Theoretical simulations of environment-sensitive dynamical systems for advanced reservoir computing applications

Licentiatavhandling, 2018

The possibility of building intelligent sensing substrates that both collect information about an environment and analyze it in real-time has been investigated theoretically. In a typical setup, a dynamical system is assumed to interact with the environment over time. The system operates as a reservoir computer acting as a reservoir of states. Due to the reservoir-environment interaction, the information about the environment is encoded in the state of the reservoir. The information stored in the system can be inferred (decoded) by analyzing the reservoir state, which is done by observing how a system responds to an external stimulus being an external drive signal. This signal is optimized to ensure that under different environmental conditions the reservoir visits distinct regions of the configuration space. If such a behavior is possible, then a relatively simple readout layer can be used to achieve efficient sensing. These ideas have been examined theoretically by simulating various networks of environment-sensitive elements: the memristor, the capacitor, the constant phase element, and the organic field effect transistor element. It was found that heterogeneity of the network is important for sensing. The simulations were done in the context of ion sensing, which is an extremely complex, many-body, and multi-scale modeling problem. A generic electrical circuit simulator has been developed with a focus on understanding transient dynamics. The constant phase element has been identified as an important primitive that is essential for modeling the experimental data. A new algorithm has been develop to model its transient behavior. Likewise, the same was done for the organic electrochemical transistor. To quantify the sensing capacity of an environment sensitive network a precise mathematical measure has been introduced, the state separability index, and evaluated in numerical experiments. The theoretical work has been supported by the related set of experiments.