fit.spectrum: Fit a galaxy spectrum with a stellar template and a Gaussian...
In oczoske/slacR: R-package to measure v and sigma from VIMOS IFU cubes

Description Usage Arguments Details Value Author(s) See Also

This function determines velocity and velocity dispersion of a galaxy by optimizing over a merit.function given by convolving a stellar template spectrum and comparing to the galaxy spectrum in the least squares sense. The line profile can be parameterized as Gaussian or a Gauss-Hermite series of order 4.

fit.spectrum(galaxy, template, p0 = c(0, 230, 0, 0),
             LOSVD=c("gauss", "gh", "gauss-hermite"),
             noise = NULL, lambda = NULL, errors = 0, mask = NULL,
             SN.lim = 0, niter = 3, nsig = 3, verbose = FALSE,
             plot = FALSE, plot.errors = FALSE, 
             simple = FALSE, o.mult=4, o.add=6, ...)

`galaxy`	an object of class `spectrum` containing the galaxy spectrum that is to be fitted
`template`	an object of class `spectrum` containing the stellar template spectrum
`p0`	four-component `vector` containing the starting values for the parameters that are to be determined from the template fit: velocity `v` and velocity dispersion `sigma`, both in km/s. The Gauss-Hermite coefficients `h3` and `h4` are ignored if `LOSVD=="gauss"`.
`LOSVD`	Shape of the LOSVD to be fit to the spectrum. Defaults to Gaussian but can also be set to a Gauss-Hermite series, to order 4.
`noise`	an object of class `spectrum` or a vector along `galaxy$lambda` containing an estimate of the noise for each pixel; or, a single number giving the average noise value. If missing, the noise will be estimated from the wavelength region `lambda`. If noise is a `spectrum`, then the wavelength component should be identical to `galaxy$lambda`
`lambda`	a two-component `vector` giving the limits of a wavelength region within which level, noise and signal-to-noise of the galaxy spectrum will be determined
`errors`	Number of Monte Carlo simulations that will be performed to give an estimate of the error distribution; defaults to zero
`mask`	Normally, the mask used in the fit is included in the spectrum `galaxy`. It is possible to specify an additional `mask`, which is combined with `galaxy$mask` by logical AND. Any format that is understood by `convert.mask` is allowed: a data.frame with two columns `x1` and `x2` giving the left and right limits of wavelength regions that are to be masked; or, the name of a file containing such a data.frame; or, a logical vector along `galaxy$lambda`, with `TRUE` for pixels that shall be included in the fit
`SN.lim`	Only galaxy spectra which have signal-to-noise larger than this value will be analysed; defaults to zero, i.e. all spectra will be analysed.
`niter`	number of iterations in the sigma-clipping
`nsig`	data points with residuals larger than `nsig` times the standard deviation of the fit will be masked
`verbose`	`logical`. If `TRUE`, informative messages will be printed
`plot`	`logical`. If `TRUE`, the galaxy spectrum will be plotted along with the best fitting model and residuals.
`plot.errors`	`logical`. If `TRUE`, then a contour plot of the `merit.function` in the `v`–`sigma` plane will be plotted and the estimates from the error runs will be marked
`simple`	`logical`. If `TRUE`, a list of only scalar vectors will be returned (this is required when the function is called by `fit.cube`)
`o.mult`	order of polynomial multiplying the convolved template spectrum
`o.add`	order of polynomial added to the convolved template spectrum
`...`	Further parameters to be passed to `merit.function`

Plotting the errors does not work yet, rewrite of gof.contours is required.

An object of class kinfit. This is the original spectrum with added parameters describing the fit and its results. Of particular importance are the first and second velocity moments, v1 and v2, computed by integrating over the best-fit LOSVD.

If errors is larger than zero, the list includes summaries of the error distribution:

`dv1.low, dv1.high, dv2.low, dv2.high`	68% confidence levels for the first and second velocity moments
`dv.low, dv.high`	68% confidence limits for velocity
`dsig.low, dsig.high`	68% confidence limits for velocity dispersion
`v1.err, v2.err, v.err, sig.err, gof.err, h3.err, h4.err`	velocities, velocity dispersions and corresponding `merit.function` values determined from each Monte Carlo simulation (only if `simple=FALSE`)