Here you find a list of our previous talks, including slides and recording, if available.

Climate modelling across scales – from single molecules to the Paris agreement

Speaker: Prof. Philip Stier
Date: 25th June, 2021
Materials: Slides & Recording
Abstract: Climate change is one of the most pressing global challenges. Yet, advances in climate science, mitigation and adaptation remain held back by our ability accurately and reliably predict future climate on relevant scales. In this presentation, I will introduce the audience to climate modelling and the associated key uncertainties, opening the lid of what is widely perceived (and used) as a black box. We will follow a process chain from the scale of individual molecules, cloud droplets, all the way to the global radiation balance – and learn how this relates to the remaining carbon budget to keep Earth’s temperature below the targets of the Paris agreement. Looking forward, we will discuss an ongoing major transformation of climate modelling, which will resolve some of the uncertainties discussed – and introduce others.

Modelling cerebral perfusion during acute ischaemic stroke using a multiscale model

Speaker: Raymond Padmos
Date: 18th June, 2021
Materials: Slides & Recording
Abstract: An acute ischaemic stroke is caused by the sudden blockage of a major cerebral vessel by a thrombus. Every year, millions of people suffer a stroke, resulting in disability and possibly death. The sudden loss of blood flow to tissue, i.e. perfusion, leads to the formation of a cerebral infarct. Understanding how an infarct forms during a stroke can help medical decision making and treatment development. To understand infarct formation, it is necessary to understand how a stroke affects brain tissue perfusion. In this talk, I will present some of my recent work in modelling brain perfusion and predicting infarct volume. In addition, I will present a brief overview of the INSIST project, a consortium dedicated to developing in silico trials for the treatment of acute ischaemic stroke.

The impact of different hydrodynamics on astrophysical predictions

Speaker: Joey Braspenning
Date: 11th June, 2021
Materials: Slides & Recording
Abstract: To form new stars, galaxies need cold gas. This gas can only originate from intermediate temperature gas in the galaxies’ halo, which rapidly cools. However, observing this gas is near impossible. To add to the problem, astrophysicists don’t know quite how it can survive in the first place. In trying to solve this riddle, they turn to simulations. Though large cosmological simulations have made great strides in furthering our understanding of galaxy formation and evolution, they lack the resolution to resolve the intricate dynamics of this intermediate temperature gas. Recently, astronomers have utilized more idealized, high resolution simulations trying to understand what happens when cold and hot gas meet. The past 3 years has seen a proliferation of research in this field. Seemingly every group added their own physics modules: chemistry, cooling, conduction and magnetic fields are just a start. Unfortunately, a handful of different hydrodynamical codes, which should all solve the same equations, is used. This adds to the difficulty of making an honest comparison. All codes make legitimate physical choices which highlights different approaches to understanding the same physical problem. Similarity and comparability between the codes is implicitly assumed when the literature is read. In this talk I will focus on stripping away all the fancy physics and comparing quite how similar those basic codes are in non-trivial problems. Studying the simulations both quantitatively and qualitatively, I will show that there are significant differences between the codes and try to understand where they originate. Then I will translate this to an astrophysical setting and consider the impact these differences have on observable quantities. These are not just technical considerations, but could have a large impact on predictions theorists feed to observational astronomers. It also highlights how we all have to be aware that in choosing a simulation software, which is often calibrated on a narrow set of test problems, we are unconsciously biasing our results by the design choices made in that software.

Detecting resilience loss in ecosystems

Speaker: Dr. Juan C. Rocha
Date: 4th June, 2021
Materials: Slides & Recording
Abstract: Ecosystems around the world are at risk of critical transitions due to increasing anthropogenic pressures and climate change. Yet, it is unclear where the risks are higher or where in the world ecosystems are more vulnerable. Here I measure resilience of primary productivity proxies for marine and terrestrial ecosystems globally. Preliminary results show that up to 30% of global terrestrial ecosystems show symptoms of resilience loss. These early warning signals affect all types of biomes, but by large Arctic and boreal forests are the most affected. Although our results are likely an underestimation, they enable the identification of risk areas as well as the potential synchrony of some transitions. Mapping where ecosystems are likely to undergo critical transitions or long transients can help prioritize areas for management interventions and conservation. Our results pave the way towards developing an ecological resilience observatory.

How to use information to promote cooperation?

