Python package for informative/non-informative decomposition of variables in a dynamical system.
The variables in a complex dynamical system are entangled: the evolution of a given variable depends on the others. As consequence, it is possible to gain some understanding about one variable by observing the evolution of other variables in the system. This has led to the development of methods that tries to assess how variables are informative to each other in a dynamical system.
The proposed informative/non-informative decomposition (IND for short) aims as splitting a given source variable into a component that contains all the information to predict the state of a target variable in the future; and a non-informative (residual) contribution that shares no information with the future state of the target variable.
To discern what informative means, IND uses the Shannon mutual-information (Shannon, 1948). The details of the algorithm can be found in Arranz & Lozano-Duran (2024).
The best way of learning how to use IND is to use the notebooks.
Shannon, C.E., 1948. A mathematical theory of communication. The Bell system technical journal, 27(3), pp.379-423.
Arranz, G. and Lozano-Durán, A., 2024. Informative and non-informative decomposition of turbulent flow fields.