A repository for the Second IPTA Mock Data Challenge datasets.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


International Pulsar Timing Array Mock Data Challenge #2



  • MDC2 is open and accepting submissions.
  • The deadline for submitting analyses for MDC2 is, March, 15 2019.
  • The datasets were updated on December, 1 2018 to reflect a more "usual" fit for the timing model. Please use these updated data sets in any submissions. Please contact the organizers if you have any questions or requests.

The International Pulsar Timing Array (IPTA) is a galactic-scale gravitational-wave observatory that monitors an array of millisecond pulsars. The timing precision of these pulsars is such that one can measure the correlated changes in pulse arrival times accurately enough to search for the signature of a stochastic gravitational-wave background. As we add more pulsars to the array, and extend the length of our dataset, we are able to observe at ever lower gravitational-wave frequencies. As our dataset matures we are approaching a regime where a detection is expected, and therefore testing current data analysis tools is crucial, as is the development of new tools and techniques.


Here we provide the second IPTA Mock Data Challenge. The purpose of this challenge is to foster the development of detection tools for pulsar timing arrays and to cultivate interaction with the international gravitational-wave community. For more information about the IPTA mock datasets see the MDC2 website or the announcement on the ArXiv.

Participating in the MDC

In order to participate in an MDC just fork this repository. You will need to register with GitHub in order to do so. From your forked version of the repository you can clone or download the repo using the Clone or download button. Please fork the repo, rather than just downloading directly from this version. This will allow us to keep track of how many people are interested in the MDC. Also, send an email to the Data Challenge Working Group Co-Chair, Jeffrey S. Hazboun. This will allow us to update participants about new MDC releases and the currently open datasets. This README will record any announcements made about ongoing MDCs.

Summary of MDC2 Datasets

In the following table we summarize the contents of the 6 different datasets included in MDC2. Details about the noise parameters injected into Group 1 can be found along with the datasets in the group1 directory in the form of a json file.

Group.Dataset Time Span Freq (MHz) Observing Cadence Noise Signals
g1.d1a(b) 15 yrs 1440 30 days WN SB
g1.d2a(b) 15 yrs 1440 30 days WN,RN SB
g1.d3a(b) 15 yrs 1440 30 days WN SB+SS
g2.d1 15 yrs 800, 1440 30 days - -
g2.d2 15 yrs 800, 1440 30 days - -
g2.d3 15 yrs 800, 1440 30 days - -

WN=white noise, RN=red noise, SB=stochastic background, SS=single source

The datasets included in MDC2 include parameter files containing the timing model parameters for a pulsar, and timing files containing the integrated times-of-arrival or TOAs of pulsar pulses. These files are referred to as par and tim files, respectively.

Data Set Production Notes

  • These data sets were made using IPTA DR2 files as starting place and refit after signals were added. They no longer represent real astrophysical objects, and can only be used for data analysis testing.
  • Signals were simulated using libstempo, a Python wrapper for TEMPO2, but with its own fitting function.
  • Signals were then fit with TEMPO2.
  • Those pulsars with significant red noise contain TEMPO2 red noise parameters, TNRedAmp, TNRedGam and TNRedC. Only a few of the pulsars contain these parameters.

The TOAs in the tim file contain Doppler shifts from the motion of the Earth in its orbit, relative proper motion between the solar system and the pulsar system and any orbital motion of the pulsar, if in a binary system. The par files contain the relevant parameters needed to construct such a model to remove these effects, but its construction is left to the participant. In addition, various clock corrections, observatory location coordinates and other observation dependent effects must be taken into account to build an adequate timing model for a given pulsar. There is a rich history in the pulsar timing community of scientist-written and maintained code bases useful for timing pulsars and, more recently, for doing gravitational wave analyses. While the participant is free to write their own code, there are three publicly available sets of software for modeling pulsar TOAs, TEMPO, TEMPO2 and PINT. While new software development is welcome, the participant is encouraged to use one of these existing packages for building a pulsar timing model, and to focus their time on the data analysis algorithm development.

Use of Existing Gravitational Wave Analysis packages

The existence of a number of well developed PTA gravitational wave data analysis packages needs to be recognized and commented on here. Among the many reasons for holding a MDC, two are relevant when discussing these previous code bases; to encourage new researchers to become involved with PTA data analysis and to foster the cross pollination of analysis techniques. In order to facilitate the first of these, the development of an entirely independent code for PTA analysis is not required for participation in the MDC, though obviously welcome and encouraged.

Besides the existing analysis codes (TempoNest, 42, PAL2, piccard and NX01...) there also exists a code built as a framework for constructing a PTA analysis developed by the NANOGrav collaboration, called enterprise.

Since these tools are accessible widely, we do not prohibit their use in the MDC, but will categorize them differently when discussing them in any manuscript that summarizes the MDC. Entries based on an existing code base (unless submitted by the developer of that code) must add some significant new ability to the base code in order to warrant their submission, and, of course, cite the original developer. An example of such an addition could be a new way of modeling noise in a pulsar, or using the output from the original code to calculate a different statistic.

If you have any questions please contact Jeffrey S. Hazboun.