Speaker: Dr. Fernando P. Santos
Date: 28th May, 2021
Materials: Slides & Recording
Abstract: Social dilemmas of cooperation pervade human societies and, with the advent of AI, artificial multi-agent systems. In such settings, cooperation is socially desirable yet hard to incentivize. Reputations and information about individuals’ behaviors are key enablers of reciprocity and consequently cooperation. In this regard, it is fundamental to understand 1) which pieces of information are necessary for cooperation to emerge and 2) how information-based interventions fare against other types of incentives (e.g., monetary). In this talk, I will present new evolutionary game theoretical models to study the co-evolving dynamics of reputations and cooperation in dynamical social systems. I will introduce a way of quantifying the complexity of cooperation based on reputations and shed light on the trade-offs between simplicity and efficiency in that context. Finally, I will present a model and simulations revealing the pernicious role of social perception biases in public goods dilemmas and show how changing the information available to individuals can prompt cooperation and collective action.

Community detection for correlation matrices and an application to portfolio risk modelling

Speaker: Dr. Ioannis Anagnostou
Date: 30th April, 2021
Materials: Slides & Recording
Abstract: One of the most challenging aspects in the analysis and modelling of financial markets is the presence of an emergent, intermediate level of structure standing in between the microscopic dynamics of individual financial entities and the macroscopic dynamicsof the market as a whole. This mesoscopic level of organisation is often sought for via factor models that ultimately decompose the market according to geographic regions and economic industries. However, at a more general level, the presence of mesoscopic structure might be revealed in an entirely data-driven approach, looking for a modular and possibly hierarchical organisation of the empirical correlation matrix between financial time series. Thecrucial ingredient in such an approach is the definition of an appropriate null model for the correlation matrix. Recent research showed that community detection techniques developed for networks become intrinsically biased when applied to correlation matrices. For this reason, a method based on Random Matrix Theory has been developed, which identifies the optimal hierarchical decomposition of the system into internally correlated and mutually anti-correlated communities. Building upon this technique, here we resolvethe mesoscopic structure of the CDS market and identify groups of issuers that cannot be traced back to standard industry/region taxonomies, thereby being inaccessible to standard factor models. We use this decomposition to introduce a novel default risk modelthat is shown to outperform more traditional alternatives.

VVUQ using the VECMA toolkit: Examples in migration and local epidemiological agent-based modelling

Speaker: Dr. Derek Groen
Date: 23th April, 2021
Materials: Slides & Recording
Abstract: In this talk Dr. Groen will highlight the VECMA toolkit, which enables researchers to perform verification, validation, uncertainty quantification and sensitivity analysis on their simulations irrespective of their scientific domain or scale of execution. For instance he will explain how users can mix and match different components to speed up development, perform automated validation routines and quickly run large studies to analyse the behavior of their models under different conditions or with different assumptions. In particular Dr. Groen will highlight two applications: (i) the Flee migration model, which is applied in several regions in Africa in a collaboration with the Save The Children NGO to forecast where people will migrate to when conflicts erupt, and (ii) the Flu And Coronavirus Simulator, which is used to make hyperlocal epidemiological predictions in collaboration with local Hospital Trusts in North and West London.

From Simulation to Reality in Robotics

Speaker: Dr. Edward Johns
Date: 16th April, 2021
Materials: Recording
Abstract: Robot simulators have been used for decades, due to the ability to rapidly test new methods for robot control. But in recent years, a new use of robot simulators has emerged: to collect simulated data for large-scale machine learning, such as deep learning, to train robot controllers. In this talk, I will discuss the methods we have developed in my lab to enable robot controllers to be trained in simulation, and then transferred to the real world, with little or no real-world data at all.

Emergence of property rights in competitive resource allocation

Speaker: Dr. Clélia de Mulatier
Date: 9th April, 2021
Materials: Recording
Abstract: We study a system of multiple agents competing for resources with different payoffs, and explore the possible equilibrium strategies for agents with non-aggressive and non-cooperative behaviors. We show that the system can reach different types of equilibria depending on the crowdedness of the resources. In under-crowded systems, resources are equally shared between agents. However, as crowdedness increases, the system becomes conducive to the emergence of property rights. By analogy with physics, the agents’ equilibrium society thus transitions from being liquid-like to being semi-crystalized. We characterize these phases theoretically, as well as numerically with reinforcement learning simulations.

