Ginestra Bianconi
Cascades in single and multilayer networks
Percolation theory plays a pivotal role in network science as it sheads light on the fundamental structural properties of a network that determine its robustness when a fraction of node is initially damaged.
Despite the fact that the percolation transition is second order, cascade of failure events that abruptly dismantle a network are actually occurring in real systems, with major examples ranging from large electric blackouts to the sudden collapse of ecological systems.
Here we build a large deviation theory of percolation characterizing the response of a sparse network to rare events. This general theory includes the second order phase transition observed typically for random configurations of the initial damage but reveals also discontinuous transitions corresponding to rare configurations of the initial damage for which the size of the giant component is suppressed
Moreover we will discuss recent results regarding cascade of failure on multilayer networks and theoretical principles to design multilayer networks which boost their robustness.
Wei Koong Chai
Networks in Human Engineered Infrastructures
Abstract: Networks appear in various contexts and systems ranging from natural (e.g., hub protein, human brain structure, social networks) to manmade networks. This talk provides a sample collection of my work employing tools and concepts of network science in human engineered infrastructures. The talk covers investigation on:
(1) the robustness of smart power grid against cascading failures whereby there exist inter-dependency between the grid and the communication networks,
(2) caching problems in computer networks and the Internet whereby network nodes are equipped with caching capabilities to enhance information or content delivery, and
(3) spreading phenomenon in transportation systems whereby novel epidemic model is used to understand spreading process induced by agent mobility.
.Jitendra Agarwal
Network communities and infrastructure criticalit
Contributors: Jitendra Agarwal (speaker) and Giulio Galva
Infrastructure networks evolve according to societal needs. Clusters of elements are found in regions where the demand for service is higher, while the system is sparser in areas of lower activity. These clusters are defined network communities, in analogy with human communities. Several community detection algorithms exist and we use stability optimisation to obtain information at different levels of description. Communities are functional subsystems within the network, but the elements critical to their performance are not always detected by global analyses. To overcome this limitation, we use different community measures to analyse and reconcile information obtained at different scales of analysis. A model of the UK railway network is used as a case study. The results obtained from the examination of its community structure are used in conjunction with system-wide metrics to produce a richer picture of its criticality
Andrea Santoro
Pareto optimality in multilayer network growth
Multiplex networks provide a natural framework for analysing transportation systems. Within this context, nodes represent locations of interest, edges stand for connections between two locations, and each layer accounts for a different type of connections among the same set of nodes. In this work we propose a minimal, zero-parameter model of multi-layer network growth which incorporates the concurrent tendency of carriers to maximise their profit and to minimise competition with other carriers. The creation of edges is subject to the local optimisation of a multi-objective cost function, representing a trade-off between efficiency and competition. Interestingly, the proposed model is able to accurately reproduce the micro-, meso-, and macro-scale structure of the six continental airline transportation networks, and provides a simple, reasonable explanation for the observed high heterogeneity in the size of airports and in the distribution of routes across the globe.
The analysis of the feasible points in the resulting efficiency-competition plane reveals that the airlines whose route network is closer to the corresponding Pareto front are indeed those which attain higher per-route revenues. These results shed light on the fundamental role played by multiplexity and multi-objective optimisation principles in shaping the structure of large-scale transportation systems, and provide new insights on potential strategies for individual airlines to increase their revenues by a clever selection of new routes.
Eiko Yoneki
Efficient Large-Scale Graph Processing
The emergence of big data requires fundamental new methodology for data analysis, processing, and information extraction. The main challenge here is to perform efficient and robust data processing, while adapting to the underlying resource availability in a dynamic, large-scale computing environment. Do we really need high performance computers or a large cluster computing? I would introduce our recent work on the graph processing in a commodity single computer, which requires secondary storage as external memory. Executing algorithms results in access to such secondary storage and performance of I/O takes an important role, regardless of the algorithmic complexity or runtime efficiency of the actual algorithm in use.