[SciPy-user] floating point exception in interpolate.splrep

[Taro Sato]

Travis Oliphant <oliphant.travis <at> ieee.org> writes:

> 
> Taro Sato wrote:
> > with a weight vector, I never get this error.  It seems the
> > combination of x,y, and w that ends up causing the error (especially
> > something in relation to the combination of y and w), but I haven't
> > been able to nail down the cause.
> >
> 
> Hmm..  I could not reproduce your problems.   But, I did get an error on 
> trying to read the data which might be related to what you are 
> experiencing.  I re-wrote your data-reader to use N.fromfile and I am 
> able to process both data sets on my system.  Attached is a new test.py 
> file.
> 


I tried the revised version of the data-reader, but I still get exactly the same
error as before on my machine.  Could this be a plat-form dependent compilation
issue?  I have the same problem with both my desktop and laptop running Debian,
with pretty much the same setting (so these won't help me track down...)

I build scipy/numpy from source from the subversion repository, following
exactly the unofficial instructions written by Steve Baum (without atlas).  I
have gcc version 3.3.5 (Debian 1:3.3.5-13).

I'd appreciate any suggestions for me to track down the cause!

Thanks,
Taro