A widget for graphing a parameterized function while exploring parameter variations. Also can adjust the parameters automatically to fit data. Function, parameters, independent variable can be specified by the user. This widget may be saved in a session.
The FunctionFitter starts out with entries for graphing a single exponential with two parameters. When a parameter value is changed the graph is replotted.
Enter space separated names of parameters in a string dialog. Default value editors for these parameters will appear in the rightmost box.
Enter an expression involving the independent variable and the parameters (args). Any valid top level hoc expression is acceptable.
Starts fitting the function to the data with respect to the DataWeights
by adjusting the checked argument values. See
Stop the praxis fitter when it finishes its current/next cycle. At this point it contains a computation of the quadratic form of the parameter phase space (printed in the terminal window).
Plot the function on each call to the error function during fitting. Things are slower if this box is checked.
Instead of fitting all the data according to the data weights, use only 5 equally spaced points in each of the two central data regions. Things can be much faster if this box is checked.
The values of the arguments are used in the plot. When a value is changed the function is replotted. When fit data is present, a checkbox is added to the left of each argument button. When the box is checked then the fitter adjusts the value for a best fit. When not checked the parameter is treated as a constant.
Get data from a file. The format is the number of data points followed by pairs of x,y data.
Pops up a panel of parameters with their range limits. When the fitter calls the error function and one of the parameters is outside its range the error function will return a value of 1e6.
Pops up a panel of data weight intervals and weights. The first interval ranges from the beginning of the data to the interval 1 endpoint. From the interval3 endpoint to the end of the data, the weight is 0. The entire interval is given the weight indicated. Intervals can be manipulated directly by the AdjustWeightRegions tool of the Graph menu.