----------------------------------------------------------------------------------
@MSGID:
<aa2d38fc-1df2-4703-b1b3-67adf45e2793n@googlegroups.com> d4f07709
@REPLYADDR Farzad Tatar
<tatarfarzad1992@gmail.com>
@REPLYTO 2:5075/128 Farzad Tatar
@CHRS: CP866 2
@RFC: 1 0
@RFC-Message-ID:
<aa2d38fc-1df2-4703-b1b3-67adf45e2793n@googlegroups.com>
@TZUTC: -0700
@PID: G2/1.0
@TID: FIDOGATE-5.12-ge4e8b94
Dear All,
I am using Dgetri to invert a square matrix that I have. The
matrix consists of the Radial Basis values of a grid that I have, i.e.,
RBF(Nnode,Nnode).
I have used two different compilers to invert the matrix. The first
one is in the environment of Microsoft powerstaion 4 and the other is
using inter fortran. The problem is that when I create the RBF matrix,
all the components in two different codes are the same but after the
inversion, they are not, though the values are close.
calculated after the inverting with intel:
89588.4169487697 -400462.914663885 993255.716366269
-1815020.42991486 2737745.90690058 -3621529.66069135
4359302.34907216 -4892909.07940900 5208790.59570593
-5324438.63545066
the old calculation, (I have trust in this results because it is
verified) this one takes a much longer time for inversion, though.
92178.015799056420000 -414516.040633608200000 1035342.675828539000000
-1907313.242167841000000 2903705.975803736000000 -3881289.913598144000000
4726376.390789632000000 -5372746.814054917000000 5798863.224470832000000
-6015613.975911859000000 6052730.233410349000000 -5947828.868681677000000
I was thinking of the precision problem, but it cannot be the case
because the calculation of the RBF function, which includes calculating the
distance of each point in the grid with respect to the other points, is
correct for both codes. They are equal to each other. Now, I think maybe
the Intel compiler is somehow reducing the accuracy in the calculation
in the inversion step. Any idea is highly welcome. Thank you in
advance.
Cheers,
Farzad
Thank you for
--- G2/1.0
* Origin: usenet.network (2:5075/128)
SEEN-BY: 5001/100 5005/49 5015/255 5019/40 5020/715
848 1042 4441 12000
SEEN-BY: 5030/49 1081 5058/104 5075/128
@PATH: 5075/128 5020/1042 4441