Analytical Sciences


Abstract − Analytical Sciences, 19(7), 1037 (2003).

An Improved Trilinear DecompositionAlgorithm Based on a Lagrange Operator
Jian-Zhong LU, Hai-Long WU, Jian-Hui JIANG, Ning LONG, Cui-Yun MO, and Ru-Qin YU
State Key Laboratory of Chemo/Biosensingand Chemometrics, College of Chemistry and Chemical Engineering, HunanUniversity, Changsha 410082, P. R. China
An improved trilinear decompositionalgorithm based on a Lagrange operator (LO) is developed in this paper,which introduces a Lagrange operator and penalty terms in the loss functionto improve the performance of the algorithm. Compared to the traditionalparallel factor (PARAFAC) algorithm, the algorithm not only may convergemuch faster, but also overcome the sensibility to estimate the number ofcomponents. A set of simulated and measured excitation/emissionfluorescence data were treated by both the proposed and traditional PARAFACalgorithm to compare their efficiencies. The analytical results obtainedwith real chemical system containing aspirin and its metabolic productsshow that the trilinear decomposition methodology is a promising tool toobtain spectral and composition information from mixtures without chemicalseparation.