[Numpy-discussion] hold parameters

josef.pktd@gmai... josef.pktd@gmai...
Tue Aug 2 04:07:20 CDT 2011

On Mon, Aug 1, 2011 at 8:43 PM, Bevan Jenkins <bevan07@gmail.com> wrote:
> Hello,
> I have a function that I fitting to a curve via scipy.optimize.leastsq.  The
> function has 4 parameters and this is all working fine.
> For a site, I have a number of curves (n=10 in the example below).  I would
> like to some of the parameters to be the best fit across all curves (best fit
> for a site) while letting the other parameters vary for each curve.  I have
> this working as well.
> The issue I have is like to be able to vary this for a run.  That is do a run
> where parameter1 is best fit for entire site, whith the remaining three
> varying per curve. Then on the next run, have two parameters being held or
> fitted for all curves at one.  Or be able to do a run where all 4 parameters
> are fit for each individual curve.
> Using my e.g. below, if I change the 'fix' dict, so that 'a','b', and 'c' are
> True, with 'd' False, then I will have to change the zip to
> for a,b,c in zip(a,b,c):
>    solve(a,b,c,d)
> I would prefer to find a way to do this via code.  I hope this example makes
> sense.  The code below is all within my objective function that is being
> called by scipy.optimize.leastsq.
> import numpy as np
> def solve(a,b,c,d):
>    print a,b,c,d
>    #return x*a*b*c*d
> fix = {"a":True,"b":True,"c":False,"d":False}
> n=10
> params = np.array([0,1,2,3]*n)
> params = params.reshape(-1,4)
> if fix["a"] is True:
>    a = params[0,0]
> else:
>    a = params[:,0]
> if fix["b"] is True:
>    b = params[0,1]
> else:
>    b = params[:,1]
> if fix["c"] is True:
>    c = params[0,2]
> else:
>    c = params[:,2]
> if fix["d"] is True:
>    d = params[0,3]
> else:
>    d = params[:,3]
> res=[]
> for c,d in zip(c,d):
>    res = solve(a,b,c,d)
>    #res = solve(a,b,c,d)-self.orig
> #return np.hstack(res)**2

I'm not a fan of named individual parameters for function arguments
when the number of arguments varies, *args

What I'm using is a full parameter array with nan's

fixed = np.array([nan, nan, c, d])  #fix c,d

def func(args):
   fixed[np.isnan(fixed)] = args
   a,b,c,d = fixed

to set starting values allstartvals = np.array([a0, b0, c0, d0])

startvals = allstartvals[np.isnan(fixed)

optimize.leastsq(func, startvals, other_args)

or something like this.
I find it easier to keep track of the parameters, if I just have an
array or tuple.

for an alternative, Travis used a different way in the scipy.stats
implementation of partially fixed parameters for distributions fit
with named arguments. (I don't remember the details)


> _______________________________________________
> NumPy-Discussion mailing list
> NumPy-Discussion@scipy.org
> http://mail.scipy.org/mailman/listinfo/numpy-discussion

More information about the NumPy-Discussion mailing list