In intdgesv3.f : there is something I don't understand

Vincent Lejeune vincent.lejeune at institutoptique.fr
Fri Jun 5 23:27:07 CEST 2009


Hi,

I'm trying to reproduce the behaviour of the intdgesv3 function (that is, the 
"backslash" function between 2 double matrix).

After several check of input argument, if the, let's say A matrix in the A\B 
expression, is a square matrix, the function copy the A matrix in another 
matrix AF, call DGETRF that compute the LU decomposition of the AL matrix 
inplace, and then call DGETRS, which resolve the A*X=B system, with the help 
of the previous LU decomposition. The matter is, that DGETRS is called on the 
lB pointer : according to the documentation of the DGETRS function, the lB 
matrix should be modified, that means that the B matrix is not the same after 
the computation.

A quick run into Scilab prooves me that this assertion is wrong, B stay 
unchanged before and after using A\B.

Which step miss I ?

Thanks, Vincent.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.scilab.org/pipermail/dev/attachments/20090605/4af887a1/attachment.htm>


More information about the dev mailing list