When invoking PSS, three parameters have command line positions defined -- these are the same for all modes.
> PSS <input_image> <background> <results>If the information present in your dataset is sufficient for PSS to work with (as it is in the demo file), then PSS could be run using the following,
> PSS MY_PSS_DEMO 1.01 MY_SOURCES \using the ADAM command line token
\ to suppress all
prompts for which defaults exist. However, for the purposes
of a useful tutorial, the prompts
and their uses will be explained.
Start with simply the input image name,
> PSS MY_PSS_DEMO PSS Version 1.6-1 EXPERT - Expert mode /NO/ >and PSS responds with its version number followed by the EXPERT parameter prompt. Accepting
NOEXPERT means PSS will minimise user interaction by assuming
sensible defaults for control parameters ( usually chosen to minimise
processing time ). Specifying EXPERT true will result in between
five and ten extra prompts depending on the PSS mode. Accepting the
defaults for
all these prompts will give exactly the same behaviour as non-expert mode.
Next PSS requires the source model to be specified,
PSF system options :
ANAL EXOLE PWFC TABULAR WFC
XRT_HRI XRT_PSPC
POLAR(psf,rbin[,abin]) RECT(psf,xbin[,ybin])
PSF - Choose PSF to use for source model /'WFC'/ >
Filter Lexan/C/B4C
The ASTERIX psf system (see Appendix A) allows the PSS
source model to be manipulated in an instrument
independent fashion. For datasets produced by an
instrument interface a default for PSF parameter should be
offered. The default in this case is WFC -- the demonstration is
a simulated WFC survey image. On accepting this default the WFC psf
routine displays the WFC filter being used.
In search mode PSS compares the input data and model in a box which is passed over the image. The size of this box is the principle factor controlling both the accuracy of the results, and the speed with which they are produced.
Using a psf box of radius 4 pixels Using a psf constant across the fieldIn non-expert mode PSS chooses a box size which encloses a minimum of 68% of the psf energy. This is a reasonable compromise between accuracy and speed (see Section 4.4). In expert mode, the user has control over this box size. The second line of text indicates that PSS will access the psf only once for the whole field. The alternative, a fully varying psf, is computationally more expensive but required for some detectors (eg. ROSAT PSPC). PSS only chooses a varying psf as default if the user supplies a psf model specification (see Section A.4), thus stating explicitly that the psf varies.
In SEARCH mode a subset of the image may be searched. This may be useful in avoiding areas of bad background subtraction, or simply to cut down the processing time. PSS informs the user of the extent of the image and prompts for a slice.
X_CORR axis range is from 29.5 to -29.5 arcmin Y_CORR axis range is from -29.5 to 29.5 arcmin SLICE - Section of dataset to search /'*:*,*:*'/ >The default offered is the entire image. To select the central 40 arcminutes box the reponse to this prompt would have been
20:-20,-20:20 (note that the range values must confirm to the
direction of increase of the axis values). See
Ref 5,`` User Interface,Lists and Ranges"
for a description of the ASTERIX range syntax.
The SOPT prompt selects the fitting statistic, the default is always
CASH,
SOPT - Statistic option (CASH,GAUSSIAN) /'CASH'/ >
PSS now requires a background model. There are two options, either a file name or a constant value. The latter is a suitable choice here as the tutorial dataset was created using a flat background. The units for such a value are image data units (counts in this case) per pixel.
BGND - Background model > 1.01 Using background value 1.01 for entire image...PSS informs you that it is using a constant background.
In SEARCH mode PSS constructs a map of the statistic covering the area selected using the SLICE parameter. In non-expert mode, the point spacing of this map is the same as the pixel spacing in the input image.
First pass - grid spacing 1 pixels MAP - Significance map /!/ > sig_mapConstructing the significance map is usually the most time consuming part of PSS. Once it is complete, the option to output the map to an external file is given -- enter a file name or type RETURN to ignore the prompt. If output, the map inherits all the ancillary information associated with the input dataset, enabling significance contours to be overlaid on the original data, for example. The map becomes the default ASTERIX graphics object, and so could be displayed on an already opened device with,
> DRAW \
The range of the significance map is stated and a threshold for source detection selected.
Significance varies from 0 to 26.53978 SIGMIN - Significance threshold /5/ >Once SIGMIN is supplied PSS searches for peaks in the map above this threshold -- a peak being defined as a point higher than all surrounding points. PSS lists all peaks higher than 80% of the supplied threshold. This 20% factor is allowed on the first pass as the initial search grid will not find the highest part of a source peak unless the position of best source fit lies exactly on a grid point. A second threshold value may be supplied after the first, separated by a comma.
SIGMIN - Significance threshold /5/ > 3,5This mechanism allows less peaks below the first threshold to progress to fitting (see below) and possibly extension testing in expert mode, but the second threshold is applied before sources are output.
The list of source candidates is printed,
Src X Y Signif
1 0.38 0.38 26.540
2 16.62 9.63 9.789
The table shows source number, image position and peak significance. PSS
now performs a grid search around each source candidate, but with the grid
centred on the best source position. This process is repeated a few times
to optimise the source position.
Second pass Symmetric flux errors at 68% confidence, and position errors at 90%PSS now finds the errors on the source parameters. The error confidence levels are chosen automatically in non-expert mode. Once the errors have been found PSS prints another source list with the user's significance threshold applied firmly, sorted into increasing Right Ascension.
Src RA DEC X Y Signif Flux
1 00 01 07.7 +00 10 04 16.93 10.07 9.892 95.579
2 23 59 59.8 -00 00 01 -0.04 -0.01 27.463 358.160
OUT - Source search results file (! for none) /@SRCLIST/ > my_sources
The option is also given to output these results to a source
search results file (SSDS). This is an HDS file which ASTERIX
software uses to store such information. If the printed data
is sufficient, reply with ! to this prompt. Note however,
that the SSDS contains all error information and provides a
simple mechanism for passing source positions around the ASTERIX
system.