1. Hunting for Vela and Crab in the DC2 data
A tutorial on pulsar analysis with the LAT Science Tools
Massimiliano Razzano
(INFN Pisa)
DC II Kickoff Meeting
(SLAC, March )
02/03/2006
M.Razzano - DC II Kickoff Meeting

2. Summary
In this tutorial we will see how to use the LAT Science Tools to perform a pulsar timing analysis in the DC 2;
The dataset used is the first week of DC 2, available at:
Everyone can gain pratice and reproduce this tutorial !
The Science Tools version is v6r4p1. Updated tools with respect to previous tutorials (e.g. at the Collaboration Meeting);
We choose Vela and Crab, famous and bright g-ray pulsars (we use just 1 week…)
Step-by-step procedure to test periodicity and recover pulse profile for g-ray pulsars with radio counterpart;
Included command prompt screenshots for your reference;
02/03/2006
M.Razzano - DC II Kickoff Meeting

3. gtpulsardb, gtephcomp
The pulsar tools
Select the region and apply the cuts on time, energy etc…;
gtselect
Apply the barycentric corrections;
gtbary
Retrieve pulsar ephemerides from the pulsar database;
gtpulsardb, gtephcomp
Test periodicity;
gtpsearch
Assign phase;
gtpphase
Display phase curve;
What do you prefere: fplot, fv, ROOT, Hippodraw, etc..(here we use fv…)
LAT pulsar Tools
Pulsar tools developed mainly by M.Hirayama and J.Peachey (
02/03/2006
M.Razzano - DC II Kickoff Meeting

4. Selecting the region of interest For this task we use gtselect
The first week of DC2
The radius of the sky region selected for the analysis should be chosen in order to maximize S/B ratio;
A radius comparable with PSF is ok (a good choice could be 2 PSF).
PSF depends on energy, so an energy-dependend cut is optimal (as was done in EGRET)
We make here a rough selection: E>100MeV, and PSF(100MeV) ≈3°R=6°
For this task we use gtselect
02/03/2006
M.Razzano - DC II Kickoff Meeting

5. Selecting Vela and Crab
Note that time is in MET !
In a similar way as for Vela, we cut on Crab, but another source seems to be there. To avoid possible confusion on the periodicity test, we use a 3° radius
Crab
02/03/2006
M.Razzano - DC II Kickoff Meeting

6. For this task we use gtbary
The barycentric corrections
The photon arrival times are affected by the motion of GLAST through Solar System and by relativistic effects. These effects are compensated by the barycentric corrections
Corrections are:
Conversion TTTDB;
Geometric corrections due to lighttravel time from GLAST location to Solar System Barycenter;
Relativistic delay due to gravitaional field of Sun (e.g. Shapiro delay);
For this task we use gtbary
Pay attention:gtbary usually overwrites the input file!
The barycentric corrections convert the photons arrival times, (expressed in Terrestrial Time TT at the spacecraft), to the arrival times at the Solar System Barycenter (expressed in Barycentric Dynamical Time TDB)
02/03/2006
M.Razzano - DC II Kickoff Meeting

7. Retrieving pulsar ephemerides
Pulsar rotational frequency decreases with time (period increase). Usually the behaviour of frequency with time is approximated by a 2-nd order polynomial:
The parameters f0,f1, f2 are called ephemerides and t0 is the epoch, i.e. the time where the ephemerides refers to (usually t0 is in Modified Julian Date MJD)
The pulsar ephemerides are stored in our pulsar database (D4).
In order to test the periodicity of a pulsar and assign to each photon a rotational phase, it is necessary to know the ephemerides (e.g. from radio)
For this task we use gtpulsardb and gtephcomp
02/03/2006
M.Razzano - DC II Kickoff Meeting

8. A quick look at the pulsar database…
Download pulsar database (D4) from GSSC Data Center.
Pay attention to names: look at the D4..
(m stands for “-”, p stands for “+”)
02/03/2006
M.Razzano - DC II Kickoff Meeting

9. There are ephemerides for our pulsar?
Because in real life the ephemerides timing solution comes from observations, it is possible that there are no radio ephemerides coverage of our observation time. With gtephcomp it is possible to check that.
Strict options force to use ephemerides that cover the input time
In this case the tool checks if there are ephemerides that contains our time (in this case the middle time of the observation.
It also estimates the ephemerides at our input time
In this case we have valid ephemerides covering the observation (MJD )
We will see with Crab the other possibility = no radio ephemerides available…
02/03/2006
M.Razzano - DC II Kickoff Meeting

10. For this task we use gtpsearch
Periodicity tests
For this task we use gtpsearch
Once we know that D4 contains valid ephemerides of Vela, we can test if our gamma source has the same periodicity of the radio counterpart. In this case we have identified the source as a gamma-ray pulsar!
Tests against the null hypotesis:
H0 = no periodicity
When we will find the periodicity, we can assign phases to each photon and build the lightcurve
Tests implemented:
Chi-squared test (Leahy et al. 1983,ApJ 266;
Z2n test (Buccheri et al A&A128),Rayleigh test;
H test (De Jager et al., 1989 A&A 221)
02/03/2006
M.Razzano - DC II Kickoff Meeting

