Studies on effect of condition number on selectivity of multivariate molecular spectral analysis

This paper reports on the multivariate molecular spectral analysis , including equations suitable for the linear and inconsistent equations sets. The experimental results of the simultaneous spectrophotometric determination indicate that condition number rapidly increases with increase of component number, so that the mean relative standard deviation (MRSD) of the computation results obtained by modified pure spectrum ordinary least square method (OLS-M) steeply increases; but slow increase of MRSD of the computation results obtained by partial least squares method (PLS) shows its stronger ability resisting random error