MAKIMG -- make a starting image for doppler tomography or for simulating
data.
Parameters:
NSIDE -- Size of images (which are square)
NIMAGE -- Number of images. Blended lines can be dealt with by
assigning an image for each line. It is also possible
for the same image (bar a scaling factor) to represent
more than one line.
IMAGE -- The name of the image file to create
VPIX -- Size of pixels in km/s. The product of NPIX and VPIX should
more than cover the width of the line of interest. VPIX should
be less than or equal to the km/s/pixel of the data. Note
there is no problem if VPIX is much smaller, except in the
extra CPU and memory required. If you have blended lines
the images need only extend over each line, not the whole
blend.
NWAVE -- Number of lines. e.g you could map the Paschen series and
the CaII triplet with 2 images, one for CaII and one for the
hydrogen lines, but you would have to specify a wavelength for
each line to be mapped. NWAVE >= NIMAGE
WAVE(NWAVE) -- Central wavelength of each line in Angstroms
If (NWAVE .NE. NIMAGE)
NWHICH(NWAVE) -- Defines which image corresponds to each line. e.g.
If you had two image accounting for three lines
you might set NWHICH = 1,2,2 which means that the
first image corresponds to the first wavelength while
the second image corresponds to the other 2 lines.
SCALE(NWAVE) -- Defines scale factor to apply to each line (allows for
different lines to have same image but a different
scaling.) These values can be optimised outside MEM
using optscl. e.g. on the above example you might
have SCALE = 1.,1.,0.5
FWHM -- Full width half maximum of instrument, km/s. Don't make too
large or it will be attempting an impossible deconvolution.
Too small does not matter in this respect, but a finite
FWHM is good for smoothing out numerical noise.
GAMMA -- Systemic velocity of system km/s.
NSUB -- Subdivision factor to account for phase smearing from finite
exposure length. Each exposure will be divided into NSUB
exposures uniformly spread through the exposure and then
trapezoidally averaged. If your exposures are short compared
to the orbital period, set = 1. If you do use it, only
apply towards the end of the iterations because it slows
things down (by a factor NSUB).
OFFSET -- Constant to add to all phases to account for zero point
error in the ephemeris. This rotates the entire image but
the image must be re-iterated after any change to OFFSET
for this to be seen.
NDIV -- The projections of model to data is done first onto a
fine grid which is then blurred and binned onto the data
array. This is usually faster than blurring each pixel
directly onto the coarse grid. NDIV is the factor by
which each data pixel is split to give the fine array.
NDIV must be large enough so that the FWHM is well
sampled. e.g if FWHM=2, then NDIV>2 should be used.
There is no reason not to change NDIV as you iterate
and you could for example start with NDIV = 1 and only raise
it for the final few iterations. Note that large NDIV values
will slow things down so be careful.
(Use 'setobj 5 image.more.doppler.ndiv' to set ndiv = 5 for
example)
USE -- This specifies whether the HJDs or phases should be used
to compute the phases during iterations. The option of
HJDs is provided to allow one to account for changes in the
ephemeris, however the method of choice should normally be
'phases'. The reason is that if the data comes from a
phase-folding operation (pbin inside molly) while the
phases are correct, the HJDs are not. e.g. say you average
two spectra at phase 0.5 but 1 cycle apart. The HJD will be
set half-way between i.e. phase 0!. The phase on the other
hand is correctly dealt with. You will be OK with either
option if you are dealing with straight spectra.
The two options are:
HJDs -- use HJDs
Phases -- use phases
and you can change them by editing the NDF with setobj.
Now parameters defining shape of image:
FANCY -- .FALSE. you just get a constant image. .TRUE. then you
are prompted for many more parameters to define a gaussian
spot, a spiral pattern and a power law in radius.
If(FANCY)
XCEN -- X centre of gaussian spot (km/s)
YCEN -- Y centre of gaussian spot (km/s)
WIDTH -- FWHM of gaussian spot (km/s)
PGAUSS -- Peak intensity of gaussian
XSP -- X centre of spiral pattern and power law disc (km/s)
YSP -- Y centre of spiral pattern and power law disc (km/s)
VLOW -- Lowest disc velocity (outer disc) (km/s)
VHIGH -- Highest disk velocity (inner disc) (km/s)
EXPON -- exponent between them (intensity scales as V**EXPON)
ANGLE -- Opening angle of two armed spiral pattern (degrees)
CURVE -- Difference in angle between outer and inner velocities
PPOWER -- Peak intensity of power law background
PSPIRAL -- Peak intensity of spiral shock
Else
CONST -- Constant per pixel
Don't worry about the levels you set; these can be tweaked with optscl.
This command belongs to the class: essential