PERIOD - The essential periodicity search software

If you are searching for periodicities in your data, then the new Unix release of the highly respected package PERIOD is just what you need. Many of you will already be familiar with the VMS version that Vik Dhillon (La Palma) wrote and will be aware that it is extremely easy to use on any type of data, be it radial velocity, X-ray, or even gravitational waves. Until now, the only limitation on its use has been that it was not available under Solaris or OSF. This has now changed, and that is not all.

Porting PERIOD to Unix allowed Vik and I to upgrade the code considerably, making it even more robust, accurate, and simple to use. We also grabbed the opportunity to extend the functionality to allow FITS OGIP files to be read as input. This means that users working with data from sources such as ROSAT, EXOSAT, VELA, or ASCA will find life much easier as PERIOD will act directly upon their data without the need for any conversion process. Indeed, for a number of users, PERIOD is now an essential part of their software arsenal.

What does it do?

To whet your appetite, Figure 1 shows some noisy cyclical data and Figure 2 shows a periodogram generated from it by PERIOD. The application successfully identified the 2 periodicities in the data. The time taken to do this was around 2 minutes from sitting down at the terminal. In a recent trial, a completely new user was able to get his first periodogram out within 5 minutes by following the examples given.

For those of you unfamiliar with PERIOD, it employs a simple menu-driven package with help available from the menu at any time. Operations that may be undertaken from the menus include:

Input and output

  • Reading raw data from both ASCII tables and FITS format OGIP files.
  • Storing calculated fits in a log file.
  • Outputting power spectra to ASCII files.
  • Outputting modelled/modified data.
  • Invoking the QDP plotting/fitting package (allows labelling).
Examining the data
  • Generating a power spectrum showing periodicities.
  • Folding the data on a given period.
  • Fitting sine curves to the input data.
  • Setting data points to unity to investigate spectral leakage.
Manipulating the raw data
  • Removing background trends in the data via polynomial fits.
  • Creating test data.
  • Adding noise to data.
  • Adding or subtracting known periods to/from the data.
The bulk of the serious computational work is carried out by a group of routines that employ a variety of sophisticated techniques that search the data for periodicities - even when the data points are unequally spaced. The software has been written so that a user need not have in-depth knowledge of any of the methods employed, but may employ them as black-boxes. These include:
  • Chi-squared analysis of a folded sine fit versus frequency.
  • Cleaned and Discrete Fourier power spectra.
  • Phase dispersion minimization (PDM Stellingwerf).
  • Lomb-Scargle normalized periodograms.
  • String-length (Dworetsky) versus frequency estimates.
  • Calculating periods from the periodogram.
  • Calculating estimate significance.
It is worth remembering that significance estimates for periods derived by most methods are notoriously unreliable, hence special effort has been expended in the current version of PERIOD to ensure that the Fisher Randomisation method (as per Nemec 1985), one of the best regarded, is employed.

Where to find out more

A document ( SUN/167) is available which describes in detail how the applications work and how they may be used. It includes a lengthy technical reference section for those wishing to see inside the black-boxes. An important feature of the document is a simple step-by-step account of a realistic analysis session which will allow readers to get working immediately on their data. A further document (SSN/25), is available for system managers describing, in detail, the simple installation process.

There is a web page for PERIOD at http://www.astro.cf.ac.uk/pub/Grant.Privett/period.html where the latest developments will be announced. If you have any comments or queries about using PERIOD, please do not hesitate to e-mail me. All user input is gratefully received.

Next section, Previous section, Contents
Grant Privett, Starlink, Cardiff, gjp@astro.cf.ac.uk, Sep-96