[SciPy-user] linear (polynomial) fit with error bars

[Robert Kern]

On Thu, Apr 10, 2008 at 7:09 AM, massimo sandal <massimo.sandal at unibo.it> wrote:

>  massimo at calliope:~/Python/linfit$ python linfit.py
>  Traceback (most recent call last):
>    File "linfit.py", line 31, in <module>
>      w, success = optimize.leastsq(errfunc, [0,0], args=(xval,yval),
>  diag=weigths)
>    File "/usr/lib/python2.5/site-packages/scipy/optimize/minpack.py",
>  line 262, in leastsq
>      m = check_func(func,x0,args,n)[0]
>    File "/usr/lib/python2.5/site-packages/scipy/optimize/minpack.py",
>  line 12, in check_func
>      res = atleast_1d(apply(thefunc,args))
>    File "linfit.py", line 4, in errfunc
>      return Y-(a[0]*X+a[1])
>  ValueError: shape mismatch: objects cannot be broadcast to a single shape
>
>  which baffles me. What should I look for to understand what I am doing
>  wrong?

Print out the various individual parts of that expression to make sure
they are compatible (or use a debugger to do the same interactively).
E.g.

  print Y
  print X
  print a

In this case, the problem is that you are passing in X and Y as lists
instead of arrays. Since a[0] is a numpy scalar type that inherits
from the builtin int type, a[0]*X uses the list type's multiplication
which leaves you with an empty list.

Instead of constructing xval and yval as lists, use numpy.empty().

  xval=numpy.empty([numofdatapoints])
  yval=numpy.empty([numofdatapoints])
  err=numpy.empty([numofdatapoints])

Also, you don't need to "declare" the variables "w" and "success".

-- 
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