[SciPy-user] leastsq not converging with tutorial example

[Kael Fischer]

On 10/27/05, Alan G Isaac <aisaac at american.edu> wrote:
> I've lost track of this conversation:
> are you working with the old SciPy?
> If so, I get a fine fit.
> (There is no unique solution.)
> Output below.
> Cheers,
> Alan Isaac
 [clip]

Thanks for the replies - but I realize that this code works for most
people (I'm assuming that's why it is in the tutorial) and that my
observations are anomalous.  What would be most helpful, would be if
someone with a FreeBSD system with the current release of scipy
(custom or port build) would report their results.


I'm using the released 0.3.2 version of scipy.  I've gotten as far as
the minpack fortran code, and it appears that on line 313 of lmder.f
the following condition is true after a few (<= 4) function
evaluations:

        if (gnorm .le. gtol) info = 4

on my systems and with my libraries.

Thanks,
Kael