<< >> Up Title Contents

fit


fit <ik>

This command enters the fit sublevel and data set <ik> is taken as target data set. For an easy start, select a fit function with 'func' and start the fit using the 'run' command. The result can be displayed with 'plot' and saved with 'save'. For more detailed information see the list of commands below:

commands

Valid commands in this sublevel are:

@      : Executes a macro (see general help)
=      : Algebra (see general help)
cycle  : Sets maximum number of fit cycles
echo   : Echos a string on the screen (see general help)
eval   : Evaluates an expression  (see general help)
exit   : Exits fit sublevel
func   : Sets fit function
help   : Gives on-line help for 'fit' sublevel (see general help)
macro  : Writes current fit parameters to a macro file
output : Toggles fit screen output on/off
para   : Sets/modifies fit parameters
plot   : Plot results
run    : Start the fit
save   : Save fit results
show   : Show settings and fit results
system : Executes operating system command (see general help)
urf    : Sets urf value
wait   : Waits for user input (see general help)
wic    : Sets weighting scheme

cycle

cycle <n>

This command sets the maximum number of cycles to <n>. If a minimum is found in less cycles, the fit will stop before <n> cycles are finished.

func

func { "poly" | "gaus" | "lore" } [,<par>]

This command allows to set the fit function. After this command is called, KUPLOT will calculate starting values and destroy possible previous fit results, which should be saved ('save') before this command is called. The starting values for some functions use the 'smax' function. Make sure that the last 'smax' parameter <ifen> will find the correct maxima. The following table shows the currently allowed functions:

function   2d   3d   what            <par>
-----------------------------------------------------------
 "poly"    yes  no   polynom         order if the polynom
 "gaus"    yes  yes  Gaussian        number of Gaussians
 "lore"    yes  no   Lorenzian       number of Lorenzians

help

help <command>

Displays this help text, for help for a certain command type help command. See help entry on main level for more details.

macro

macro <filename>

This command saves the current parameters in a KUPLOT macro file. This allows to set the parameter values in a later session using @<filename> from the FIT sublevel.

output

output { "on" | "off" }

This command allows to specify if the progress of the fit should be displayed on the screen ("on") or not ("off").

para

para
para <ip>,<value>,<pinc>
para { "save" | "load" }

This command allows to modify the fit parameters. If the command is called alone, the current parameters are displayed. A single parameter <ip> can be set to a value <value>. The value <pinc> specifies if the parameter should be refined (<pinc> = 1) of should be kept fixed (<pinc> = 0).

A set of parameters can be internally stored ('para save') and loaded later back in the parameter array ('para load'). Each save command will overwrite previously stored values.

plot

plot

This command displays the current fit result with the current plot settings (like the plot command in the main level).

run

run

This command starts the fit.

save

save

This command saves the current fit results. The output filename is the filename of the input data set with the following extensions:

  .erg  : text file with fit results
  .fit  : calculated data set
  .dif  : difference between observed and calculated data set

show

show [ { "general" | "fit" | "para" } ]

This command shows fit settings and results. If the command is called without parameters, all information is printed on the screen. The command 'show general' will display only general fit settings, 'show fit' the fit results like R values and 'show para' will list the resulting parameters.

urf

urf <u>

This command sets the URF (some German: Unterer Relaxations Faktor) for the fit. This value determines who 'fast' the fit will move to its minimum or how much the parameter values are changed in each cycle depending on the deviations. A small value (e.g. 0.1) might lead to a fast convergence but might also miss the minimum. A larger value (e.g. 100.0) will give a slow convergence which more certain finds the minimum, but might be caught in local minima rather than in the global one.

Understood ? Well just try different values until your fit converges nicely to the global minimum.

wic

wic <w>

This command defines the weighting scheme to be used for the fit. The valid values for <w> are:

  "one"  : w(y) = 1.0              "squa" : w(y) = y**2
  "sqrt" : w(y) = SQRT(y)          "inv"  : w(y) = 1/y
  "log"  : w(y) = LOG(y)           "isq"  : w(y) = 1/SQRT(y)
  "lin"  : w(y) = y                "dat"  : w(y) = sigma y from data set


<< >> Up Title Contents