[SciPy-user] optimize.fmin_bfgs only returning a scalar for xopt?

David Warde-Farley dwf@cs.toronto....
Thu Apr 3 14:21:27 CDT 2008


Hi folks,

I just checked out a fresh copy of scipy this morning and I'm having  
an odd problem with optimize.fmin_bfgs (and actually fmin_cg too).

Namely, the function I give it (and the initial guess) use 1d-array  
arguments of length > 1, but I only get a scalar back (specifically a  
numpy.float64) in the first tuple position, which, if I'm reading the  
docs right, should contain the value of x that minimizes the function  
you provided.

This happens whether or not I supply a gradient function.   
Furthermore, my objective function actually raises an exception if  
it's called with a scalar arg instead of an array of the right length.

Is this a bug? Can anyone confirm this behaviour in r4076? Or am I  
doing it wrong?

Thanks,

David



More information about the SciPy-user mailing list