wirawan0/wpylib
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

* wpylib.math.fitting.fit_func_base: changed default fitting method to leastsq.

Browse Source
This commit is contained in:
Wirawan Purwanto
2015-06-04 16:06:21 -04:00
parent b93d40d0fc
commit 70ae2812c3
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
math/fitting/__init__.py
View File

@@ -41,8 +41,8 @@ The following methods are currently supported for scipy.optimize:
See the documentation of `scipy.optimize` for more details. See the documentation of `scipy.optimize` for more details.
The `fmin` algorithm is the slowest although it is fairly foor proof to The `fmin` algorithm is the slowest although it is fairly foor proof to
converge it (it may take many iterations). converge it (it may take very many iterations).
The leastsq` algorithm is the best but it requires parameter guess that is The `leastsq` algorithm is the best, but it requires parameter guess that is
reasonable. reasonable.
I don't have much success with `anneal`--it seems to behave erratically in I don't have much success with `anneal`--it seems to behave erratically in
my limited experience. YMMV. my limited experience. YMMV.
@@ -492,7 +492,7 @@ class fit_func_base(object):
fit_default_opts["lmfit:leastsq"] = dict(xtol=1e-8, epsfcn=1e-6) fit_default_opts["lmfit:leastsq"] = dict(xtol=1e-8, epsfcn=1e-6)
debug = 0 debug = 0
dbg_params = 1 dbg_params = 1
fit_method = 'fmin' fit_method = 'leastsq' # changed 20150529 from fmin. Leastsq is much faster.
fit_opts = fit_default_opts fit_opts = fit_default_opts
#fit_opts = dict(xtol=1e-5, maxfun=100000, maxiter=10000, disp=0) #fit_opts = dict(xtol=1e-5, maxfun=100000, maxiter=10000, disp=0)
def fit(self, x, y, dy=None, fit_opts=None, Funct_hook=None, Guess=None): def fit(self, x, y, dy=None, fit_opts=None, Funct_hook=None, Guess=None):