The information flow through a combinatorial threshold linear network of polymer-networked engineered nanoparticle composites has been found to exhibit primitive neuromorphic computing behavior. Our systematic analysis of a 4-node, 1-sink network reveals specific conditions for the emergence of Limit Cycles (LCs), which could serve as a mechanism for information storage in affine networks. By examining various input profiles, we establish quantitative relationships between input parameters and the resulting LC characteristics. We demonstrate that peak amplitudes and frequencies of these oscillatory attractors can function as system outputs within specific input regimes, enabling predictable system input-output relationships for computational operations. We perturb the network to assess system robustness by introducing additional structural sinks (creating 5-node and 6-node structures). We find that certain network architectures maintain stable LC behavior despite structural modifications, suggesting the potential for scalability in more complex implementations.
Harazinska et al. (Mon,) studied this question.