Modelling cascading interactions in climate and social tipping dynamics in the Earth system: Risks and opportunities

Speaker: Dr. Jonathan Donges
Date: 26th March, 2021
Materials: Recording
Abstract: Tipping elements in the Earth’s climate system are continental-scale subsystems that are characterized by a nonlinear threshold behavior. These include biosphere components (e.g. the Amazon rainforest and coral reefs), cryosphere components (e.g. the Greenland and Antarctic ice sheets) and large-scale atmospheric and oceanic circulations (e.g. the thermohaline circulation, ENSO and Indian summer monsoon). Once operating near a threshold or tipping point that may be approached due to anthropogenic climate change, these components can transgress into a qualitatively different state by small external perturbations. The large-scale environmental consequences could impact the livelihoods of millions of people. In this seminar, Dr. Jonathan Donges reports on recent research on modelling individual tipping elements such as the Antarctic Ice Sheet, reinforcing (positive) feedbacks on anthropogenic global warming mediated by cryospheric tipping elements, interactions between climate tipping elements and the risk for resulting tipping cascades. Finally, he will present work on the potentials for positive social tipping dynamics that could help to achieve the rapid decarbonization of the world’s social-economic systems needed to stabilize the Earth’s climate in line with the Paris climate agreement.

Interdisciplinarity Can Aid the Spread of Better Methods Between Scientific Communities

Speaker: Dr. Paul Smaldino
Date: 19th March, 2021
Materials: Slides & Recording
Abstract: Why do bad methods persist in some academic disciplines, even when they have been clearly rejected in others? What factors allow good methodological advances to spread across disciplines? I will discuss some key features determining the success and failure of methodological spread between the sciences. I will then introduce a formal model that considers factors like methodological competence and reviewer bias towards one’s own methods.The model helps to show how self-preferential biases can protect poor methodology within scientific communities, and lack of reviewer competence can contribute to failures to adopt better methods. However, input from outside disciplines, especially in the form of peer review and other credit assignment mechanisms, can help break down barriers to methodological improvement.

Rational optimization of drug-membrane selectivity by computational screening

Speaker: Bernadette Mohr
Date: 12th March, 2021
Materials: Slides & Recording
Abstract: Mitochondria are organelles of eucaryiotic cells involved in a number of physiological pathways. Cardiolipin (CL) is a phospholipid unique to the inner mitochondrial membrane. It plays a central role in mitochondrial functions and dynamics, and CL abnormalities are implicated in diseases. Our goal is to find compounds with high selectivity that can act as CL probes. We explore the capabilities of using a coarse-grained (CG) model to find structures with certain properties. The 5-bead-type reduced Martini force field (T5) is a physics-based model that incorporates both the essential chemical features with a robust treatment of statistical mechanics. It simplifies the molecular representation through a small set of bead types that encode a variety of functional groups. This offers two advantages: first, many molecules map to the same CG representation and second, screening boils down to systematically varying among the set of CG bead types available. We have combined coarse-grained free energy calculations with deep representational learning and Bayesian optimization to efficiently screen the chemical space represented by all T5 compounds up to 400 Da molecular weight. The chemical-space exploration provides general design rules to further optimize selectivity over known CL probes.
Note: We also had two 10 minute talks by Maarten van den Ende and Casper van Elteren, who are both PhD students in the Computational Science Lab at the UvA. Maarten is working on interactive social networks in complex systems, while Casper is looking at how to disrupt criminal networks using a complexity science approach.

Refining the causal loop diagram: A tutorial for maximizing the contribution of domain expertise in computational system dynamics modeling

Speaker: Loes Crielaard
Date: 5th March, 2021
Materials: Slides & Recording
Abstract: Complexity science and systems thinking are increasingly recognized as a relevant paradigm for studying systems where biology, psychology, and socio-environmental factors interact. The application of systems thinking however often stops at developing a conceptual model that visualizes the mapping of causal links within a system, e.g., a causal loop diagram (CLD). While this is an important contribution in itself, it is imperative to subsequently formulate a computational version of a CLD in order to interpret the dynamics of the modeled system and simulate ‘what if’ scenarios. We propose to realize this by deriving knowledge from experts’ mental models in the biopsychosocial domains. This tutorial paper first describes the steps required for capturing expert knowledge in a CLD such that it may result in a computational system dynamics model (SDM). For this purpose, we introduce several annotations to the CLD that facilitate this intended conversion. This annotated CLD (aCLD) includes sources of evidence, intermediary variables, functional forms of causal links, and the distinction between uncertain and known-to-be-absent causal links. We propose an algorithm for developing an aCLD that includes these annotations. We then describe how to formulate an SDM based on the aCLD. The described steps for this conversion help identify, quantify, and potentially reduce sources of uncertainty and obtain confidence in the results of the SDM’s simulations. We utilize a running example that illustrates each step of this conversion process. The approach described in this paper facilitates and advances the application of computational science methods to biopsychosocial systems.

