to the whole measured signal. Implementation of constraints such as normalization, closure, nonnegativity, unimodality and local rank/selectivity during the non-linear optimization and their effect on the calculation of the feasible solutions are studied in detail. Calculation of the band boundaries is shown for different simulated and real two-way data examples of increasing complexity. Copyright (C) 2001 John Wiley & Sons, Ltd.