ctools

The tool has been developed and somewhat tested. Phase information is derived using the GModelTemporalPhaseCurve class and the user supplied reference values for:

• “MJD” (reference MJD)
• “P0” (phase at reference MJD)
• “F0” (frequency at reference MJD in Hz)
• “F1” (first frequency derivative at reference MJD in sec^(-1))
• “F2” (second frequency derivative at reference MJD in sec^(-2))

The phase information is successfully appended to the events file, however precision issues can become a problem. These issues appear to only occur when a reference date far from the MJD of the data is chosen (more than a few years difference in the dates) or when a large value for the first/second derivate are chosen. For the time being, a warning is printed into the ctphase log file when either (F1 * dt) or (F2 * dt * dt) is greater than 10^8 (dt is seconds between the reference MJD and the MJD of the events). When this occurs the user should check that the distribution of derived phases makes sense. The method should be stable for most sensible choices of the parameters.

Tests of the above issue have been done simulating events with ctobssim and then passing the resulting events file through ctphase. These results are documented below:

Example plot for 3 minute period:

• Data MJD: 51544.5
• Reference MJD: 59000
• P0: 0.0
• F0: 0.00555556 (1/180)
• F1: 0.0
• F2: 0.0
  

Example plot for sensible, uniform distribution of phases as a function of event time:

• Data MJD: 51544.5
• Reference MJD: 59000
• P0: 0.0
• F0: 1.0
• F1: 1e-8
• F2: 0
  

Example plot showing precision issues (generated phases do not appear uniformly distributed)

• Data MJD: 51544.5
• Reference MJD: 59000
• P0: 0.0
• F0: 1.0
• F1: 1.0e-8
• F2: 1.0e-8
  

Added ability to obtain the phase information from a model xml file. User needs to specify two parameters:

• inmodel (model XML file containing source information)
• source (Name of the source in 'inmodel’ from which phase information will be taken)

If either of these parameters is not specified, then the phase information is derived from the parameters in the ctphase.par file. An exception is thrown if either of the following occurs:

• Model XML file does not define a source with name 'source’
• 'source’ does not define a temporal model with phase information.

This functionality was tested using a temporal model for the Crab of:

  <source name="Crab" type="PointSource">
    <spatialModel type="PointSource">
      <parameter name="RA"  scale="1" value="83.6331" min="-360" max="360" free="0"/>
      <parameter name="DEC" scale="1" value="22.0145" min="-90"  max="90"  free="0"/>
    </spatialModel>
    <spectrum type="PowerLaw">
      <parameter name="Prefactor"   scale="1e-16" value="5.7"  min="1e-07" max="1000.0" free="1"/>
      <parameter name="Index"       scale="-1"    value="2.48" min="0.0"   max="+5.0"   free="1"/>
      <parameter name="PivotEnergy" scale="1e6"   value="0.3"  min="0.01"  max="1000.0" free="0"/>
    </spectrum>
    <temporal type="PhaseCurve" file="model_phase_info.fits">
      <parameter name="Normalization" scale="1" value="1.0"     min="0.0" max="1000.0"   free="0"/>
      <parameter name="MJD"           scale="1" value="51544.5" min="0.0" max="100000.0" free="0"/>
      <parameter name="Phase"         scale="1" value="0.0"     min="0.0" max="1.0"      free="0"/>
      <parameter name="F0"            scale="1" value="1.0"     min="0.0" max="1000.0"   free="0"/>
      <parameter name="F1"            scale="1" value="0.1"     min="0.0" max="1000.0"   free="0"/>
      <parameter name="F2"            scale="1" value="0.01"    min="0.0" max="1000.0"   free="0"/>
    </temporal>
  </source>

'model_phase_info.fits’ is attached and defines node information in the following way:

Generating events from only the above Crab model and processing through ctphase produces the following distribution of events (as a function of phase):

Processing the same data through ctphase, but using random values for the phase information produces the following (seemingly random) distribution of events. This is the expected behavior.

I presume that the features requested are now implemented and a pull of this code can be done. Note, a pull of the associated gammaLib branch is also necessary for appropriate functionality, otherwise errors will be produced by GModelTemporalPhaseCurve.

Relevant repositories for pull request:

• GammaLib: git@cta-gitlab.irap.omp.eu:jcardenzana/gammalib.git
• GammaLib branch: 1982-add-phase-info
• CTools: git@cta-gitlab.irap.omp.eu:jcardenzana/ctools.git
• CTools branch: 1982-add-phase-info-ctphase

I integrated the GammaLib branch 1982-add-phase-info.

I’ve seen that you allowed an empty filename in the GModelTemporalPhaseCurve::read() method, however I’m worried about the possible side effects of that. In particular, a phase object without any nodes does not really make sense. What is the Use Case of this functionality? For the moment I have not merged in that part.

I merged the ctphase tool into devel.

I also added a page to the reference manual.

I’ve made the above change and verified that the results are consistent with those presented previously. I’ve attached my test python script (with supplemental files) that produces the events, then stores the phase information and plots the distribution of events in “phase”. Note that the 'model_phase_info.fits’ file attached in a previous post is also necessary to get the node information in the pulse profile.

Relevant repositories for pull request:

• GammaLib: git@cta-gitlab.irap.omp.eu:jcardenzana/gammalib.git
• GammaLib branch: 1982-add-phase-info
• CTools: git@cta-gitlab.irap.omp.eu:jcardenzana/ctools.git
• CTools branch: 1982-add-phase-info-ctphase

Thanks. Note that I modified two parameter names so that they are compliant with the names of other tools:

source -> srcname
p0     -> phase

I modified a bit the script to avoid using astropy (which I don’t have installed): test.py