Bulletin of the American Physical Society
APS March Meeting 2014
Volume 59, Number 1
Monday–Friday, March 3–7, 2014; Denver, Colorado
Session Z17: Complex Networks and their Applications II |
Hide Abstracts |
Sponsoring Units: GSNP Room: 402 |
Friday, March 7, 2014 11:15AM - 11:27AM |
Z17.00001: Task-Based Cohesive Evolution of Dynamic Brain Networks Elizabeth Davison Applications of graph theory to neuroscience have resulted in significant progress towards a mechanistic understanding of the brain. Functional network representation of the brain has linked efficient network structure to psychometric intelligence and altered configurations with disease. Dynamic graphs provide us with tools to further study integral properties of the brain; specifically, the mathematical convention of hyperedges has allowed us to study the brain's cross-linked structure. Hyperedges capture the changes in network structure by identifying groups of brain regions with correlation patterns that change cohesively through time. We performed a hyperedge analysis on functional MRI data from 86 subjects and explored the cohesive evolution properties of their functional brain networks as they performed a series of tasks. Our results establish the hypergraph as a useful measure in understanding functional brain dynamics over tasks and reveal characteristic differences in the co-evolution structure of task-specific networks. [Preview Abstract] |
Friday, March 7, 2014 11:27AM - 11:39AM |
Z17.00002: Evaluation of the satellite power system survivability with the selfish algorithm. Svetlana V. Poroseva, Jacob Lowe, Bryan E. Kaiser Engineering networks (electric power, gas, water, transportation systems, etc.) are traditionally designed for normal operating conditions. Reliability analysis provides tools for describing the network's performance under such conditions. In the modern society, the likelihood of adverse conditions has dramatically increased along with the scale and cost of the network's failure. For some networks such as, for example, those in spacecrafts or in military applications, adverse conditions are normal. Due to differences in the mathematical formulation, reliability analysis is not applicable to networks under adverse conditions. Instead, survivability analysis should be applied. Survivability analysis due to the network's topology is the emerging discipline. In Poroseva et al, IEEE ESTS, 2005, a network survivability analysis based on a probabilistic approach was proposed for networks with heterogeneous nodes. Later, the selfish algorithm was developed to evaluate the topological survivability of networks. The approach is applicable to networks with multiple sources and sinks. The application of this algorithm to a satellite electric power subsystem will be demonstrated at the conference. [Preview Abstract] |
Friday, March 7, 2014 11:39AM - 11:51AM |
Z17.00003: A Simulation of Cooperation and Competition in Insurgent Networks Michael Gabbay Insurgencies are often characterized by multiple groups who share a common foe in the national government but have independent organizations which may differ with respect to social identities, ideologies, strategies, and their use of violence. These groups may cooperate in various ways such as conducting joint attacks, pooling resources, and establishing formal alliances or mergers. However, they may also compete with each other over popular support, recruitment of fighters, funding, allies, and ultimately military dominance. A network coevolution model of insurgent factional dynamics is presented which accounts for factors driving cooperation and competition. The model is formulated as a system of coupled ODEs which evolves network ties between insurgent groups along with group policies concerning the targets of violence. Simulation results are presented showing sharp transitions in network structure as model parameters are varied. Connections are drawn between the model results and empirical data from the Iraqi insurgency. [Preview Abstract] |
Friday, March 7, 2014 11:51AM - 12:03PM |
Z17.00004: Scientific impact: the story of your big hit Roberta Sinatra, Dashun Wang, Pierre Deville, Chaoming Song, Albert-Laszlo Barabasi A gradual increase in performance through learning and practice characterize most trades, from sport to music or engineering, and common sense suggests this to be true in science as well. This prompts us to ask: what are the precise patterns that lead to scientific excellence? Does performance indeed improve throughout a scientific career? Are there quantifiable signs of an impending scientific hit? Using citation-based measures as a proxy of impact, we show that (i) major discoveries are not preceded by works of increasing impact, nor are followed by work of higher impact, (ii) the precise time ranking of the highest impact work in a scientist's career is uniformly random, with the higher probability to have a major discovery in the middle of scientific careers being due only to changes in productivity, (iii) there is a strong correlation between the highest impact work and average impact of a scientist's work. These findings suggest that the impact of a paper is drawn randomly from an impact distribution that is unique for each scientist. We present a model which allows to reconstruct the individual impact distribution, making possible to create synthetic careers that exhibit the same properties of the real data and to define a ranking based on the overall impact of a scientist. [Preview Abstract] |
Friday, March 7, 2014 12:03PM - 12:15PM |
