24 June 2005 08:23:20 AM SVD_BASIS: (C++ version) Compiled on Jun 24 2005 at 08:22:58. Given a PDE for which: C is the number of components of the solution at any single point, P is the number of points where a solution is given, N is the number of solution vectors, L is the number of modes to be extracted. Then we let M = C*P be the abstract spatial dimension. Set up A, the M by N matrix of solution vectors, Get A = U * S * V', the singular value decomposition. The first L columns of U are our modes. Compiled on Jun 24 2005 at 08:22:58. How many basis vectors (L) are to be extracted? L = 4 You specify a consecutive sequence of file names by giving the first "base" file name. If there are no more sequences to enter, just hit RETURN. Enter a new base file name, or RETURN: 1: "data01.txt" According to the first base file, The number of solution components C = 3 The number of solution points P = 4 The "size" of each solution M = (C*P) = 12 You specify a consecutive sequence of file names by giving the first "base" file name. If there are no more sequences to enter, just hit RETURN. Enter a new base file name, or RETURN: RETURN was entered. Presumably, there are no more file sequences. The number of data files N = 5 The data is stored in an M by N matrix A. The "spatial" dimension M is 12 The number of data points N is 5 The data has been read into the matrix A. SINGULAR_VECTORS For an MxN matrix A in general storage, The LINPACK routine DSVDC computes the singular value decomposition: A = U * S * V' The leading singular values: 1 75.7027 2 11.1519 3 7.5139 4 2.64009 SVD_BASIS: Ready to write the left singular vectors to files. Do you want comments in the header of the file? (These begin with the "#" character.) (Y/N) Enter "Y" or "N": Writing first file svd_001.txt Writing last file svd_004.txt SVD_BASIS: Normal end of execution. 24 June 2005 08:23:21 AM