Dgecov Dgecovs Dgecovy Dgels Dgelsd Dgelss Dgelsy Dgetsls

Dgelsd() Sub Dgelsd ( M As  Long, N As  Long, A() As  Double, B() As  Double, S() As  Double, RCond As  Double, Rank As  Long, Info As  Long, Optional Nrhs As  Long = 1  ) Solution to overdetermined or underdetermined linear equations Ax = b using the singular value decomposition (SVD) (Divide and conquer method) Purpose This routine computes the minimum norm solution to a real linear least squares problem: minimize 2-norm(| b - A*x |) using the singular value decomposition (SVD) of A. A is an M x N matrix which may be rank-deficient. Several right hand side vectors b and solution vectors x can be handled in a single call; they are stored as the columns of the M x Nrhs right hand side matrix B and the N x Nrhs solution matrix X. The problem is solved in three steps: (1) Reduce the coefficient matrix A to bidiagonal form with Householder transformations, reducing the original problem into a "bidiagonal least squares problem" (BLS). (2) Solve the BLS using a divide and conquer approach. (3) Apply back all the Householder transformations to solve the original least squares problem. The effective rank of A is determined by treating as zero those singular values which are less than RCond times the largest singular value. Parameters [in] M Number of rows of the matrix A. (M >= 0) (If M = 0, returns with Rank = 0) [in] N Number of columns of the matrix A. (N >= 0) (If N = 0, returns with Rank = 0) [in,out] A() Array A(LA1 - 1, LA2 - 1) (LA1 >= M, LA2 >= N) [in] M x N matrix A. [out] A() has been destroyed. [in,out] B() Array B(LB1 - 1, LB2 - 1) (LB1 >= max(M, N), LB2 >= Nrhs) (2D array) or B(LB - 1) (LB >= max(M, N), Nrhs = 1) (1D array) [in] M x Nrhs right hand side matrix B. [out] B() is overwritten by the N x Nrhs solution matrix X. If M >= N and Rank = N, the residual sum of squares for the solution in the i-th column is given by the sum of squares of elements N to M-1 in that column. [out] S() Array S(LS - 1) (LS >= min(M, N)) The singular values of A in decreasing order. The condition number of A in the 2-norm = S(0)/S(min(M, N)-1). [in] RCond RCond is used to determine the effective rank of A. Singular values S(i) <= RCond*S(0) are treated as zero. If RCond < 0, machine precision is used instead. [out] Rank The effective rank of A, i.e., the number of singular values which are greater than RCond*S(0). [out] Info = 0: Successful exit = -1: The argument M had an illegal value (M < 0) = -2: The argument N had an illegal value (N < 0) = -3: The argument A() is invalid. = -4: The argument B() is invalid. = -5: The argument S() is invalid. = -9: The argument Nrhs had an illegal value. (Nrhs < 0) = i > 0: The algorithm for computing the SVD failed to converge; i off-diagonal elements of an intermediate bidiagonal form did not converge to zero. [in] Nrhs (Optional) Number of right hand sides, i.e., number of columns of the matrices B and X. (Nrhs >= 0) (If Nrhs = 0, returns with Rank = 0) (default = 1) Reference LAPACK Example Program Compute the least squares solution of the overdetermined linear equations Ax = b and its variance, where ( -1.06 0.48 -0.04 ) A = ( -1.19 0.73 -0.24 ) ( 1.97 -0.89 0.56 ) ( 0.68 -0.53 0.08 ) ( 0.3884 ) B = ( 0.1120 ) ( -0.3644 ) ( -0.0002 ) Sub Ex_Dgelsd() Const M = 4, N = 3 Dim A(M - 1, N - 1) As Double, B(M - 1) As Double, Ci(N - 1) As Double Dim Sigma(N - 1) As Double, RCond As Double, Rank As Long, Info As Long Dim I As Long A(0, 0) = -1.06: A(0, 1) = 0.48: A(0, 2) = -0.04 A(1, 0) = -1.19: A(1, 1) = 0.73: A(1, 2) = -0.24 A(2, 0) = 1.97: A(2, 1) = -0.89: A(2, 2) = 0.56 A(3, 0) = 0.68: A(3, 1) = -0.53: A(3, 2) = 0.08 B(0) = 0.3884: B(1) = 0.112: B(2) = -0.3644: B(3) = -0.0002 RCond = 0.0001 Call Dgelsd(M, N, A(), B(), Sigma(), RCond, Rank, Info) If Info <> 0 Then Debug.Print "Error in Dgelss: Info =", Info Exit Sub End If Debug.Print "X =", B(0), B(1), B(2) Debug.Print "Rank =", Rank, "Info =", Info End Sub Example Results X = -0.82 -0.94 0.740000000000001 Rank = 3 Info = 0