Summary
International Symposium on Nonlinear Theory and Its Applications
2015
Session Number:A2L-A
Session:
Number:A2L-A-3
Barycentric Coordinates Revisited: Relaxation with Linear Programming and its Evaluations
Yoshito Hirata, Hideyuki Suzuki, Kazuyuki Aihara,
pp.10-13
Publication Date:2015/12/1
Online ISSN:2188-5079
Summary:
Barycentric coordinates are first used in Mees, Int. J. Bifurcat. Chaos (1991) to model a short nonlinear time series faithfully, while his formulation is restricted to low-dimensional dynamics because it employs triangulation. We recently relaxed his formulation by using linear programming (Hirata et al., Chaos (2015)). Using this relaxation, we can generate prediction and freeruns from a high dimensional time series. In this talk, we will review these recent advancements on barycentric coordinates and discuss some indices that evaluate locally the modelling accuracy for barycentric coordinates.