Pulsation Search (gtpspec) Tutorial

Also see:

Tutorials

SciTools References:

Popup Navbar Windows:

The gtpspec tool searches for pulsations in a wider range of frequencies, using the Discrete Fast Fourier Transfer (FFT) technique to compute power spectrum density. It can be used as a part of a so-called blind period search, in which data are examined for pulsations at any frequency.

Known Issues: When running gtpspec multiple times using the GUI, plot windows from previous runs reappear after they are closed manually. Thus, there is no way to permanently close plot windows without exiting the GUI.

Prerequisites

  • Event data file in FT1 format, also known as a photon data file. (See Extract LAT Data.)
  • Orbit file to use for the barycentric correction
  • Pulsar ephemeris database file which contains binary orbital parameters of a pulsar of interest

Sample Files. To try the examples in this section, you can download the following fake data files:

Note: The output of gtpspec consists of text describing the result of the pulsation search and an optional plot. This tool also creates an output file when requested. The output file contains the result of computation, i.e., the search result in the text output and the data array to plot, for future reference.

How to Perform a Simple Search

Assuming No Frequency Variations

The example below shows how to perform a pulsation search in the simplest case, where one can assume no significant frequency variations exist throughout the observation. The time origin of this search is taken to be the center of the observation. That means, if a pulsation is found, the pulse frequency should be interpreted as measured at the center of the observation (i.e., the time origin).

The command shown below produces the following output message that describes the search criteria and explains the search result.

A plot similar to the following will also be produced:

gtpspec example 1 (graphical output)

This plot very likely shows a pulsation consistent with the ephemeris of the simulated pulsar.

Correcting for Frequency Variations

In a full blind search for pulsations, it may be desired that a time derivative of pulse frequency is also scanned. This tool does not directly perform such a full search, but it supports frequency derivatives in a pulsation search. The gtpspec tool corrects for variations in the frequency, taking one set of frequency derivatives at a time. For the following example, the frequency derivatives were taken from the Monte Carlo truth of the simulated pulsar.

gtpspec example 2 (graphical output)

The effect of correcting for frequency variation was to increase the value of the maximum statistic, or the height of the main peak.

Getting More Information on Computation

If gtpspec is invoked with the chatter parameter set to 4 or larger, it will display brief summaries of loaded ephemerides, arrival time corrections, and time systems being used, followed by a description of test condition and test result. This feature shows you additional information on computations being performed, which may help you understand the tool. Also, the extra information printed may contain a clue to diagnose certain types of problems. So, try a high chatter when you are unsure about the tool's behavior. Note that in the following example, the middle part of periodgram data is omitted.

Going Beyond The Simple Case

How to Perform a Full Pulsation Search

When searching for a pulsation, you may need to take into account the time derivatives of pulse frequency that are also unknown. By running gtpspec multiple times with different sets of frequency derivatives, you can sweep a desired frequency range and its derivatives to perform a full blind search for pulsations in a range of pulse frequencies of interest. Note that a chance probability in a text output is for a given set of frequency parameters only. When running gtpspec multiple times, the chance probability of the entire pulsation search must be computed separately using the information shown in the text output such as the number of independent trials and the maximum statistic.

Working in the Period Domain

All of the examples above used inputs in the form of frequencies and derivatives of frequencies, but it is also possible to enter the input data as periods and derivatives of periods. However, these values are immediately converted to the frequency domain, and the search is still performed in the frequency domain. To enter period information instead, select PER for the ephemeris style. For example, the following input should produce the same plot as the example Correcting For Frequency Variations.

Owned by: Masaharu Hirayama 04/17/2009
Last updated by: Masaharu Hirayama 04/17/2009