Having confidence in your confidence: parameter estimation post-processing

Document #:

LIGO-P1400054-v2

Document type:

P - Publications

Abstract:

Stochastic samplers are often used to perform Bayesian inference on data sets. It is often assumed that once the samples are produced then this is the end of the data processing and the treatment of the samples will not greatly affect the results. In this paper we re-emphasise some of the biases that can enter the extracted information, due to the processing of these samples. We suggest a straightforward method to solve these issues called the 2-stage kD-tree. This method has already been used as a necessary part of testing algorithms that are used to extract gravitational wave signals from ground-based interferometer data.

Files in Document:

• kdtrees.pdf (618.5 kB)

Topics:

• Compact Binaries

Authors:

• Trevor Sidery
• Will Farr
• Jonathan Gair
• Ilya Mandel