Delta-Chi2 Model Comparison¶
What it is¶
Several PQC event detectors (transient, step, DM-step, dip, bump, glitch) accept candidates using improvement in weighted fit:
Why PQC uses it¶
It provides a simple, fast statistic for ranking candidate event templates and applying practical acceptance thresholds.
Weighted least-squares context¶
With residuals \(y_i\), uncertainties \(\sigma_i\), and model \(m_i\), weights are \(w_i=1/\sigma_i^2\):
Interpretation¶
larger \(\Delta\chi^2\) means stronger evidence for event model
threshold choice controls detector aggressiveness
Assumptions and caveats¶
assumes sigma values are meaningful relative weights
not a full Bayesian model comparison
scanning many candidate epochs/windows introduces a look-elsewhere effect, so absolute statistical significance is approximate
Small worked example¶
If \(\chi^2_{\mathrm{null}}=120\) and \(\chi^2_{\mathrm{model}}=92\),
then \(\Delta\chi^2=28\).
If detector threshold is 25, this candidate is accepted.
References¶
Edwards, R. T., Hobbs, G. B., & Manchester, R. N. (2006). “tempo2, a new pulsar timing package - II. The timing model and precision estimates.” MNRAS, 372(4), 1549-1574.
Lorimer, D. R., & Kramer, M. (2005). Handbook of Pulsar Astronomy. Cambridge University Press.