[SciPy-user] linalg.scipy use of "GMRESREVCOM.f.scr" for complex-type systems

[idesbald van den bosch]

Hi there,

I would like to reuse directly in C++ code the "GMRESREVCOM.f.scr" fortran
routine that is used in scipy.

The problem is that I need it for complex matrices and vectors (A, b, x),
and "GMRESREVCOM.f.scr" has been written for double-precision floats.

I know that scipy can solve complex-type linear systems using "
GMRESREVCOM.f.src", but I don't understand how it does it.

Apparently, by looking at the scipy "iterative.py" source code, scipy has a
dictionary of linalg methods (ended by "revcom") that work for "floats" and
"complex":

dir(_iterative) = ['__doc__', '__file__', '__name__', '__version__',
'as_column_major_storage', 'cbicgrevcom', 'cbicgstabrevcom', 'ccgrevcom',
'ccgsrevcom', 'cgmresrevcom', 'cqmrrevcom', 'cstoptest2', 'dbicgrevcom',
'dbicgstabrevcom', 'dcgrevcom', 'dcgsrevcom', 'dgmresrevcom', 'dqmrrevcom',
'dstoptest2', 'has_column_major_storage', 'sbicgrevcom', 'sbicgstabrevcom',
'scgrevcom', 'scgsrevcom', 'sgmresrevcom', 'sqmrrevcom', 'sstoptest2',
'zbicgrevcom', 'zbicgstabrevcom', 'zcgrevcom', 'zcgsrevcom', 'zgmresrevcom',
'zqmrrevcom', 'zstoptest2']


My question is: how is 'zgmresrevcom' built, since the fortran source code "
GMRESREVCOM.f.src" has been written for double-precision floats???

Thanks for help anyone!

Ides
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20060516/d644f98b/attachment.html>