dgssvd zgssvd

dgssvd() void dgssvd ( char  jobv, char  jobu, int  m, int  n, const double  val[], const int  ptr[], const int  ind[], int  base, int  format, double  s[], const char *  which, int  nev, double  tol, double  resid[], int  ncv, int  ldv, double  v[], int  ldu, double  u[], int  maxiter, double  workd[], double  workl[], int  lworkl, double  workv[], int  iwork[], int *  info  ) Singular value decomposition (SVD) of a general sparse matrix (driver) Purpose The singular values (s) and corresponding right singular vectors (v) for the the matrix A are computed by solving the symmetric eigenvalue problem: (A^T*A)*v = s^2*v where A is an m by n real general sparse matrix. The left singular vectors are also computed by u = A*v/s if required. The sparse matrix is stored in CSC or CSR format. The eigenvalue problem is solved by the implicitly restarted Lanczos method (IRLM) using arpack routines dsaupd and dseupd. Parameters [in] jobv = 'N': Do not compute right singular vector(s). = 'V': Compute right singular vector(s). [in] jobu = 'N': Do not compute left singular vector(s). = 'U': Compute left singular vector(s). (jobv must also be 'V'.) [in] m Number of rows of the matrix. (m >= 0) (If m = 0, returns without computation) [in] n Number of columns of the matrix. (n >= 0) (If n = 0, returns without computation) [in] val[] Array val[lval] (lval >= nnz) Values of nonzero elements of input matrix (where nnz is the number of nonzero elements). [in] ptr[] Array ptr[lptr] (lptr >= n + 1) Column pointers (if CSC) or row pointers (if CSR) of input matrix. [in] ind[] Array ind[lind] (lind >= nnz) Row indices (if CSC) or column indices (if CSR) of input matrix (where nnz is the number of nonzero elements). [in] base Indexing of ptr[] and ind[]. = 0: Zero-based (C style) indexing: Starting index is 0. = 1: One-based (Fortran style) indexing: Starting index is 1. [in] format Sparse matrix format. = 0: CSR format. = 1: CSC format. [out] s[] Array s[ls] (ls >= nev) Contains the singular values of A. The values are returned in ascending order. [in] which = "LA": Compute the nev largest (algebraic) singular values. = "SA": Compute the nev smallest (algebraic) singular values. = "LM": Compute the nev largest (in magnitude) singular values. = "SM": Compute the nev smallest (in magnitude) singular values. = "BE": Compute nev singular values, half from each end of the spectrum. When nev is odd, compute one more from the high end than from the low end. [in] nev Number of singular values to be computed. (0 < nev < n) [in] tol Stopping criterion: the acceptable relative accuracy of the Ritz value. If tol <= 0, machine precision is assumed. [out] resid[] Array resid[lresid] (lresid >= n) The residual vector. [in] ncv This will indicate how many Lanczos vectors are generated at each iteration. (nev < ncv <= n) (ncv >= 2*nev is recommended) [in] ldv Leading dimension of the array v. (ldv >= n) [out] v[] Array v[ldv * lv] (lv >= nev) nev right singular vectors. Not referenced if jobv = 'N'. [in] ldu Leading dimension of the array u. (ldu >= m) [out] u[] Array u[ldu * lu] (lu >= nev) nev left singular vectors. Not referenced if jobu = 'N'. [in] maxiter Maximum number of Lanczos update iterations allowed. [out] workd[] Array workd[lworkd] (lworkd >= 3*n) Distributed work array used for reverse communication. [out] workl[] Array workl[lworkl] Local work array. [in] lworkl Size of array workl[]. (lworkl >= ncv^2 + 8*ncv) [out] workv[] Array workv[lworkv] (lworkv >= n*ncv) Double work array. [out] iwork[] Array iwork[liwork] (liwork >= max(ncv, 3)) Integer work array. The following values are returned in iwork[0], ..., iwork[2]. iwork[0]: Number of Lanczos update iterations taken. iwork[1]: Number of Ritz values that satisfy the convergence criterion (nconv). iwork[2]: Total number of OP*x operations (numop). [out] info Return code. = 0: Successful exit. < 0: The (-info)-th argument is invalid. = 1: Maximum number of iterations taken. = 3: No shifts could be applied during a cycle of the implicitly restarted Lanczos iteration. A possible remedy is to increase ncv relative to nev (ncv >= 2*nev is recommended). = 11: Initial residual vector is zero. = 12: Failed to build a Lanczos factorization. = 13: Error return from LAPACK eigenvalue calculation. = 14: dsaupd did not find any eigenvalues to sufficient accuracy. = 15, 16: Internal code error.