[pars,criteria,message,output] = fits(a, model, pars, options, constraints, ...) : fit data set on a model
@iData/fits find best parameters estimates in order to minimize the
fitting criteria using model 'fun', by mean of an optimization method
described with the 'options' structure argument.
Additional constraints may be set by fxing some parameters, or define
more advanced constraints (min, max, steps). The last arguments controls the fitting
options with the optimset mechanism, and the constraints to apply during optimization.
The fit can be applied sequentially and independently onto iData object arrays.
[pars,...] = fits(a, model, pars, options, lb, ub)
uses lower and upper bounds as parameter constraints (double arrays)
[pars,...] = fits(a, model, pars, options, fixed)
indicates which parameters are fixed (non zero elements of array).
[pars,...] = fits(a, model, pars, 'optimizer', ...)
uses a specific optimizer and its default options=feval(optimizer,'defaults')
See below for suggested best optimizers.
[pars,...] = fits(a, model, pars, options, constraints, args...)
send additional arguments to the fit model(pars, axes, args...)
[optimizers,functions] = fits(iData)
returns the list of all available optimizers and fit functions.
fits(iData)
displays the list of all available optimizers and fit functions.
You may create new fit models with the 'ifitmakefunc' tool.
When the iData object contains a Monitor value, the fit is performed on
Signal/Monitor.
When parameters, options, and constraints are entered as a string with
name=value pairs, the string is interpreted as a structure description, so
that options='TolX=1e-4; optimizer=fminpso' is a compact form for
options=struct('TolX','1e-4','optimizer','fminpso').
To set a constraint on a model parameter, define the 'constraint' input argument
or set the constraint directly on the model parameters with:
model.parameter='fix' % to lock its value during a fit process
model.parameter='clear' % to unlock value during a fit process
model.parameter=[min max] % to bound value
model.parameter=[nan nan] % to remove bound constraint
model.parameter='' % to remove all constraints on 'parameter'
model.Constraint='' % to remove all constraints
The default fit options.criteria is 'least_square', but others are available:
least_square (|Signal-Model|/Error).^2 non-robust
least_absolute |Signal-Model|/Error robust
least_median median(|Signal-Model|/Error) robust, scalar
least_max max(|Signal-Model|/Error) non-robust, scalar
least_rfactor (|Signal-Model|/Error).^2/(Signal/Error).^2 non-robust
max_corrcoef 1-corrcoeff(Signal, Model) scalar
max_likelihood
Type <a href="matlab:doc(iData,'Fit')">doc(iData,'Fit')</a> to access the iFit/Fit Documentation.
Type <a href="matlab:doc(iData,'Optimizers')">doc(iData,'Optimizers')</a> to access the Optimizers Documentation.
input: a: object or array (iData)
when given as an empty iData, the list of optimizers and fit models
is shown.
model: model function (char/iFunc/function handle)
the model is converted into an iFunc object, with 'p' as parameters
and 'x,y,...' as axes (1st axis 'x' refers to rows, 'y' to columns)
from a string: 'signal = expression(p, x,y,...);'
'expression(p, x,y,...)'
from function handle: @(p,x,..)expression or @function(p,x,...)
when set to empty, the 'gauss' 1D function is used (and possibly extended to multidimensional).
pars: initial model parameters (double array).
when set to empty the starting parameters are guessed.
when set to 'current', the current model parameter values are used.
Named parameters can be given as a structure or string 'Amplitude=...; Width=...'
options: structure as defined by optimset/optimget (char/struct)
if given as a char, it defines the algorithm to use and its default
options (single optimizer name or string describing a structure).
when set to empty, it sets the default algorithm options (fmin).
options.TolX
The termination tolerance for x. Its default value is 1.e-4.
options.TolFun
The termination tolerance for the function value. The default value is 1.e-4.
This parameter is used by fminsearch, but not fminbnd.
options.MaxIter
Maximum number of iterations allowed.
options.MaxFunEvals
The maximum number of function evaluations allowed.
options.optimizer
Optimization method. Default is 'fminsearch' (char/function handle)
the syntax for calling the optimizer is e.g. optimizer(criteria,pars,options,constraints)
Best optimizers are:
fminpso: Particle Swarm Optimization
fminpowell: Powell with Coggins line search
fminhooke: Hooke-Jeeves direct search
fminralg: Shor R-algorithm
fminsimpsa: Simplex/simulated annealing
fminimfil: Unconstrained Implicit filtering
options.criteria
Minimization criteria. Default is 'least_square' (char/function handle)
the syntax for evaluating the criteria is criteria(Signal, Error, Model)
options.OutputFcn
Function called at each iteration as outfun(pars, optimValues, state)
The 'fminplot' function may be used.
options.Display
Display additional information during fit: 'iter','off','final'. Default is 'iter'.
options.Diagnostics
When set to 'on' or 1, returns the correlation coefficient and Hessian matrix
constraints: fixed parameter array. Use 1 for fixed parameters, 0 otherwise (double array)
OR use empty to not set constraints
OR use a structure with some of the following fields:
constraints.min: minimum parameter values (double array)
constraints.max: maximum parameter values (double array)
constraints.step: maximum parameter step/change allowed.
constraints.fixed: fixed parameter flag. Use 1 for fixed parameters, 0 otherwise (double array)
constraints.eval: expression making use of 'p', 'constraints', and 'options'
and returning modified 'p'
or function handle p=@constraints.eval(p)
OR use a string 'min=...; max=...' to define the structure
output:
pars: best parameter estimates (double array)
criteria: minimal criteria value achieved (double)
message: return message/exitcode from the optimizer (char/integer)
output: additional information about the optimization (structure)
algorithm: Algorithm used (char)
funcCount: Number of function evaluations (double)
iterations: Number of iterations (double)
parsHistory: Parameter set history during optimization (double array)
criteriaHistory: Criteria history during optimization (double array)
modelValue: Last model evaluation (iData)
parsHistoryUncertainty: Uncertainty on the parameters obtained from
the optimization trajectory (double)
ex: a=iData('sv1850.scn'); p=fits(a,'','','fminpowell');
o=fminpowell('defaults'); o.OutputFcn='fminplot';
[p,c,m,o]=fits(a,'gauss',p,o); b=o.modelValue; figure; plot(a,b)
Version: Nov. 26, 2018
See also iData, fminsearch, optimset, optimget, ifitmakefunc, Models, iFunc/fits