Permutation entropy with vector embedding delays

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Permutation entropy (PE) is a statistic used widely for the detection of structure within a time series. Embedding delay times at which the PE is reduced are characteristic timescales for which such structure exists. Here, a generalized scheme is investigated where embedding delays are represented by vectors rather than scalars, permitting PE to be calculated over a (D-1)-dimensional space, where D is the embedding dimension. This scheme is applied to numerically generated noise, sine wave and logistic map series, and experimental data sets taken from a vertical-cavity surface emitting laser exhibiting temporally localized pulse structures within the round-trip time of the laser cavity. Results are visualized as PE maps as a function of embedding delay, with low PE values indicating combinations of embedding delays where correlation structure is present. It is demonstrated that vector embedding delays enable identification of structure that is ambiguous or masked, when the embedding delay is constrained to scalar form.

Original languageEnglish
Article number062205
Pages (from-to)1-8
Number of pages8
JournalPhysical Review E
Volume96
Issue number6
DOIs
Publication statusPublished - 12 Dec 2017

Fingerprint Dive into the research topics of 'Permutation entropy with vector embedding delays'. Together they form a unique fingerprint.

Cite this