Dans cette thèse, nous étudions les graphes aléatoires en utilisant des outils de la théorie des matrices aléatoires et l’analyse probabilistique afin de résoudre des problèmes clefs dans le domaine des réseaux complexes et Big Data. Le premier problème qu’on considère est de détecter un sous graphe Erdős–Rényi G(m,p) plante dans un graphe Erdős–Rényi G(n,q). Nous dérivons les distributions d’une statistique basée sur les propriétés spectrales d’une matrice définie du graphe. Ensuite, nous considérons le problème de la récupération des sommets du sous graphe en présence de l’information supplémentaire. Pour cela nous utilisons l’algorithme «Belief Propagation». Le BP sans informations supplémentaires ne réussit à la récupération qu’avec un SNR effectif lambda au-delà d’un seuil. Nous prouvons qu’en présence des informations supplémentaires, ce seuil disparaît et le BP réussi pour n’importe quel lambda. Finalement, nous dérivons des expressions asymptotiques pour PageRank sur une classe de graphes aléatoires non dirigés appelés « fast expanders », en utilisant des techniques théoriques à la matrice aléatoire. Nous montrons que PageRank peut être approché pour les grandes tailles du graphe comme une combinaison convexe du vecteur de dégré normalisé et le vecteur de personnalisation du PageRank, lorsque le vecteur de personnalisation est suffisamment délocalisé. Par la suite, nous caractérisons les formes asymptotiques de PageRank sur le Stochastic Block Model (SBM) et montrons qu’il contient un terme de correction qui est fonction de la structure de la communauté. / In this thesis, we study random graphs using tools from Random Matrix Theory and probability to tackle key problems in complex networks and Big Data. First we study graph anomaly detection. Consider an Erdős-Rényi (ER) graph with edge probability q and size n containing a planted subgraph of size m and probability p. We derive a statistical test based on the eigenvalue and eigenvector properties of a suitably defined matrix to detect the planted subgraph. We analyze the distribution of the derived test statistic using Random Matrix Theoretic techniques. Next, we consider subgraph recovery in this model in the presence of side-information. We analyse the effect of side-information on the detectability threshold of Belief Propagation (BP) applied to the above problem. We show that BP correctly recovers the subgraph even with noisy side-information for any positive value of an effective SNR parameter. This is in contrast to BP without side-information which requires the SNR to be above a certain threshold. Finally, we study the asymptotic behaviour of PageRank on a class of undirected random graphs called fast expanders, using Random Matrix Theoretic techniques. We show that PageRank can be approximated for large graph sizes as a convex combination of the normalized degree vector and the personalization vector of the PageRank, when the personalization vector is sufficiently delocalized. Subsequently, we characterize asymptotic PageRank on Stochastic Block Model (SBM) graphs, and show that it contains a correction term that is a function of the community structure.

In this thesis we consider graph partition of a particular kind of complex networks referred to as power law graphs. In particular, we focus our analysis on the market graph, constructed from time series of price return on the American stock market. Two different methods originating from clustering analysis in social networks and image segmentation are applied to obtain graph partitions and the results are evaluated in terms of the structure and quality of the partition. Along with the market graph, power law graphs from three different theoretical graph models are considered. This study highlights topological features common in many power law graphs as well as their differences and limitations. Our results show that the market graph possess a clear clustered structure only for higher correlation thresholds. By studying the internal structure of the graph clusters we found that they could serve as an alternative to traditional sector classification of the market. Finally, partitions for different time series was considered to study the dynamics and stability in the partition structure. Even though the results from this part were not conclusive we think this could be an interesting topic for future research. / I denna uppsats studeras graf partition av en typ av komplexa nätverk som kallas power law grafer. Specifikt fokuserar vi på marknadengrafen, konstruerad av tidsserier av aktiepriser på den amerikanska aktiemarknaden. Två olika metoder, initialt utvecklade för klusteranalys i sociala nätverk samt för bildanalys appliceras för att få graf-partitioner och resultaten utvärderas utifrån strukturen och kvaliten på partitionen. Utöver marknadsgrafen studeras aven power law grafer från tre olika teoretiska grafmodeller. Denna studie belyser topologiska egenskaper vanligt förekommande i många power law grafer samt modellerns olikheter och begränsningar. Våra resultat visar att marknadsgrafen endast uppvisar en tydlig klustrad struktur för högre korrelation-trösklar. Genom att studera den interna strukturen hos varje kluster fann vi att kluster kan vara ett alternativ till traditionell marknadsindelning med industriella sektorer. Slutligen studerades partitioner för olika tidsserier för att undersöka dynamiken och stabiliteten i partitionsstrukturen. Trots att resultaten från denna del inte var entydiga tror vi att detta kan vara ett intressant spår för framtida studier.

In this thesis, new methods for proving concentration inequalities for Poisson functionals are developed. The focus is on techniques that are based on logarithmic Sobolev inequalities, but also results that are based on the convex distance for Poisson processes are presented. The general methods are applied to a variety of functionals associated with random geometric graphs. In particular, concentration inequalities for subgraph and component counts are proved. Finally, the established concentration results are used to derive strong laws of large numbers for subgraph and component counts associated with random geometric graphs.

