HELP · SEE_ALSO

## NAME

dispfity -- Compute FOS wavelength dispersion coefficients.

## USAGE

`dispfity table dcout`

## DESCRIPTION

This routine computes the FOS wavelength dispersion coefficients
using the sample positions for the spectral lines found by the `linefind`
task. For each grating mode the dispersion relation is described as
follows:

w = c0 + (c1 * x) + (c2 * x**2) + (c3 * x**3)

where:

w - is the wavelength, x - is the sample position, c0, c1, c2, and c3 - are the fitted coefficients.For the prism modes the dispersion relation is described by the following two equations:

w = c0 + (c1 / xx) + (c2 / xx**2) + (c3 / xx**3) + (c4 / xx**4) xx = x - xzero xzero - is a fitted coefficient.A weighted least squares fit of the wavelengths as a function of the found sample positions in the input table is performed. For the prism mode, the non-linear least squares fit (i.e., the Levenberg-Marquardt Method) is used. Starting "guesses" for the fitted coefficients are taken from an input dispersion coefficient table, specified by the

`dcin`parameter.

Lines where the approximate and found position differ by more
than the value of `maxdev` are not used in the fit. Lines having a peak count
less then the value of `minint` are also not used.

The least squares fit is repeated a maximum of `ntry` times. Residuals
are computed as the difference between the found and the fitted positions.
Lines having a residual greater than `nsig` times the RMS of the
residuals are not used in subsequent fits.

## PARAMETERS

- table = "linefind" [file name]
- Input table containing the found locations; this table is generated
by the
`linefind`task. This table also contains a column`WEIGHT`, which is used to weight the lines in the fit.

- dcout = "dispfit" [file name]
- Name of the output table that will contain the dispersion coefficients.
This table will have the following columns:
DETECTOR char*5 APER_ID char*3 APER_POS char*6 FGWA_ID char*3 POLAR_ID char PASS_DIR integer COEFF_0 double COEFF_1 double COEFF_2 double COEFF_3 double COEFF_4 double XZERO double NLINES integer NGOOD integer RMSFIT double

- (dcin = fos$prismdc") [file name]
- File name of the previous dispersion coefficient table. This parameter
is required only for the
prism modes;
`dcin`provides initial guesses for the non-linear least squares fit. It will be necessary to obtain this file from the HST archive.

- (maxdev=2.0) [real]
- Maximum deviation---in sample or diode units---of the found line
location from the approximate
location. Lines having deviations larger than
`maxdev`are not used in the fit.

- (minint=0.0) [real]
- Minimum peak intensity, or counts in the line, for inclusion in the fit.

- (order=3) [integer]
- Order of the polynomial fit for the grating modes. This parameter is ignored for the prism modes.

- (nsig=2.5) [real]
- Rejection limit. Lines having residual values greater than
`nsig`times the RMS residual are deleted from subseqent fit attempts.

- (ntry=3) [integer]
- Maximum number of times to try the fit; bad data is rejected after each try.

- (niter = 20) [integer]
- Number of iterations in the non-linear least squares fit. This parameter is used only for the prism mode.

## EXAMPLES

1. Perform a fit for the table `stabh40`, writing the output to `dtab`.

fo> dispfity stabh40 dtab

2. Perform the fit again using a 3rd order polynomial. Reject lines having peak counts below 30.

fo> dispfity stabh40 dtab order=3 minint=30.0

## BUGS

## REFERENCES

Howard Bushouse, STSDAS

## HELP

For assistance using this or any other tasks, please contact help@stsci.edu or call the help desk at 410-338-1082.

## SEE ALSO

linefind, waveoffset, fitoffsety