Hello, I am having some trouble with the routine LA_SVD I have a covariance matrix that is generated as covar=xvar#transpose(yvar) covar=covar/mtot where xvar=fltarr(9450,2044); yvar=fltarr(12221,2044) and mtot is a constant Then I use LA_SVD,covar,W,U,V,/DOUBLE,STATUS=stat My question is: How do I get the expanded coefficients associated with each variable (xvar,yvar)? They should be the time coeffcients of the 2044 modes of variability between both my variables. How is LA_SVD different from the former nr_svd? Thank you, Rod