Disentangling Glitches

Document #:

LIGO-G1500573-v1

Document type:

G - Presentations (eg Graphics)

Abstract:

A simple physically driven model whereby noise glitches in gravitational wave data are viewed as linear or multi-linear combinations with random amplitudes and delays of a few independent elementary waveforms (representing the transfer functions between un-accessible disturbance entry-points and the data channel) is presented. The number of (independent) elementary consituents is estimated via principal component analysis. A shifted indipendent-component-analysis algorithm is proposed as an effective tool for disentangling observed glitches into their elementary constituents. Numerical experiments are presented, and ensuing perspectives for active glitch subtraction and noise (re)-gaussianization are discussed.

Files in Document:

• Osaka_poster_final_A0_rev.pdf (1.1 MB)

Topics:

• Other data analysis

Authors:

• Innocenzo M Pinto
• Maria Principe

Keywords:

active noise subtraction glitchy noise independent component aanalysis noise statistics,

Associated with Events:

GWPAW 2015 held on 17 Jun 2015 in Osaka, Japan