No description available.

Spectral graph theory widely increases the interests in not only discovering new properties of well known graphs but also proving the well known properties for the new type of graphs. In fact all spectral properties of proverbial graphs are not acknowledged to us and in other hand due to the structure of nature, new classes of graphs are required to explain the phenomena around us and the spectral properties of these graphs can tell us more about the structure of them. These both themes are the body of our work here. We introduce here three models of random graphs and show that the eigenvalue counting function of Laplacians on these graphs has exponential decay bound. Since our methods heavily depend on the first nonzero eigenvalue of Laplacian, we study also this eigenvalue for the graph in both random and nonrandom cases.

No description available.

[pt] Nesta dissertação, discutiremos o método de equações diferenciais de Wormald, que possui muitas aplicações recentes em Combinatória. Esse método explora a interação entre a matemática discreta e contínua e pode ser usado para provar concentração em uma grande quantidade de processos aleatórios discretos. Em particular, estudaremos o processo livre de H e o algoritmo guloso aleatório para gerar conjuntos independentes em hipergrafos. Esses processos tem sido amplamente estudados nos últimos anos, culminando com o recente grande avanço de Tom Bohman e Patrick Bennett em 2016, que obtiveram uma cota inferior para hipergrafos com certas condições de densidade. Nós não só reproduzimos sua demonstração mas também obtemos um resultado mais forte (expandindo seu resultado para hipergrafos mais esparsos) e analisamos o caso de hipergrafos lineares, com o intuito de progredir rumo a uma conjectura de Johnson e Pinto sobre o processo livre de Q2 no hipercubo Qd. / [en] In this dissertation, we will discuss Wormald s differential equations method, which has recently had many intriguing applications in Combinatorics. This method explores the interplay between discrete and continuous mathematics and it can be used to prove concentration in a number of discrete random processes. In particular, we will discuss the H-free process and the random greedy algorithm to obtain independent sets in hypergraphs. These processes had been extensively studied through the past few years, culminating in the recent breakthrough of Tom Bohman and Patrick Bennett in 2016, who obtained a lower bound for hypergraphs with certain density conditions. We not only reproduce the proof given by them but also obtain a stronger result (expanding their result to sparser hypergraphs) and we analyze the case of linear hypergraphs, in order to make progress towards a conjecture by Johnson and Pinto concerning the Q2-free process in the hypercube Qd.

[pt] O estudo de desvios, e em particular grandes desvios, tem uma história longa na teoria de probabilidade. Nas últimas décadas muitos artigos consideraram essas questões no contexto de subgrafos de grafos aleatórios G(n, p) e G(n, m). Esta dissertação considera a cauda inferior para o número de triângulos no grafo aleatório G(n, m). Duas abordagens estão consideradas: Martingales, a partir artigo de Christina Goldschmidt, Simon Griffiths e Alex Scott; e Teoria Espectral de Grafos, a partir do artigo de Joe Neeman, Charles Radin e Lorenzo Sadun. Essas duas abordagens conseguem encontrar o comportamento da cauda em dois regimes diferentes. Na dissertação discutiremos a visão geral do artigo de Goldschmidt, Griffiths e Scott, e discutiremos em detalhes o artigo de Neeman, Radin e Sadun. Em particular, exploraremos a conexão entre a cauda inferior do número de triângulos e o comportamento dos autovalores mais negativos da matriz de adjacência. Veremos que a contagem tende a depender, essencialmente, do autovalor mais negativo. / [en] The study of deviations, and in particular large deviations, has a long history in Probability Theory. In recent decades many articles have considered these questions in the context of subgraphs of the random graphs G(n, p) and G(n, m). This dissertation considers the lower tail for the number of triangles in the random graph G(n, m). Two approaches are considered: Martingales, based on the article of Christina Goldschmidt, Simon Griffiths and Alex Scott; and Spectral Graph Theory, based on the article of Joe Neeman, Charles Radin and Lorenzo Sadun. These two approaches manage to find the behavior of the tail in two different regimes. In this dissertation we give an overview of the article of Goldschmidt, Griffiths and Scott, discuss in detail the article of artigo Neeman, Radin and Sadun. In particular, we shall explore the connection between the lower tail of the number of triangles and the behavior of the most negative eigenvalues of the adjacency matrix. We shall see that the triangle count tends to especially depend on the most negative eigenvalue.

