Issue |
EPL
Volume 114, Number 4, May 2016
|
|
---|---|---|
Article Number | 41001 | |
Number of page(s) | 6 | |
Section | The Physics of Elementary Particles and Fields | |
DOI | https://doi.org/10.1209/0295-5075/114/41001 | |
Published online | 15 June 2016 |
Edge detecting new physics the Voronoi way
1 Physics Department, University of Florida - Gainesville, FL 32611, USA
2 Department of Physics and Astronomy, University of Hawaii - Honolulu, HI 96822, USA
Received: 22 February 2016
Accepted: 25 May 2016
Edge detection is an important tool in the search for and exploration of physics beyond the standard model. Ideally one would be able to perform edge detection in a relatively model-independent way, however most analyses rely on more detailed properties (i.e. “shapes” or likelihood distributions) of the variable(s) of interest. We therefore present a sketch of how edge detection can be accomplished using Voronoi tessellations, focusing on the case of two-dimensional distributions for simplicity. After deriving some useful properties of the Voronoi tessellations of simplified distributions containing edges, we propose several algorithms for tagging the Voronoi cells in the vicinity of kinematic edges in real data and show that the efficiency of our methods is improved by the addition of a few Voronoi relaxation steps via Lloyd's method. Our results suggest specifically that Voronoi-based methods should be useful for relatively model-independent edge detection, and, more generally, that the wider adaptation of Voronoi tessellations may be useful in collider physics.
PACS: 14.80.Ly – Supersymmetric partners of known particles / 07.05.Kf – Data analysis: algorithms and implementation; data management / 12.60.-i – Models beyond the standard model
© EPLA, 2016
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.