Z17.00005: Global Analysis of Food and Nutrition: What the Human Body Wants Seunghyeon Kim, Mathias Foo, Jaeyun Sung, Yong-Su Jin, Pan-Jun Kim There is currently an abundance of quantitative information regarding foods we consume, such as their total nutrient composition and daily nutritional requirements. In this study, we systematically analyzed such large-scale data of foods to better understand how the composition of foods affects their overall nutritional value. Herein, we constructed two types of networks that reflect nutritional data from about 700 food products: 1) The Food-food Network, in which each edge connects a pair of foods having similar nutritional contents; and 2) the Nutrient-nutrient Network, which is based on co-occurrence patterns of different nutrients across foods. By adopting the insight we obtained from the topological properties of these networks, we present a novel measure to quantify the overall nutritional value of a food, which we call the Nutritional Fitness (NF). Some nutrients can hinder foods from having high NF, acting as ``nutritional bottlenecks.'' Interestingly, a food's NF is not only affected by individual nutrients, but also pairs of nutrients. To this effect, foods with very high NFs tend to have unique nutrient pairs not observed from the majority of foods. To summarize, our study provides insight into how NF and nutrients are intricately related. [Preview Abstract] |
Friday, March 7, 2014 12:15PM - 12:27PM |
Z17.00006: Distinguishing fiction from non-fiction with complex networks David M. Larue, Lincoln D. Carr, Linnea K. Jones, Joe T. Stevanak Complex Network Measures are applied to networks constructed from texts in English to demonstrate an initial viability in textual analysis. Texts from novels and short stories obtained from Project Gutenberg and news stories obtained from NPR are selected. Unique word stems in a text are used as nodes in an associated unweighted undirected network, with edges connecting words occurring within a certain number of words somewhere in the text. Various combinations of complex network measures are computed for each text's network. Fisher's Linear Discriminant analysis is used to build a parameter optimizing the ability to separate the texts according to their genre. Success rates in the 70\% range for correctly distinguishing fiction from non-fiction were obtained using edges defined as within four words, using 400 word samples from 400 texts from each of the two genres with some combinations of measures such as the power-law exponents of degree distributions and clustering coefficients. [Preview Abstract] |
Friday, March 7, 2014 12:27PM - 12:39PM |
Z17.00007: ABSTRACT WITHDRAWN |
Friday, March 7, 2014 12:39PM - 12:51PM |
Z17.00008: Community Structure of a Bank-Firm Credit Network in Japan Hiroshi Iyetomi, Yuki Matsuura We study temporal change of community structure in a Japanese credit network formed by banks and listed firms through their financial relations over the last 30 years. The credit connectedness is regarded as a potenital source of systemic risk. Our network is a bipartite graph consisting of two species of nodes connected with bidirectional links. The direction of links is identified with that of risk flows and their weights are relative credit/loan with respect to the targets. In a partial credit network obtained only with the links pointing from firms toward banks, the city banks forms one major community in most of the time period to share risk when firms go wrong. On the other hand, a partial network only with the links from banks toward firms is decomposed into communities of similar size each of which has its own city bank, reflecting the main-bank system in Japan. Finally we take overlapping parts of the two community sets to find cores of the risk concentration in the credit network. [Preview Abstract] |
Friday, March 7, 2014 12:51PM - 1:03PM |
Z17.00009: Complex dynamics and scale invariance of one-dimensional memristive networks Yuriy Pershin, Valeriy Slipko, Massimiliano Di Ventra There is currently a great interest in resistive systems with memory, also called as ``memristive systems'', and their potential applications [1,2]. In this talk we show that even the simplest one-dimensional network formed by the most common memristive elements with voltage threshold bears nontrivial physical properties [3]. In particular, by taking into account the single element variability we find (1) dynamical acceleration and slowing down of the total resistance in adiabatic processes, (2) dependence of the final state on the history of the input signal with same initial conditions, (3) existence of switching avalanches in memristive ladders, and (4) independence of the dynamics voltage threshold with respect to the number of memristive elements in the network (scale invariance). An important criterion for this scale invariance is the presence of memristive systems with very small threshold voltages in the ensemble. [1] Y. V. Pershin and M. Di Ventra, Advances in Physics 60, 145-227 (2011). [2] M. Di Ventra and Y. V. Pershin, Nature Physics 9, 200 (2013). [3] Y. V. Pershin, V. A. Slipko, and M. Di Ventra, Phys. Rev. E 87, 022116 (2013). [Preview Abstract] |
Friday, March 7, 2014 1:03PM - 1:15PM |
Z17.00010: Scaling of Various Dominating Sets in Scale-Free and Empirical Networks N. Derzsy, F. Molnar Jr., E. Czabarka, L. Szekely, B. Szymanski, G. Korniss We develop ensemble-based graph theoretical methods to approximate the size of minimum dominating sets (MDS) in scale-free networks. The MDS is found by a sequential greedy algorithm and the scale-free network samples are generated using the configuration model. Depending on the considered maximum degree cutoff, we analyze two subtypes of scale-free networks. We study the upper bound of random dominating sets and find that the numerical bound is lower than the analytical one. We propose a degree-based probabilistic selection that for a limiting case provides the smallest probabilistic dominating set. The method relies on a degree cutoff parameter and nodes having degrees above this cutoff are added to the dominating set (CDS). Our results reveal that with an optimal degree cutoff the CDS size is very close to the MDS. We provide analytical estimates for the uncorrelated version of scale-free networks to support our conjecture. We propose an efficient method to control the assortativity (measured by Spearman's rho) in networks. Applying this technique for the construction of our scale-free network ensembles, we provide a comprehensive analysis on the behavior of the probabilistic in comparison with the greedily selected dominating sets with respect to multiple network features. [Preview Abstract] |
