In some occasions there may arise the necessity to perform photon selections without any binning of the selected photons into images, light curves, spectra or profiles. This would be done via Projection pipelines and command files where no BIN commands are defined. The output of the photon selection would be a new Photon Event Table containing only the filtered photons. In the following examples a few pipelines performing pure photon selection are shown:
Midas 001> SELECT/RING cursor SELECT TIME 10000,20000 *events *outlist Midas 002> SELECT/AMPL >150 input=events output=hard_photons Midas 003> SELECT/SECTOR center=(100,-1200) angles=180,360 *events *list
The usage of pure selection pipelines and command files is not advisable. They should be created only when a specific scientific task must be carried out on the resulting photon list itself. For instance, if a user is interested in having a spectrum from a point source within a given time interval, he will not use the sequence of commands
Midas 001> SELECT TIME 200000 to 300000 in=events out=sublist1 Midas 002> SELECT/RING cen=(100,200) rad=500 *sublist1 *sublist2 Midas 003> BIN AMPL 1 input=sublist2 output=spectrum
This would take generally a long execution time and moreover it would create the two huge (undesired) tables sublist1.tbl and sublist2.tbl. Instead, with the single pipeline
Midas 004> SELECT TIME 200000,300000-
SEL/RING (100,200),500-
BIN AMPL 1 *events *spectrum
would be generated the same result in a very short time and without
ever filling the working disk area with huge files.
To keep photon event tables containing selected photons is almost never efficient. The photon events tables produced with Projection are not provided a sky map structure for fast access (a sky map structure could be recreated via the MAKE/PEM command, but here we go again with some extra time lost). This is because the photon event tables which are produced are supposed to be generally much smaller than the original input ROD photon event table.
Therefore it would be generally a very bad choice to decide to perform -- for instance -- a time screening of photons before defining any required spatial selection, like in the following example:
Midas 005> SELECT TIME 10000 to 20000 in=events out=time_screened_list
There is no hope of working efficiently by accessing the new table time_screened_list. It is much more efficient to define the time screening in each pipeline which is used in the production of the binned datasets:
Midas 006> SELE TIME 10000,20000 SEL/RING cursor BIN TIME 0.7 *events *rate
Thanks to the sky map structure the execution time is in this case definitely much shorter.