Systemic Performance Measures from Distributional Zeta-Function

TL;DR

This paper introduces a novel approach using the Distributional Zeta-Function to develop new systemic performance measures for network synthesis, linking spectral properties to performance enhancement and exploring physical implications.

Contribution

It proposes the use of the Distributional Zeta-Function for constructing new systemic performance measures and relates network topology to spectral properties for performance optimization.

Findings

Few operations needed for performance enhancement.

Optimal link placement improves network performance.

Advantages demonstrated in network synthesis applications.

Abstract

We propose the use of the Distributional Zeta-Function (DZF) for constructing a new set of Systemic Performance Measures (SPM). SPM have been proposed to investigate network synthesis problems such as the growing of linear consensus networks. The adoption of the DZF has shown interesting physical consequences that in the usual replica method are still unclarified, i.e., the connection between the spontaneous symmetry breaking mechanism and the structure of the replica space in the disordered model. We relate topology of the network and the partition function present in the DZF by using the spectral and the Hamiltonian structure of the system. The studied objects are the generalized partition funcion, the DZF, the Expected value of the replica partition function, and the quenched free energy of a field network. We show that with these objects we need few operations to increase the percentage of performance enhancement of a network. Furthermore, we evalue the location of the optimal added links for each new SPM and calculate the performance improvement of the new network for each new SPM via the spectral zeta function, $\mathcal{H}_{2}$-norm, and the communicability between nodes. We present the advantages of this new set of SPM in the network synthesis and we propose other methods for using the DZF to explore some issues such as disorder, critical phenomena, finite-temperature, and finite-size effects on networks. Relevance of the results are discussed.