Scilab Website | Contribute with GitLab | Mailing list archives | ATOMS toolboxes


res_with_prec

computes the residual r = Ax-b with precision

Syntax

[r,norm2_r] = res_with_prec(A, x, b)

Arguments

A

real or complex sparse matrix (m x n)

x

column vector (n x 1) or matrix (n x p)

b

column vector (m x 1) or matrix (m x p)

r

column vector (m x 1) or matrix (m x p)

norm2_r

scalar or vector (1 x p) when b is a m x p matrix

Description

This function computes the residual of a linear system r = Ax - b (together with its 2-norm) with the additional precision provided on "Intel like" FPU (80 bits in place of 64) if the compiler translate "long double" to use it. Else one must get the same than using A*x - b at the scilab level. In both cases using [r, nr] = res_with_prec(A,x,b) is faster than r = A*x - b (and faster than r = A*x - b; nr = norm(r)).

When p > 1, norm2_r(i) is the 2-norm of the vector r(:,i).

Examples

[A] = ReadHBSparse(SCI+"/modules/umfpack/demos/bcsstk24.rsa");
C_ptr = taucs_chfact(A);
b = rand(size(A,1),1);
x0 = taucs_chsolve(C_ptr, b);
norm(A*x0 - b)
norm(res_with_prec(A, x0, b))

See also

  • taucs_chsolve — solves a linear s.p.d. system A*X = B from Cholesky factors of the sparse A
Report an issue
<< ReadHBSparse UMFPACK Interface taucs_chdel >>

Copyright (c) 2022-2023 (Dassault Systèmes)
Copyright (c) 2017-2022 (ESI Group)
Copyright (c) 2011-2017 (Scilab Enterprises)
Copyright (c) 1989-2012 (INRIA)
Copyright (c) 1989-2007 (ENPC)
with contributors
Last updated:
Mon Nov 07 15:05:59 CET 2022