11. Periodicity test on the profile
Example of c2 periodicity test
Frequency space is scanned in fractions of Fourier resolution fF=1/T, where T is the duration of the observation. It represents the spacing between 2 independent frequencies in FT.
For our data fF 1/(86400*7)≈1.6E-06 Hz
Here we use gtpsearch (/periodsearch package) and use as example the c2 test
For each photon a phase is assigned for a trial frequency, then an histogram of phase bin mj is created;
In absence of pulsation every phase bin will have the same mean count mexpected;
The quantity:
is computed;
For large counts is every bin, S is distributed as a c2
Then it is possible to test the non-periodicity hypothesis and give a chance probability p(c2>S)
Vedere quant’e’ l’esposizione totale
02/03/2006
M.Razzano - DC II Kickoff Meeting

12. Testing Vela periodicity (I)
As first approximation we use the DB option that tell gtpsearch to extract ephemerides from D4.
Pay attention to specify the psrdbfile parameter!
The cancelpdot parameter tells to take into account the period derivatives
02/03/2006
M.Razzano - DC II Kickoff Meeting

13. Testing Vela periodicity (II)
As a first iteration we adopt as central frequency the frequency of the maximum at the previous run.
The frequency has the same value as before, we can proceed to a refined scan in frequency
02/03/2006
M.Razzano - DC II Kickoff Meeting

14. Testing Vela periodicity (III)
After this last run we’ve found that our source has the periodicity of the radio counterpart.
fmax= Hz with a freq. sampling of≈8E-8 Hz
We can compare with the value f0= Hz
|f0 – fmax| < freq. sampling
02/03/2006
M.Razzano - DC II Kickoff Meeting

15. Folding at the radio ephemerides For this task we use gtpphase
Once the periodicity is known, to each photon can be assigned a rotational phase as:
Gtppphase add a PULSE_PHASE column. We make an histogram with fv
... ) ( 6 1 2 φ( 3 + - = t f
dt=t-t0, t0 is the epoch
For this task we use gtpphase
Here we choose to use the D4 ephemerides, but they can be also entered manually
02/03/2006
M.Razzano - DC II Kickoff Meeting

16. Z2n test and H test Z2n test H test
Other 2 tests are implemented in gtpsearch: Z2n test and H-test. For more details, please see references.
The other 2 tests give similar results.
The number of bins is Z2n is equivalent to the number of harmonics we want to consider.
Z2n has p.d.f of χ22n
(See for details:
Buccheri et al. 1983, A&A128)
Z2n test
The H test is more efficient for unknown-a-priori lightcurves
(see for details:
De Jager et al., 1989 A&A 221)
H test
02/03/2006
M.Razzano - DC II Kickoff Meeting

17. What happens if Crab is not in D4?
It can be possible that no radio ephemerides cover our observation. To perform timing analysis in this case we’ve shifted ephem of Crab outside the DC2
It is possible to use the D4 entries in the D4 to estimate ephemerides in our dataset (by using the 2 nd order approx.)
With strict=yes we see that no radio ephem are available
T is in MET !
Then we will use strict=no to extrapolate our ephemerides
at the center of the observation time
For this task we use gtpephcomp
We will use these extrapolated data to test periodicity
02/03/2006
M.Razzano - DC II Kickoff Meeting

18. Testing periodicity (I)
We use gtpsearch by entering manually ephemerides
In this case the ephemerides validity ends just before DC2 starts, so extrapolation works fine.
In addition Crab is bright, but this will be not the case for other pulsars in DC2…
For instance, this is ad hoc example: at least Crab has a complete ephemerides coverage in the official DC2 pulsar database…
02/03/2006
M.Razzano - DC II Kickoff Meeting

19. Testing periodicity (II)
We make another iteration with refined freq-scan
We then assign phases according to the found frequency and we find lightcurve
02/03/2006
M.Razzano - DC II Kickoff Meeting

20. Concluding remarks
This tutorial shows basic analysis steps using the Science Tools. More refined analysis can be done.
No spectral analysis shown. You can make spectral analysis as for other sources (e.g. with likelihood or with other specific packages, e.g XSpec);
When retrieving ephemerides, please refer to our D4 (most of DC2 pulsars are not in the real sky…);
For major info on Pulsar Tools, please see the User Workbook or the Pulsar Tools page at:
Take this tutorial as a template to have pratice with basic analysis for DC2;
You can also have exercises by simulating pulsars with gtobssim and PulsarSpectrum;
There are many pulsars out there: good luck and enjoy!
02/03/2006
M.Razzano - DC II Kickoff Meeting