| Matrix Vector Multiply in IDL [message #33044] |
Tue, 26 November 2002 08:25 |
K Banerjee
Messages: 14
Registered: September 2001
|
Junior Member |
|
|
Folks,
I am trying to port some IDL code to C/C++ and I
am stuck on a matrix multiplication problem. As I
understand things, IDL stores an M by N matrix as
an N by M array. Here are the dimensions of the
relevant arrays:
KX: 148 by 1, a vector in the linear algebra sense.
Fac1: 148 by 3 array, a 3 by 148 matrix.
fac1tr: transpose of Fac1
; Explictily carrying out the matrix-vector multiply:
fac1tr = transpose(Fac1)
tempBetas = fltarr(3)
for i = 0, 2 do begin
for j = 0, 147 do begin
tempBetas(i) = tempBetas(i) + fac1tr(i, j) * KX(i)
endfor
endfor
Here are the values in tempBetas:
tempBetas[ 0] = [ 9.24656e-06]
tempBetas[ 1] = [ -5.54790e-06]
tempBetas[ 2] = [ 745.434]
The corresponding values from my C/C++ code are:
element[0] = [-4.01067e-07]
element[1] = [2.75595e-08]
element[2] = [745.434]
As you can see, the above values are "close" to the values
computed in IDL.
; Now carrying out IDL's matrix vector multiply:
Betas=Fac1##transpose(KX)
The values of Betas end up being:
Betas:[ 0] = [ 4.35200]
Betas:[ 1] = [ -2.40834]
Betas:[ 2] = [ 795.034]
Why don't tempBetas and Betas have the same values? How is using
"##" different from computing the matrix-vector multiply with
2 nested for loops? (I am probably missing something obvious!)
Thanks.
K. Banerjee
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
