["BayesicFitting"]

Global	Documentation
Home	Manual
Readme	Glossary
Notes	Design
Examples	Classes
	Trouble
	Style

class AmoebaFitter( MaxLikelihoodFitter )	Source

Fitter using the simulated annealing simplex minimum finding algorithm,

See also: AnnealingAmoeba

Author Do Kester

Examples

# assume x and y are Double1d data arrays.
x = numpy.arange( 100, dtype=float ) / 10
y = 3.5 * SIN( x + 0.4 )                    # make sine
numpy.random.seed( 12345L )                 # Gaussian random number generator
y += numpy.random.randn( 100 ) * 0.2        # add noise
sine = SineModel( )                         # sinusiodal model
lolim = numpy.asarray( [1,-10,-10], dtype=float )
hilim = numpy.asarray( [100,10,10], dtype=float )
sine.setLimits( lolim, hilim )              # set limits on the model parameters
amfit = AmoebaFitter( x, sine )
param = amfit.fit( y, temp=10 )
stdev = amfit.getStandardDeviation( )       # stdevs on the parameters
chisq = amfit.getChiSquared( )
scale = amfit.getScale( )                 # noise scale
yfit  = amfit.getResult( )                # fitted values
yfit  = sine( x )                         # fitted values ( same as previous )
yband = amfit.monteCarloError( )               # 1 sigma confidence region
# for diagnostics ( or just for fun )
amfit = AmoebaFitter( x, sine )
amfit.setTemperature( 10 )                # set a temperature to escape local minima
amfit.setVerbose( 10 )                    # report every 10th iteration
plotter = IterationPlotter( )             # from BayesicFitting
amfit.setPlotter( plotter, 20 )            # make a plot every 20th iteration
param = amfit.fit( y )

Notes

AmoebaFitter is not guaranteed to find the global minimum.
The calculation of the evidence is an Gaussian approximation which is only exact for linear models with a fixed scale.

Author : Do Kester.

AmoebaFitter( xdata, model, **kwargs )

Create a new Amoeba class, providing inputs and model.

Parameters

xdata : array_like
independent input values
model : Model
the model function to be fitted
kwargs : dict
     Possibly includes keywords from
         MaxLikelihoodFitter : errdis, scale, power
         IterativeFitter : maxIter, tolerance, verbose
         BaseFitter : map, keep, fixedScale

fit( data, weights=None, par0=None, keep=None, size=None, seed=4567, temp=0, limits=None, maxiter=1000, tolerance=0.0001, cooling=0.95, steps=10, verbose=0, plot=False, accuracy=None, callback=None )

Return Model fitted to the data array.

When done, it also calculates the hessian matrix and chisq.

Parameters

data : array_like
the data vector to be fitted
weights : array_like
weights pertaining to the data
The weights are relative weights unless scale is set.
accuracy : float or array_like
accuracy of (individual) data
par0 : array_like
initial values of teh parameters of the model
default: from model
keep : dict of {int:float}
     dictionary of indices (int) to be kept at a fixed value (float)
     The values of keep are only valid for this fit
     See also AmoebaFitter( ..., keep=dict )
size : float or array_like
step size of the simplex
seed : int
for random number generator
temp : float
temperature of annealing (0 is no annealing)
limits : None or list of 2 floats or list of 2 array_like
     None : no limits applied
     [lo,hi] : low and high limits for all values
     [la,ha] : low array and high array limits for the values
maxiter : int
max number of iterations
tolerance : float
stops when ( |hi-lo| / (|hi|+|lo|) ) < tolerance
cooling : float
cooling factor when annealing
steps : int
number of cycles in each cooling step.
verbose : int
     0 : silent
     1 : print results to output
     2 : print some info every 100 iterations
     3 : print some info all iterations
plot : bool
plot the results.
callback : callable
     is called each iteration as
     val = callback( val )
     where val is the minimizable array

Methods inherited from MaxLikelihoodFitter

Methods inherited from IterativeFitter

Methods inherited from BaseFitter