Dans cette thèse, nous étudions les propriétés de concentration des fonctions propres du laplacien discret sur des graphes réguliers de degré fixé dont le nombre de sommets tend vers l'infini. Cette étude s'inspire de la théorie de l'ergodicité quantique sur les variétés. Par analogie avec cette dernière, nous développons un calcul pseudo-différentiel sur les arbres réguliers : nous définissons des classes de symboles et des opérateurs associés, et nous prouvons un certain nombre de propriétés de ces classes de symboles et opérateurs. Nous montrons notamment que les opérateurs sont bornés dans L², et nous donnons des formules de l'adjoint et du produit. Nous nous servons ensuite de cette théorie pour montrer un théorème d'ergodicité quantique pour des suites de graphes réguliers dont le nombre de sommets tend vers l'infini. Il s'agit d'un résultat de délocalisation de la plupart des fonctions propres dans la limite des grands graphes réguliers. Les graphes vérifient une hypothèse d'expansion et ne comportent pas trop de cycles courts, deux hypothèses vérifiées presque sûrement par des suites de graphes réguliers aléatoires. / N this thesis, we study concentration properties of eigenfunctions of the discrete Laplacian on regular graphs of fixed degree, when the number of vertices tend to infinity. This study is made in analogy with the Quantum Ergodicity theory on manifolds. We construct a pseudo-differential calculus on regular trees by defining symbol classes and associated operators and proving some properties of these classes of symbols and operators. In particular we prove that the operators are bounded on L² and give adjoint and product formulas. We then use this theory to prove a Quantum Ergodicity theorem on large regular graphs. This is a property of delocalization of most eigenfunctions in the large scale limit. We consider expander graphs with few short cycles (for instance random large regular graphs). These hypothesis are almost surely satisfied by sequences of random regular graphs.

Modelos e teorias científicas surgem da necessidade do homem entender melhor o funcionamento do mundo em que vive. Constantemente, novos modelos e técnicas são criados com esse objetivo. Uma dessas teorias recentemente desenvolvida é a da Criticalidade Auto-Organizada. No Capítulo 2 desta tese, apresentamos uma breve introdução a Criticalidade Auto-Organizada. Tendo a criticalidade auto-organizada como pano de fundo, no Capítulo 3, estudamos a dinâmica Bak-Sneppen (e diversas variantes) e a comparamos com alguns algoritmos de otimização. Apresentamos no Capítulo 4, uma revisão histórica e conceitual das redes complexas. Revisamos alguns importantes modelos tais como: Erdös-Rényi, Watts-Strogatz, de configuração e Barabási-Albert. No Capítulo 5, estudamos o modelo Barabási-Albert não-linear. Para este modelo, obtivemos uma expressão analítica para a distribuição de conectividades P(k), válida para amplo espectro do espaço de parâmetros. Propusemos também uma forma analítica para o coeficiente de agrupamento, que foi corroborada por nossas simulações numéricas. Verificamos que a rede Barabási-Albert não-linear pode ser assortativa ou desassortativa e que, somente no caso da rede Barabási-Albert linear, ela é não assortativa. No Capítulo 6, utilizando dados coletados do CD-ROM da revista Placar, construímos uma rede bastante peculiar -- a rede do futebol brasileiro. Primeiramente analisamos a rede bipartida formada por jogadores e clubes. Verificamos que a probabilidade de que um jogador tenha participado de M partidas decai exponencialmente com M, ao passo que a probabilidade de que um jogador tenha marcado G gols segue uma lei de potência. A partir da rede bipartida, construímos a rede unipartida de jogadores, que batizamos de rede de jogadores do futebol brasileiro. Nessa rede, determinamos várias grandezas: o comprimento médio do menor caminho e os coeficientes de agrupamento e de assortatividade. A rede de jogadores de futebol brasileiro nos permitiu analisar a evolução temporal dessas grandezas, uma oportunidade rara em se tratando de redes reais. / Models and scientific theories arise from the necessity of the human being to better understand how the world works. Driven by this purpose new models and techniques have been created. For instance, one of these theories recently developed is the Self-Organized Criticality, which is shortly introduced in the Chapter 2 of this thesis. In the framework of the Self-Organized Criticality theory, we investigate the standard Bak-Sneppen dynamics as well some variants of it and compare them with optimization algorithms (Chapter 3). We present a historical and conceptual review of complex networks in the Chapter 4. Some important models like: Erdös-Rényi, Watts-Strogatz, configuration model and Barabási-Albert are revised. In the Chapter 5, we analyze the nonlinear Barabási-Albert model. For this model, we got an analytical expression for the connectivity distribution P(k), which is valid for a wide range of the space parameters. We also proposed an exact analytical expression for the clustering coefficient which corroborates very well with our numerical simulations. The nonlinear Barabási-Albert network can be assortative or disassortative and only in the particular case of the linear Barabási-Albert model, the network is no assortative. In the Chapter 6, we used collected data from a CD-ROM released by the magazine Placar and constructed a very peculiar network -- the Brazilian soccer network. First, we analyzed the bipartite network formed by players and clubs. We find out that the probability of a footballer has played M matches decays exponentially with M, whereas the probability of a footballer to score G gols follows a power-law. From the bipartite network, we built the unipartite Brazilian soccer players network. For this network, we determined several important quantities: the average shortest path length, the clustering coefficient and the assortative coefficient. We were also able to analise the time evolution of these quantities -- which represents a very rare opportunity in the study of real networks.