>Has anyone got any complex eof routines for NCL that they are willing to
>make available?
Well, if nobody is going to volunteer code ...
NCAR's Climate Analysis Section has an f90 subroutine
to do this. It uses LAPACK "subroutine cheev" to solve the
complex Hermitian matrix.
It makes the assumption that the number of spatial pts (M)
exceeds the number of temporal pts. This is the case with which
we generally encounter.
Pls note:
(1) ***NO SUPPORT is offered for this****
(2) The f90 subroutine uses complex types. If u r going
to call this from NCL, the following must be done.
It is not that difficult.
(a) u must write a f77 wrapper subroutine so WRAPIT
can automatically parse the argument list.
(b) The user must explicitly allocate arrays
for the returned real/imaginary parts.
(c) The wrapper subroutine must explicitly transfer
the returned complex arrays to separate
real and imaginary arrays. NCL does not
have a complex data type.
(3) Unlike NCL's eof routines, this code does not allow
missing data.
(4) Did I mention, *no support* is offered for this code. :-)
The code will be sent off-line.
Regards
_______________________________________________
ncl-talk mailing list
ncl-talk@ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
This archive was generated by hypermail 2b29 : Tue Oct 05 2004 - 14:16:44 MDT