Ordinal Partition Networks - Extracting dynamics from time series data

Over the last 8 years, several techniques have been developed to extract signatures of a deterministic dynamical system from time series dynamics, via a network construction. However, most of these rely on delay reconstruction via Takens Theorem and they do not directly encode temporal information. I will describe a coarse grained partitioning of the underlying state space that can be achieved without an explicit embedding and which directly encodes the underlying dynamics. Using this we are able to construct networks that encode invariants of the underlying dynamical system and allow both characterisation and diagnosis of experimental time series data.