[SciPy-user] curve_fit step-size and optimal parameters
Wed Jun 10 02:58:36 CDT 2009
If you try to fit the frequency with the least-squares distance the
problem is not only nonlinearity
but rather the fact that the objective functions has many local minimizers.
At least that's what I have observed in a toy example once.
Has anyone experience what to do in that case? (Maybe use L1 norm instead?)
On Mon, Jun 8, 2009 at 10:19 PM, Robert Kern<firstname.lastname@example.org> wrote:
> 2009/6/8 Stéfan van der Walt <email@example.com>:
>> 2009/6/8 Robert Kern <firstname.lastname@example.org>:
>>> On Mon, Jun 8, 2009 at 14:59, ElMickerino<email@example.com> wrote:
>>>> My question is, how can I get curve_fit to use a very small step-size for
>>>> the phase, or put in strict limits, and to therefore get a robust fit. I
>>>> don't want to tune the phase by hand for each of my 60+ datasets.
>>> You really can't. I recommend the A*sin(w*t)+B*cos(w*t)
>>> parameterization rather than the A*sin(w*t+phi) one.
>> Could you expand? I can't immediately see why the second
>> parametrisation is bad.
> The cyclic nature of phi. It complicates things precisely as the OP describes.
>> Can't a person do this fit using non-linear
>> least-squares? Ah, that's probably why you use the other
>> parametrisation, so that you don't have to use non-linear least
> If you aren't also fitting the frequency, then yes. If you are fitting
> for the frequency, too, the problem is still non-linear.
> Robert Kern
> "I have come to believe that the whole world is an enigma, a harmless
> enigma that is made terrible by our own mad attempt to interpret it as
> though it had an underlying truth."
> -- Umberto Eco
> SciPy-user mailing list
More information about the SciPy-user