Abstract: In non-equilibrium statistical physics, a deep understanding of the role played by the entropy production is nowadays an open and fundamental question. In the seminal work of Schnakenberg (Rev. Mod. Phys. (1976), 48(4):571), he introduced an information-based formula for the entropy production of systems described by Master Equation. However, the physical meaning of such a formulation and its link with the thermodynamic entropy production is still debated. To shed some light on this problem, we study a Master Equation system with symmetric transition rates coupled to an environment by the input of an external flux current (i.e. detailed balance does not hold in this case). Interestingly, the problem can be mapped into a classical electrical circuit, and we show analytically and numerically that the Schnakenberg's entropy production is proportional to the heat dissipated by the circuit, and therefore to its classical entropy production. This result provides a strong evidence of the connection between an information-based quantity and thermodynamics. Finally, we can also extend our framework through a perturbative analysis to the case of slightly asymmetric transition rates.

Abstract: Soil-transmitted helminthiasis (intestinal worms) is a collection of neglected tropical diseases affecting more than 1 billion people worldwide. These infections cause severe morbidity in school-aged children in whom the majority of worms are harboured. De-worming treatments are available through mass drug administration (MDA); however recent research has concluded that treatment of children alone is not sufficient to break the cycle of transmission in a high transmission setting. We develop a stochastic transmission model focusing on the household structure in the population. Using Markov Chain Monte Carlo (MCMC) and Approximate Bayesian Computation (ABC) methods we fit the model to a cross-sectional study of worm burdens in rural Nigeria and evaluate the effectiveness of alternative MDA strategies that prioritise the treatment of `wormy` households. We discuss how the approach taken enables the utilisation of efficient MCMC methods up to a certain model complexity after which computational considerations necessitate the use of likelihood-free inference methods such as ABC.

Abstract: The perspective of social-ecological systems (SESs) postulates that humans and nature domains are linked in complex ways. Changes in one domain often produce changes in the other domain that might bring the SES close to a threshold that, if crossed, changes the whole identity of the SES. The proper use of early warning signals plays a critical role in preventing the crossing of an undesired threshold. For any given SES there are different views of what system’s dynamics is desirable. Thus, decisions about future states must acknowledge those multiple views. In that sense, the development of a boundary object (BO) −an artifact that facilitates communication among stakeholders− provides a means for generating the necessary knowledge about a system’s thresholds taking into consideration the salient issues and concerns, in a way that is legitimate and credible for all stakeholders. Through the case study of SESs delimited by the gray whale breeding coastal lagoons in Baja California, Mexico, we illustrate the development of a BO in the form of a hybrid model. Our hybrid model (a model that combines a dynamic-system model with an agent-based model) aims to produce an early warning signal to be used in the development of regulations for activities related to gray whale watching. Because in essence the identity of the SES is given by the gray whale, we will first elicit a set of scenarios regarding the carrying capacity of whale-watching boats through a system model. Next, we will couple the system model to an agent-based model that incorporates the social-economic domain. The BO thus produced, will be tested in the following year.

Abstract: The Louvain algorithm is one of the most widely used and fastest known algorithm for uncovering communities in complex networks. The algorithm consists of a local move heuristic where each node is assigned to its locally optimal community, after which the graph is aggregated and the procedure is reiterated on the aggregated graph. However, the local move heuristic repeatedly considers moving nodes to neighbouring communities, even though their local environment remained unchanged. We here suggest to only consider nodes which can be considered unstable: their local environment changed during the course of the local move heuristic. This fast local move algorithm shows considerably faster runtimes, speeding up the Louvain algorithm over 20 times in experiments, without degrading the quality of the uncovered partition.

Abstract: Is Complexity Economics just an extension to standard economics or a radical different way of seing economics? We think that today, 30 years after the first attempts of the Santa Fe Institute to define Complexity Economics (CE) hesitating about these two directions, the answer is now clear. CE provides a totally different framework for economic analysis. The objectif here is to define this framework more precisely. This is indeed a challenge since the first point upon almost everybody agrees is that there is no one definition of what complexity is. We think that the answer to this question depends on the level of generality at which one positions herself. At a paradigmatic level, it is possible to find an encompassing characterization of CE. Following Kuhn, a paradigm is a framework which defines the relevant problems and accepted methods, allowing for a greater efficiency in research: a common language and world apprehension that enhance the diffusion of works and canalizes investigations. We develop such definition around various level: philosophical, ontological, methodological, core concept and agenda levels. We here inform these 5 levels in order to provide a broad definition of the paradigm of CE.