The role of simulation-based science in decision-making for healthcare, government and industry

Speaker: Prof. Peter Coveney
Date: 26th February, 2021
Materials: Slides & Recording
Abstract: As our models and simulations become increasingly capable of describing the complexity of the real world, the opportunity to apply them to pressing problems that concern humanity on the largest scales also grows. This talk will survey a few contemporary examples including epidemiological modelling of the COVID-19 pandemic, climate, weather and protein structure prediction, drug discovery, and fusion energy. We shall also discuss how such models inform and are informed by machine learning and artificial intelligence approaches.

Occupational mobility and automation: A data-driven network model

Speaker: R. Maria del Rio-Chanona
Date: 19th February, 2021
Materials: Slides & Recording
Abstract: The potential impact of automation on the labor market is a topic that has generated significant interest and concern amongst scholars, policymakers, and the broader public. A number of studies have estimated occupation-specific risk profiles by examining how suitable associated skills and tasks are for automation. However, little work has sought to take a more holistic view on the process of labor reallocation and how employment prospects are impacted as displaced workers transition into new jobs. We developed a data-driven model to analyze how workers move through an empirically derived occupational mobility network in response to automation scenarios. At a macro level, our model reproduces the Beveridge curve, a key stylized fact in the labor market At a micro level, our model provides occupation-specific estimates of changes in short and long-term unemployment corresponding to specific automation shocks. We find that the network structure plays an important role in determining unemployment levels, with occupations in particular areas of the network having few job transition opportunities. In automation scenarios where low wage occupations are more likely to be automated than high wage occupations, the network effects are also more likely to increase the long-term unemployment of low wage occupations. At the end of this talk I will discuss future challenges in the labor market.

Game theoretic modeling of helping behavior in emergency evacuations

Speaker: Dr. Jaeyung Kwak
Date: 5th February, 2021
Materials: Slides & Recording
Abstract: People often help others who are in trouble, especially in emergency evacuation situations. For instance, during the 2005 London bombings, it was reported that evacuees helped injured persons to escape the place of danger. In terms of game theory, it can be understood that such helping behavior provides a collective good while it is a costly behavior because the volunteers spend extra time to assist the injured persons in case of emergency evacuations. We study the collective helping behavior in a room evacuation scenario in which two volunteers are required to rescue an injured person. We propose a game theoretic model to study the evolution of cooperation in rescuing the injured persons. We consider the existence of committed volunteers who do not change their decision to help the injured persons. With the committed volunteers, all the injured persons can be rescued depending on the payoff parameters. In contrast, without the committed volunteers, rescuing all the injured persons is not achievable on most occasions because some lonely volunteers often fail to find peers even for low temptation payoff. We have quantified various collective helping behaviors and summarized those collective patterns with phase diagrams. In the context of emergency evacuations, our study highlights the vital importance of the committed volunteers to the collective helping behavior.

Information processing in complex systems

Speaker: Dr. Rick Quax
Date: 29th January, 2021
Materials: Slides & Recording
Abstract: A network of dynamical units can generate a complex systemic behavior. Examples include human cognition emerging from a network of neural cells, ecosystems from food webs, and cellular regulatory processes from protein-protein interactions. A first important question is: which agents are the ‘drivers’ of the systemic behavior? A second question is: can we detect emergent phenomena, particularly ‘criticality’ (susceptibility to perturbations)? We attempt to address these questions in a domain-free way using information theory. As an example, surprising findings show that the most influential nodes are not the most well-connected or central nodes. This has important consequences for many applications. I will present work on addressing the above questions through analytical results, computational modeling, and data analysis.

Between empirical realism and model comprehension: Agent-based computational modelling in Analytical Sociology

Speaker: Prof. Andreas Flache
Date: 22th January, 2021
Materials: Slides
Abstract: Agent-based computational modeling (ABCM) plays a central role in Analytical Sociology (AS). ABCM attracts analytical sociologists because it combines analytic precision, ability to capture complex micro-macro interactions in social phenomena and flexibility to accommodate empirically realistic assumptions. I will discuss how meeting the goal of making empirically realistic assumptions that is put central in AS may create tension with another important aim for ABCM: full comprehension of the link between model assumptions and model implications. By means of a number of examples drawing on sociological applications of ABM, good practices are discussed for how empirical realism can be added to abstract and unrealistic models, while developing sufficient understanding of the complex dynamics generated by an ABCM. Examples draw on models of residential segregation, opinion polarization, and diffusion of innovations.