Optical tolerancing and principal component analysis

This work explores the use of linear principal component analysis (PCA) during an optical design's tolerancing analysis. Chapman et al. [Proc. SPIE3331, 102 (1998)PSISDG0277-786X] have shown the usefulness of the singular value decomposition in realizing an alignment algorithm for a system. This paper explores some insights that can be gained from performing PCA on the Monte Carlo data set obtained during the tolerancing step and comparing it with the singular components of the Jacobian (sensitivity matrix) of the system.