Santa Fe Institute

About: Santa Fe Institute is a nonprofit organization based out in Santa Fe, New Mexico, United States. It is known for research contribution in the topics: Population & Context (language use). The organization has 558 authors who have published 4558 publications receiving 396015 citations. The organization is also known as: SFI.

Heterogeneity, correlations and financial contagion

Abstract: We consider a model of contagion in financial networks recently introduced in the literature, and we characterize the effect of a few features empirically observed in real networks on the stability of the system. Notably, we consider the effect of heterogeneous degree distributions, heterogeneous balance sheet size and degree correlations between banks. We study the probability of contagion conditional on the failure of a random bank, the most connected bank and the biggest bank, and we consider the effect of targeted policies aimed at increasing the capital requirements of a few banks with high connectivity or big balance sheets. Networks with heterogeneous degree distributions are shown to be more resilient to contagion triggered by the failure of a random bank, but more fragile with respect to contagion triggered by the failure of highly connected nodes. A power law distribution of balance sheet size is shown to induce an inefficient diversification that makes the system more prone to contagion events. A targeted policy aimed at reinforcing the stability of the biggest banks is shown to improve the stability of the system in the regime of high average degree. Finally, disassortative mixing, such as that observed in real banking networks, is shown to enhance the stability of the system.

Gossip: Identifying Central Individuals in a Social Network

Abstract: Is it possible, simply by asking a few members of a community, to identify individuals who are best placed to diffuse information? A simple model of diffusion shows how boundedly rational individuals can, just by tracking gossip about people, identify those who are most central in a network according to "diffusion centrality" (a measure of network centrality which nests existing ones, and predicts the extent to which piece of information seeded to a network member diffuses in finite time). Using rich network data from 35 Indian villages, we find that respondents accurately nominate those who are diffusion central -- not just traditional leaders or those with many friends. In a subsequent randomized field experiment in 213 villages, we track the diffusion of a piece of information initially given to a small number of "seeds" in each community. Seeds who are nominated by others lead to a near tripling of the spread of information relative to randomly chosen seeds. Diffusion centrality accounts for some, but not all, of the extra diffusion from these nominated seeds compared to other seeds (including those with high social status) in our experiment.

A thousand empirical adaptive landscapes and their navigability

Abstract: The adaptive landscape is an iconic metaphor that pervades evolutionary biology. It was mostly applied in theoretical models until recent years, when empirical data began to allow partial landscape reconstructions. Here, we exhaustively analyse 1,137 complete landscapes from 129 eukaryotic species, each describing the binding affinity of a transcription factor to all possible short DNA sequences. We find that the navigability of these landscapes through single mutations is intermediate to that of additive and shuffled null models, suggesting that binding affinity-and thereby gene expression-is readily fine-tuned via mutations in transcription factor binding sites. The landscapes have few peaks that vary in their accessibility and in the number of sequences they contain. Binding sites in the mouse genome are enriched in sequences found in the peaks of especially navigable landscapes and the genetic diversity of binding sites in yeast increases with the number of sequences in a peak. Our findings suggest that landscape navigability may have contributed to the enormous success of transcriptional regulation as a source of evolutionary adaptations and innovations.

Random graph models for dynamic networks

Abstract: Recent theoretical work on the modeling of network structure has focused primarily on networks that are static and unchanging, but many real-world networks change their structure over time. There exist natural generalizations to the dynamic case of many static network models, including the classic random graph, the configuration model, and the stochastic block model, where one assumes that the appearance and disappearance of edges are governed by continuous-time Markov processes with rate parameters that can depend on properties of the nodes. Here we give an introduction to this class of models, showing for instance how one can compute their equilibrium properties. We also demonstrate their use in data analysis and statistical inference, giving efficient algorithms for fitting them to observed network data using the method of maximum likelihood. This allows us, for example, to estimate the time constants of network evolution or infer community structure from temporal network data using cues embedded both in the probabilities over time that node pairs are connected by edges and in the characteristic dynamics of edge appearance and disappearance. We illustrate these methods with a selection of applications, both to computer-generated test networks and real-world examples.

Nonergodicity and central-limit behavior for long-range Hamiltonians

Abstract: We present a molecular dynamics test of the Central-Limit Theorem (CLT) in a paradigmatic long-range-interacting many-body classical Hamiltonian system, the HMF model. We calculate sums of velocities at equidistant times along deterministic trajectories for different sizes and energy densities. We show that, when the system is in a chaotic regime (specifically, at thermal equilibrium), ergodicity is essentially verified, and the Pdfs of the sums appear to be Gaussians, consistently with the standard CLT. When the system is, instead, only weakly chaotic (specifically, along longstanding metastable Quasi-Stationary States), nonergodicity (i.e., discrepant ensemble and time averages) is observed, and robust q-Gaussian attractors emerge, consistently with recently proved generalizations of the CLT.