From a244c49256713d86521a660562dc4ee25f7621e5 Mon Sep 17 00:00:00 2001 From: Wirawan Purwanto Date: Thu, 28 May 2015 17:22:46 -0400 Subject: [PATCH] * Bug fixes and documentation update. --- math/fitting/stochastic.py | 27 ++++++++++++++++++++------- 1 file changed, 20 insertions(+), 7 deletions(-) diff --git a/math/fitting/stochastic.py b/math/fitting/stochastic.py index aa6106a..d10b8c4 100644 --- a/math/fitting/stochastic.py +++ b/math/fitting/stochastic.py @@ -15,12 +15,26 @@ wpylib.math.fitting.stochastic module Tools for stochastic curve fitting. """ +import numpy +import numpy.random + from wpylib.math.fitting import fit_func_base +from wpylib.math.stats.errorbar import errorbar class StochasticFitting(object): """Standard stochastic fit procedure. + Class attributes: + + * `func`: function ansatz to be fitted. + Set via init_func() method. + This `func` needs to be a descendant of the fit_func_base object, + or have an identical API, which are: + + - method `fit` + - method `__call__` (i.e. a callable object) + """ debug = 0 dbg_guess_params = True @@ -42,20 +56,18 @@ class StochasticFitting(object): - the uncertainty of the target points, dy """ - x = fit_func_base.domain_array(x) - if not (len(x[0]) == len(y) == len(dy)): - raise TypeError, "Length of x, y, dy arrays are not identical." - # fix (or, actually, provide an accomodation for) a common "mistake" # for 1-D domain: make it standard by adding the "first" dimension - if len(x.shape) == 1: - x = x.reshape((1, x.shape[0])) + x = fit_func_base.domain_array(x) self.samples_x = x self.samples_y = numpy.array(y) self.samples_dy = numpy.array(dy) self.samples_wt = (self.samples_dy)**(-2) + if not (len(x[0]) == len(y) == len(dy)): + raise TypeError, "Length of x, y, dy arrays are not identical." + def init_rng(self, seed=None, rng_class=numpy.random.RandomState): """Initializes a standard random number generator for use in the fitting routine.""" @@ -94,7 +106,8 @@ class StochasticFitting(object): raise RuntimeError, "Cannot determine the number of fit parameters." def nlfit1(self): - """Performs the non-stochastic, standard nonlinear fit.""" + """Performs the non-stochastic, standard nonlinear fit. + The output is given in `nlf_rec` attribute.""" from numpy.linalg import norm if self.use_dy_weights: