LAPACK/ScaLAPACK Development

The test script is below:

Code: Select all

 integer N, NRHS, LDA, LDB, INFO
    parameter(N=4, NRHS=1, LDA=4, LDB=4)
    double precision A(4,4), B(4,1)
    integer IPIV(4)
    A(1,1) = 1.; A(1,2) = 10.; A(1,3)= 1.;  A(1,4) = 1.
    A(2,1) = 2.; A(2,2) = -1.; A(2,3)= 50.; A(2,4) = -11.
    A(3,1) = 4.; A(3,2) = 8.;  A(3,3)= -7.; A(3,4) = -19.
    A(4,1) = 4.; A(4,2) = 8.;  A(4,3)= -7.; A(4,4) = 23
   
    B(1,1) = 13.; B(2,1) = 40.; B(3,1) = -14.; B(4,1) = 28.
    call dgetrs('N', N, NRHS, A, LDA, IPIV, B, LDB, INFO)
    write(*,*)B
    end

and it compiles well with:
f77 test.f -o test -llapack
and it actually generate the execute file "test", but when I execute this file, there is segment fault. I doubt that whether the algorithm in this subroutine has its own flaw. Anyone can explain this kind of phenomenon? Many thanks.

IPIV is an input argument to DGETRS and so needs to be defined before the call to DGETRS. I am sure that you realise that DGETRS is normally preceded by a call to DGETRF, which factorizes A and returns the pivot indices in IPIV.

Best wishes,

Sven Hammarling.

I got it, thanks very much.