Friday, March 7, 2014 1:15PM - 1:27PM |
Z17.00011: Geometrical structure of Neural Networks: Geodesics, Jeffrey's Prior and Hyper-ribbons Lorien Hayden, Alex Alemi, James Sethna Neural networks are learning algorithms which are employed in a host of Machine Learning problems including speech recognition, object classification and data mining. In practice, neural networks learn a low dimensional representation of high dimensional data and define a model manifold which is an embedding of this low dimensional structure in the higher dimensional space. In this work, we explore the geometrical structure of a neural network model manifold. A Stacked Denoising Autoencoder and a Deep Belief Network are trained on handwritten digits from the MNIST database. Construction of geodesics along the surface and of slices taken from the high dimensional manifolds reveal a hierarchy of widths corresponding to a hyper-ribbon structure. This property indicates that neural networks fall into the class of sloppy models, in which certain parameter combinations dominate the behavior. Employing this information could prove valuable in designing both neural network architectures and training algorithms. [Preview Abstract] |
Friday, March 7, 2014 1:27PM - 1:39PM |
Z17.00012: Modeling criticality in networks of neurons Shane Squires, Andrew Pomerance, Edward Ott, Michelle Girvan A recent series of experiments have suggested that networks of biological neurons operate near a critical point, separating two phases in which firing activity either decays or grows exponentially. In this talk, we propose and analyze a simple model of this behavior. Neurons may be connected via arbitrary networks of activating and inhibiting links, and they fire when their membrane voltage exceeds a threshold value. The main advantage of our model is that we can analyze the effects of network structure on the criticality of the system, while preserving realistic features of neurons, such as threshold-based firing behavior. At the critical point, we reproduce the empirically measured critical exponents for firing avalanches, and we discuss how changes to the network affect the tuning parameter for the phase transition. [Preview Abstract] |
Friday, March 7, 2014 1:39PM - 1:51PM |
Z17.00013: Sampling networks with prescribed degree correlations Charo Del Genio, Kevin Bassler, P\'eter Erd\H{o}s, Istv\'an Miklos, Zolt\'an Toroczkai A feature of a network known to affect its structural and dynamical properties is the presence of correlations amongst the node degrees. Degree correlations are a measure of how much the connectivity of a node influences the connectivity of its neighbours, and they are fundamental in the study of processes such as the spreading of information or epidemics, the cascading failures of damaged systems and the evolution of social relations. We introduce a method, based on novel mathematical results, that allows the exact sampling of networks where the number of connections between nodes of any given connectivity is specified. Our algorithm provides a weight associated to each sample, thereby allowing network observables to be measured according to any desired distribution, and it is guaranteed to always terminate successfully in polynomial time. Thus, our new approach provides a preferred tool for scientists to model complex systems of current relevance, and enables researchers to precisely study correlated networks with broad societal importance. [Preview Abstract] |
Friday, March 7, 2014 1:51PM - 2:03PM |
Z17.00014: Percolations on hypergraphs Bruno Coelho Coutinho, Y.-Y. Liu, H.-J. Zhou We analytically study the emergence of the giant component, two-core and core in uniform and non-uniform hypergraphs. We show that depending on the leaf definition and in the hypergraph rank distribution the 2-core can emerge as a hybrid phase transition our as a continuous phase transition and we provide a analytical condition for the existence of the hybrid phase transition. We found that in hyperpgrahs there are two meaningful versions of the greedy leaf removal (GLR), associated with two different leaves and intimately related with the vertex and edge cover problem. We study the emergence of the core for both cases, and we show that both of the cores emerge as a continuous phase transition for the considered distribution. [Preview Abstract] |
Friday, March 7, 2014 2:03PM - 2:15PM |
Z17.00015: Quantum Phase Transitions: A Network Approach David L. Vargas, David M. Larue, Lincoln D. Carr Understanding the network structure of complex systems has opened up new avenues of research in sociology, biology, technology, and physics. In this talk we present evidence that complex network measures are able to identify the phases in two well known models. We distinguish the ferromagnetic and paramagnetic phases of the transverse Ising Hamiltonian. We also identify the Mott-insulator to superfluid transition of the Bose-Hubbard Hamiltonian. The network approach to the analysis of quantum phase transitions provides us with a new set of tools to explore the many body physics of quantum phase transitions. [Preview Abstract] |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700