Previous: kstest
Up: References
Next: mpl
Previous Page: kstest
Next Page: mpl

lsf

htmlonly

• name:
  lsf
• synopsis:

  lsf inputfile [-swap_xy] -n_terms= integer [-symmetry={none | odd | even}] | -orders= integer[, integer...] [-revise_orders[= threshold]] [-chebyshev[=convert]] [-fit_output= fit_file[ ,nterms[ ,low,high,npts]]] [-difference_output= difference_file[ ,nterms]] [-sample_interval= integer] [-range= xmin,xmax] [-x_offset= value] [-x_factor= value] [-zero_constant_term] [-normalize[= term_number]] [-modify_sigmas] [-generate_sigmas] [{-sigmas= value,{absolute | fractional} | -ignore_sigmas}] [-verbose]

• description:
  lsf does error-weighted least-squares fits to any order using either ordinary polynomials or Chebyshev T polynomials. Options for fitting only even or odd polynomials, for fitting only specified orders, and for automatic elimination of orders that (though requested) are not genuinely present. Also provides difference data sets and data sets from evaluation of fitted polynomials.
• files:
  inputfile contains the data set that will be fit.
• options:

  swap_xy

  
  Requests that the roles of the abscissa and ordinate in inputfile be exchanged.

  n_terms, symmetry

  
  These options are used together to specify the number of terms to fit and the symmetry of the terms to use.

  orders

  
  Specifies a list of orders of the fitting function to be used. For polynomial fitting, order 0 is a constant term, 1 is a linear term, and so on.

  revise_orders

  
  Requests that the list of orders used in the fit be revised to eliminate orders that are not genuinely present. This is done by asserting that the fractional uncertainty in all ``genuine'' coefficients is less than the specified threshold. Fitting is iterated until all orders are ``genuine''.

  chebyshev

  
  Requests fitting of Chebyshev T polynomials instead of ordinary polynomials. If the convert keyword is given, the coefficients are expressed as ordinary polynomial coefficients.

  fit_output

  
  Requests that the first nterms terms of the fit be evaluated at npts equispaced points from low to high, with the result being placed in fit_file. The fit is summarized in a comment header in this file. The default for nterms is the number of terms in the fit. The default for low and high are the minimum and maximum abscissa values in the data. The default for npts is the the number of data points.

  difference_output

  
  Requests that the difference between the first nterms terms of the fit and the data values be evaluated and placed in difference_file. The fit is summarized in a comment header in this file. The default for nterms is the number of terms in the fit.

  sample_interval

  
  Specifies sampling of the data in inputfile so that only every point is used.

  range

  
  Specifies the range of abscissae in inputfile for which fitting will be done.

  x_offset

  
  Specifies the abscissa value about which fitting will be done. The default is 0.

  x_factor

  
  Specifies a factor by which to multiply the difference between abscissae values and the center of fitting (normally at 0, but changed by x_offset). The default is 1.

  zero_constant_term

  
  Requests that lsf attempt to eliminate the constant term in a polyomial fit by adjusting the point about which fitting is done.

  normalize

  
  Requests that the fit coefficients be normalized so that the specified term (by default the first term) is unity.

  modify_sigmas

  
  Requests that uncertainty values for the abscissae be used to alter the uncertainty values for the ordinates in order to include the former in the fitting process.

  generate_sigmas

  
  Requests that uncertainty values for the ordinates be generated by doing an initial unweighted fit and using the RMS deviation from the fit as the uncertainty for all data points.

  sigmas

  
  Specifies uncertainties for the ordinates of all points as an absolute number (the default) or as a fraction of each ordinate value.

  ignore_sigmas

  
  Requests unweighted fitting.

  verbose

  
  Requests informational output during processing.

• see also:

  
  gfit, expfit, fiteval.