PIRD – Partial Information Rate Decomposition

A Framework for the application of the principles of Partial Information Decomposition to random processes with temporal statistical structure



Partial Information Decomposition (PID) is a principled and flexible method to unveil complex high-order interactions in multi-unit network systems. Though being defined exclusively for random variables, PID is ubiquitously applied to multivariate time series taken as realizations of random processes with temporal statistical structure. To overcome the incorrect depiction of high-order effects by PID schemes applied to dynamic networks, we introduce the framework of Partial Information Rate Decomposition (PIRD). PIRD is formalized applying lattice theory to decompose the information shared dynamically between a target random process and a set of source processes, and implemented for Gaussian processes through a spectral expansion of information rates. It is then demonstrated in practice analyzing time series from large-scale climate oscillations [Faes et al., 2025] as well as  from a physiological network comprising cerebrovascular and cardiovascular variables [Sparacino et al., 2025].

OIRsimu_img

Toolbox description:
The toolbox realizes PIRD for a network of random processes mapped by multivariate time series, for the case of two target process and two or three source processes. The practical implementation is based on fitting a vector autoregressive (VAR) model on the set of (three or four) analyzed time series, and then computing spectral measures of mutual information rate (MIR) between the target and (groups of) sources from the frequency domain representation of the VAR parameters; the spectral MIR functions are then used to compute spectral redundancy rate, from which the PID principle is applied in both frequency and time domains to identify unique, redundant and synergistic rates of information shared between target and sources. In the complete implementation [Sparacino et al., 2025] (folder \full\), the VAR analysis is performed (both for the cases of two and three sources) directly in the frequency domain to obtain information decomposition at each specific frequency. In the implementation used for the letter [Faes et al., 2025] (folder \statespace\), VAR analysis is performed (only for the case of two sources) using state space models for the spectral representation and the spectral redundancy rate is integrated to get time-domain measures, so that the PIRD is executed in the time-domain only. The two implementations are compared on representative climate data in the script climate_data.mat.

The “Full” framework is demonstrated in the scripts simu_freqPIRD.m and appl_climate_data.m to demonstrate the framework in simulation 3 and in the climate data application of [Sparacino et al., 2025]

The “State space” framework is demonstrated in the scripts simuLETTER1.m and simuLETTER2.m to demonstrate the framework in the simulations of [Faes et al., 2025]


DOWNLOAD: Zip file with all scripts and functions: PIRD.zip

The code is provided free of charge. It is neither exhaustively tested nor particularly well documented. The authors accept no liability for its use. Use, modification and redistribution of the code is allowed in any way users see fit. Authors ask only that authorship is acknowledged and ref. [1] is cited upon utilization of the code in integral or partial form.

DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITY The code is supplied as is and all use is at your own risk. The authors disclaim all warranties of any kind, either express or implied, as to the softwares, including, but not limited to, implied warranties of fitness for a particular purpose, merchantability or non - infringement of proprietary rights. Neither this agreement nor any documentation furnished under it is intended to express or imply any warranty that the operation of the software will be error - free. Under no circumstances shall the authors of the softwares provided here be liable to any user for direct, indirect, incidental, consequential, special, or exemplary damages, arising from the software, or user' s use or misuse of the softwares. Such limitation of liability shall apply whether the damages arise from the use or misuse of the data provided or errors of the software.

References:

[Faes et al., 2025] - L Faes, L Sparacino, G Mijatovic, Y Antonacci, L Ricci, D Marinazzo, S Stramaglia, 'Partial Information Rate Decomposition', arXiV preprint, 2025; DOI:

[Sparacino et al., 2025] - L Sparacino, G Mijatovic, Y Antonacci, L Ricci, D Marinazzo, S Stramaglia, L Faes, 'Decomposing Multivariate Information Rates in Networks of Random Processes', arXiV preprint, 2025; DOI: