Kernel density estimations applied to gamma ray light curves
Abstract
Two kernel estimators, the naive estimator (NE) and the Fourier series estimator (FSE), are compared with the histogram estimator (HE) for a wide variety of possible unimodal light curves. Kernel estimators are shown to be objective and asymptotically unbiased, allowing errors to be approximated using the variance component only, and they are found to be smoother and more consistent than the HE. The NE is found to be better than the FSE for very narrow and weak peaks, though the FSE is optimal for broader smoother and stronger peaks.
- Publication:
-
Astronomy and Astrophysics
- Pub Date:
- December 1986
- Bibcode:
- 1986A&A...170..187D
- Keywords:
-
- Computational Astrophysics;
- Gamma Ray Astronomy;
- Histograms;
- Kernel Functions;
- Light Curve;
- Probability Density Functions;
- Data Smoothing;
- Estimators;
- Random Variables;
- Astrophysics