This is Rutger van Haasteren's Bayesian-inference pipeline for Pulsar Timing Array (PTA) data. The code interacts with Tempo2 through libstempo
The code is use mainly for single-pulsar analysis, and gravitational-wave detection purposes of full Pulsar Timing Array datasets. The modelling of the data can include:
- Error bars and EFACs: multiple per pulsar
- White noise: EQUAD/jitter noise
- Red noise: per frequency, or modelled spectrum
- DM variations: per frequency, or modelled spectrum
- Correlated signals: per frequency, or modelled spectrum, with correlations:
- Uniform: clock-error
- Dipolar: ephemeris errors
- Quadrupolar: isotropic stochastic background GR correlations
- Anisotropic: GR correlated, expanded in spherical harmonics
Many types of samplers are included:
- Metropolis-Hastings
- DNest
- twalk
- emcee (2012).
- pyMultiNest (2013)
- Reversible-Jump MCMC, based on emcee's Metropolis
For common-mode mitigation, the signals can be reconstructed mitigating arbitrary signals simultaneously. Especially handy for getting rid of DM variations. Expanded on the techniques of Lee et al. (in prep.), and similar to what is included in TempoNest (2013).
A alpha-phase interface for the generation of mock data is provided. Needs to be expanded still.
Under construction
- Python 2.7
- numpy
- scipy
- matplotlib, for plotting only
- tempo2
- libstempo
- emcee (optional)
- pyMultiNest (optional)