NAME:
                 ULY_TGM
 PURPOSE:
                 Define a TGM component
 USAGE:
                 cmp = uly_tgm(MODEL_FILE=model_file, LSF=lsf_file,      
                               TG=t_guess, LG=l_guess, ZG=z_guess,   
                               TL=t_limits, LL=l_limits, ZL=z_limits,
                               NAME=name, FIXPAR=fixpar, WL=lim_weight)

 KEYWORDS:
   MODEL_FILE    Name of the FITS file containing the TGM model, by default 
                 use: /models/elodie32_flux_tgm.fits.

   LSF           Name of the file containing a relative LSF to be injected in 
                 the template. 

   TG            [K] Guess value for the effective temperature (Teff), by
                 default = 7000.0. If an array is passed, the analysis will 
                 be repeated starting from each element to find the optimal 
                 solution.
   
   LG            [cm/s^2] Guess value for the surface gravity (Logg), by 
                 default = 2.0, it can also be given as an array to search 
                 the absolute minimum.

   ZG            [dex] Guess value for the metallicity ([Fe/H]), by 
                 default = 0.0, it can also be given as an array
                 to search the best minimum from the grid of guesses.

   TL            [K], dblarr(2), limits for the Teff,
                 by default = [3100.0,40000.0] see  ULY_TGM_INIT.

   LL            [dex],dblarr(2), limits for the Logg,
                 by default = [-0.25, 5.9] see ULY_TGM_INIT.

   ZL            [dex],dblarr(2), limits for the [Fe/H],
                 by default = [-2.5, 1.0] see ULY_TGM_INIT.

   NAME          A string, name of this component. By default a unique name is 
                 generated.
   
   FIXPAR        An array used to fix some the atmosphere parameters,
                 0 means that the parameter is free, and 1 that it is fixed.
                 A fixed parameter keeps its initial value.
                 The parameters are specified in the following order:
                 (Teff, Logg, [Fe/H]).
          
   WL            Limits for the weight of the component [min_w, max_w].
                 The weight is in data units over component units.
                 By default the weight is constrained to be positive.
                 This constraint is applicable when multiple components are
                 fitted (composite spectrum), it is ignored for a single
                 component.

   /REBIN_COEF   Computation mode, see the description below.
       
 DESCRIPTION:
     Return a TGM structure describing a TGM component that can be 
     passed to program ULYSS. 

     TGM stands for Temperature-Gravity-Metallicity: A TGM component 
     specifies an interpolated stellar spectrum evaluated as a function 
     of its atmospheric parameters (Teff, Logg and [Fe/H]) that was 
     interpolated over a stellar spectral library (e.g. ELODIE 3.2) 
     or over a grid of synthetic spectra. 

     The file passed to ULY_TGM contains the coefficients of polynomials 
     depending on the atmospheric parameter. Each wavelengh bin is represented
     separately, therefore the coefficients for each term of the development
     form a spectrum. Depending on the interpolator, there are 14 to about
     25 terms. The first polynomial interpolator was presented in 
     Prugniel and Soubiran 2001, A&A 369, 1048 for the ELODIE spectral
     libraries, it was updated in subsequent papers, in particular in 
     Wu et al. 2011, A&A 525, 71. Interpolators for the MILES library were
     presented in Prugniel et al. 2011, A&A 531, 169, and Sharma et
     al. 2015, submitted. Other interpolators are available on the
     ULySS site.

     The interpolated spectra are evaluated using the interpolator, and
     then rebinned to the WCS of the spectrum to analyse. 

     REBIN_COEF option:
     Rather than rebinning the interpolated spectrum, it first rebin the 
     coefficients, and then compute the interpolated spectrum. This option is 
     slightly less precise (because of the spline interpolation of the several
     coefficients), but is in some cases faster. This is the case when (i)
     a series of several spectra are fitted with the same CMP, and (ii) when
     the spectrum to analyse has a sampling much coarser than the interpolator.
     (Note that the impact on the precision of the interpolated spectrum
     is generally fully acceptable; it was the only available option before
     the release 1.3 of ULySS.)
     
 OUTPUT:
     TGM cmp struct, for detailed explanation please check uly_fit.pro        

 REQUIRED FUNCTION:
                 ULY_CMP_NAME
                         
 EXAMPLE:
     Fit a CFLIB star: 
     cmp = uly_tgm()
     star = uly_root+'/data/cflib_114642.fits'
     ulyss, star, cmp, lmin=3900.0, lmax=6800.0, /PLOT

 HISTORY:
                 Creation Yue WU 2008/06/18

(See pgm/uly_tgm.pro)

 NAME:            ULY_TGM_EVAL
 
 PURPOSE:           
                  Evaluate a TGM model spectrum.
 
 USAGE:
                  array = uly_tgm_eval(eval_data, para)
 
 ARGUMENTS:
   EVAL_DATA:     The interpolated stellar spectrum model coefficients, 
                  it contains three dimensions of data, specifying to
                  warm, hot and cold stellar regions respectively, in
                  fact, they are the coefficients of the evaluating 
                  polynomial function.