Abstract: There are 201 metro systems in the world used by 115 million users daily. A better understanding of rapid transit systems (RTS) can lead to improved performance, safety, and comfort. Using computer simulations, we can test different regulatory methods to compare their performance. An important subject for RTS is the control of train frequency, also known as the headway. If headways are equal then passenger waiting time at stations will be minimized. However, this configuration is unstable. There are three main causes that can deviate headways, two in the interstation segment (trains going faster or slower than expected), and one at stations (changes on passenger demand). The possible solutions demand adaptive strategies. Self-organizing methods are useful to adjust the intervals of waiting time at stations, considering the demand of passengers resulting in a supra-optimal performance in RTS due to a slower is faster effect. If the delay appears on the interstation segment the self-organizing method can correct the headway when the train arrives at station using local information of the environment. In this work we present a realistic agent-based model for RTS that evolves in discrete time and continuous space. We use it to test adaptive strategies to control headways. The model has two main components, an extended version of the Gipps Model (a microscopic car-following model) and a variation of a self-organizing method to regulate headways. To calibrate the variables of the model we consider a real scenario of a Mexico City metro line and recreate a passenger dynamics to using station entry data. We compare the performance between the actual configuration and the self-organizing regulatory methods. Our results show the potential benefit of using self-organization to regulate RTS and its implementation viability for real systems: rapid transit and public transportation in general.

Abstract: Brain, as a self-organized critical system, is the most complex system in nature, and has been always attractive for many scientists. Brain functionality depends on various features, the most important of which is the topological structure of the neuronal network. Several research studies found that the small-world property has a substantial impact on optimizing the brain function. But this is still a questionable finding, which necessitates more investigation for better understanding of the brain complex network. In this study, we investigate the importance of the effect of high connectivity on the brain network, and also on its self-organization property. In this respect, we simulate the brain structure using complex network theory, along with a spreading model for the brain dynamics. We also investigate the effect of heterogeneity of the connectivity by generating different types of networks, such as Erdos-Renyi, Small-World and Scale-Free networks. We show that by increasing the average degree, the avalanche size/duration distributions approach into the universal critical power-law behavior with mean-field solutions, independent of the network topology. On the other hand, we also show that for a network with small average degree, in spite of possessing the small-world property, the system is far from the critical state. We conclude that one of the most important factors affecting the performance of self-organization of the brain may be high connectivity, which is in agreement with the observations.

Abstract: Biology forms a complex system with a nested hierarchical structure and clear parent-child relationships, where a collective interaction of children defines properties of a parent. For example, interaction of molecular machinery defines properties of cells, whereas interaction of cells defines the properties of the whole organism. Nowadays, research of all the levels of biological organisation, from molecules to ecosystems, benefits from application of computational modelling. Arguably, joining models from different levels would allow us to understand biology much better. However, in practice, it is hard for an individual, or even a group, to develop a holistic predictive model of at least several levels of biological organisation due to complexity at each individual level. Furthermore, the diversity of implementation methods, algorithms, and programming languages that are used to create biological models makes it hard to cooperate and share. Here I present a minimalistic C++ library, which purpose is to combine modelling efforts across the levels of biological organisation, into a single computational environment. It exploits a generic programming approach to create a recursive, hierarchical template, which can be populated with user-defined models. The structure of the template imposes strict encapsulation rules on the models, while leaving developers a free choice of implementation methods, thus promoting cooperation and resolving compatibility conflicts between the models.

