Hey, Gary,
Without seeing your code it's hard to know what is right/wrong about
what you're doing. I haven't gotten facile with the Complex package
yet, and there are some pitfalls there, so I tend to work on the real
number line. That changes your two equations in two (complex)
unknowns to four equations in four unknowns. Fortunately it doesn't
take long to invert a 4x4 matrix.
This may be your problem -- the matrix inversion routines (that I
know of)) all work with real numbers. Complex numbers are
implemented as an additional dim of size 2 that runs across {real|
imaginary} axis. The matrix inversion routines are probably
threading over that dimension, giving you separate inverses of the
real 2x2 matrix and the imaginary 2x2 matrix. (That threading
ambiguity is the main pitfall in PDL::Complex number handling -
routines that are complex-number aware work great, but most such
routines are not!).
Another matrix pitfall to watch for: PDL matrices are indexed in
(column, row) order, rather than (row, column); this means that they
render correctly under print(), but that the indices hook up the
opposite way you might expect if you're used to linear algebra with
index notation. Thus, if you "print $a" and "print $b" then "print
$a x $b" should give you the same answer as if you just multiply the
two matrices by hand on paper, in that order. (Contrast, say, IDL,
which made the opposite choice -- the matrices are indexed in (row,
column) order but render incorrectly when printed). That particular
wart is due to conflicting conventions: just about everyone likes to
interpret coordinate n-tuples as happening in (x,y,z,...) order, but
the mathematical (row, column) order for matrices places the first
two in (y,x) order.
Best,
Craig
On Mar 31, 2007, at 5:59 PM, [EMAIL PROTECTED] wrote:
I'm trying to use PDL to trying to solve two simultaneous equations
in two unknowns (in which the equations and unknowns contain
complex numbers). The problem definition is fully explained at
http://www.perlmonks.org/?node=605900
When I attempt to solve for the unknowns by using the PDL inverse
matrix, I get some very bizarre results.
PDL offers promise relative to future work, but I'm on the
beginning of the learning curve. I would be grateful for any
help / explanation what I'm doing wrong.
Thanks, Gary
See what's free at AOL.com.
_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