   PARA:          Input 3 atmospheric parameters (log(Teff), Logg, [Fe/H]),
                  they will be passed to ULY_TGM_MODEL_PARAM to
                  compute out the parameters which will be applied
                  to the polynomial function for evaluating
                  the model spectrum at this given stellar
                  atmospheric parameters (Teff, log g and [Fe/H]).

 DESCRIPTION:
      Evaluate a TGM model spectrum which is interpolated as a
      function of atmospheric parameters (effective temperature,
      surface gravity and [Fe/H]). Each wavelength point is evaluated
      by using polynomials.

      This function is automatically called by the minimization routines:
      ULY_FIT, ULY_FIT_LIN to evaluate a TGM model (see ULY_TGM).
 
 OUTPUT:          
                  Evaluated stellar model spectrum               

 REQUIRED FUNCTION: 
                  ULY_TGM_MODEL_PARAM, compute the parameters 
       
 AUTHOR:
                  Yue WU, 2008/06/18

(See pgm/uly_tgm_eval.pro)

 NAME:                ULY_TGM_EXTR

 PURPOSE:
             Compute a stellar template spectrum using TGM model
                      
 USAGE:               
               model = uly_tgm_extr([<spectrum>], <pars>, MODEL_FILE=<file>)

 DESCRIPTION:
      Compute a 1D stellar template spectrum by using TGM model at a given
      Teff, Log g and [Fe/H]. Rebin it with the same sampling as <spectrum>
      and with a compatible grid.

      If the natural wavelength range of the model is wider than the one
      of <spectrum> the returned spectrum has exactly the same sampling as 
      <spectrum> (same range and same step). If instead the wavelength
      range of the model is narrower, the output spectrum will be truncated
      with respect to <spectrum> but its wavelength bins will match bins
      of <spectrum>.

      Note that although the model may be rebinned, its resolution is not 
      matched to that of <spectrum>.

      If <spectrum> is omitted, i. e. if only one argument is given, the
      returned spectrum has the original sampling of the model.
       
 ARGUMENT:
      <spectrum>    filename or spectrum structure serving as a reference 
                    for the velscale.

      <pars>        array of parameters([Teff,Logg,[Fe/H]) used
                    to evaluate the template spectrum.

 KEYWORD:              
      MODEL_FILE = <file>
         Filename of the model to use, by default:
         uly_root + '/models/elodie32_flux_tgm.fits' 
 
 RETURNS:    
      A 1D stellar spectrum as a spect structure, see ULY_SPECT_ALLOC
            
 EXAMPLE:
      To compute a template spectrum at a given atmosphere parameters
      according to uly_root+'/data/cflib_114642.fits's velscale.
            model = uly_tgm_extr(uly_root+'/data/cflib_114642.fits', $
                                 [6400., 4., 0.])
 
 HISTORY:
            Yue WU 2008/08/20 

(See pgm/uly_tgm_extr.pro)

 NAME:
               ULY_TGM_INIT
 PURPOSE: 
               Initialize a TGM component
             
 USAGE:
               cmp_new = uly_tgm_init (cmp, WAVERANGE=lamrange, $
                                       VELSCALE=velscale, QUIET=quiet)

 ARGUMENTS:   
   CMP:        Component defined by using ULY_TGM, please check it for
               more details.

 KEYWORDS:
   WAVERANGE:  Wavelength range used when the model will be
               evaluated with ULY_TGM_EVAL.

   VELSCALE:   Size of one pixel in km/s. Used when the model will be
               evaluated and log-rebinned with ULY_TGM_EVAL.

   QUIET:      verbosity control

 DESCRIPTION: 
      This initialization function is automatically executed by 
      ULY_FIT_INIT when a component has been defined by using
      ULY_TGM. A TGM model component can generate a spectrum at given 
      stellar atmospheric parameters (Teff,log g and [Fe/H]). 

      The name of this function is recorded as a member of
      cmp.init_fun in the TGM component structure (see ULY_FIT for a
      complete description of the component structure).

      ULY_TGM_INIT consists in loading the basic TGM model
      coefficients from a FITS file whose name is saved in
      *cmp.init_data (by default
      /models/elodie32_flux_tgm.fits), and also loads in the cmp 
      structure the WCS.

 NOTE
 ULY_INIT_TGM is also called by ULY_TGM_EXTR where we use the keywords
 SAMPLING, WAVERANGE, STEP, which are more general because they can make
 either lin- or log-sampled spectra.
 This more general approach may be used in ULY_FIT_INIT in the future, when
 we will implement fitting in other than log- (not however that it is not
 fully general, because uneven sampling would not be supported).

 HISTORY:
               Creation Yue WU 2008/06/18
 

(See pgm/uly_tgm_init.pro)