Abstract: Many systems in Nature can be described by a combination of local reaction rules - representing the creation or destruction of entities - with a diffusion process - the migration of entities. Reaction-diffusion equations are therefore widely used to model such systems. In accordance with the growing interest for network science, we consider these equations when evolving on complex networks. We analyze them from the self-organization point of view, and more precisely for Turing instabilities which can explain the emergence of patterns under the effect of diffusion. The reach of Turing's study now goes far beyond the original biological setting, as it is now widespread in physics, chemistry, ecology, social systems, just to name a few. We present our study of a two-species activator-inhibitor model, accounting for the presence of a time-delay impacting all the processes in the nodes of the network, possibly due to a wait-then-act strategy or simply the consequence of inertia in the concentration sites [1,2]. Our approach is based on a spectral analysis and on the properties of the scalar Lambert W-function, allowing us to obtain a complete description of the stability of the system. We explore the role of the different variables in the parameters space, and predict the formation of either stationary patterns, or Turing-like waves. With our 5 minutes presentation, we hope to put three main points across. First, why it came to us to incorporate a time-delay in a regular reaction-diffusion model. Second, suggest both a taste of the mathematical challenge this represents, and how we picked our way through it. And finally, give a feeling for the role of every parameters in the self-organizing behavior of the model, with a particular focus on the time-delay. References [1] Petit et al, EPL 111, 58002 (2015) [2] Petit et al, arXiv 1603.07122 (2016)

Abstract: For this IGNITE presentation at CCS2016, Amsterdam-based Slow Research Lab presents the material innovations of designer Maria Blaisse and how they address human-scale challenges in modeling/simulating complexity and fluid interface design. Blaisse’s work stems from a tradition of designers working at the intersection of form, mathematics, and research on nonlinear dynamics in nature (e.g., Buckminster Fuller’s “synergetics,” Paul Schatz’ “polysomatic forms”), while dovetailing with material/physical design theory that has influenced programming languages and object-oriented interfaces (e.g., Christopher Alexander’s “deep geometrical structures,” “generative grammar,” etc.). Blaisse’s current research trajectory into the formal, relational, and energy-generating potentialities of bamboo has yielded a series of flexible structures that give tangible expression to humanity’s possible near-future advances in infrastructure (architecture, transport, energy), interface design (fluid, human-scale, interactive mathematical modeling) and socio-ecological systems (systems resilience simulation, environmental bioenergetics). Format for this presentation: Slow Research Lab director Carolyn Strauss will deliver a three minute oral analysis of Blaisse’s work in relation to interaction design thinking strategies for modeling and simulating complex systems, accompanied by a performance demonstration of the bamboo structures by movement artist and complexity researcher Siobhán K. Cronin. (This video exhibits possibilities for the performance: http://bit.ly/1rCKp4j) Attendees of the presentation will gain insight into the role that design can play in expanding the frontiers of complexity research. The presentation is certain to amaze and inspire, literally igniting conversation and new pathways of interdisciplinary collaboration.

Abstract: How do multiple topics compete for the limited resource of user attention in social networks? What are the main paths in the process of information diffusion? Although much of the previous work had focused on the study of topics competition in social networks, most of them were model-based analysis, which cannot describe the temporal and spatial behaviors of users in the real world. In this work, we analyze the information sharing logs from WeChat – the largest instant messaging communication service in China – with the goal of understanding the diffusion processes. We analyzed the role of social groups and the underlying social structure in the topics competition, and presented a timeline visualization by compositing ThemeRiver with storyline style visualization. The increase and decrease of competitiveness of the topics are shown in this work.

Abstract: Social influence has been an important topic of research in the social sciences. Recently, thanks to the huge amounts of digital data produced by our daily activities, research on how much our behaviour is driven by others has produced new studies with non-trivial results. We use Last.fm's song-listen data to quantify social influence on music listenership around live events. We study live events performed by “Hyped” (i.e. trending) and “Top” (i.e. most popular) artists. We analyse how listenership changes around the time of the live event both for users who attended the event, and more importantly, for those who did not, but are friends with someone who attended. The analysis in this paper uses publicly available data from the music website Last.fm. We use three distinct types of data: (1) event attendance, (2) track listens, and (3) the Last.fm friends network. We extracted all live events in 2013 and 2014 by the most popular Hyped and Top artists. We tracked listenership of 1.3 million users over a two month time horizon—with one month of listenership data prior to the attendance of an event and one month of listenership data after the attendance of the event. Our analysis show that social influence exist in both cases of hyped and top artists but it is much stronger in the latter case. We also show how the number of "friends" would increase the effects of social influence. We argue how complex social contagion is the right model to explain these observations.