On Wednesday 26th November, Enzo Nicosia, from the School of Mathematical Sciences at Queen Mary University will give the talk
Phase transitions in network growth: from street patterns to neural networks
Mountbatten Room 4025A (53/4025A), Highfield Campus, 4pm. All welcome. Refreshments served after the talk.
Abstract: Studying how a system has evolved over time can help unraveling the elementary mechanisms responsible of its formation and understanding the complexity of the wiring patterns connecting its constituent units. We have recently found that the evolution of spatially-embedded systems, such as urban street patterns and neural networks, can be characterised by interesting phase transitions, resulting in abrupt changes of some structural descriptor of the system. In this talk we will discuss two examples of phase transition in the growth of a network, respectively corresponding to the evolution over 200 years of a large metropolitan area in Italy and to the growth of the C.elegans nematode’s neural systems, and we will present reasonable models able to accurately reproduce the observed transitions.
On Wednesday 12th November, Mervyn Freeman, from the British Antarctic Survey, Cambridge, will give the talk:
Some random walks in space weather
Please note this is being held at an earlier time (2pm)
Mountbatten Room 4025A (53/4025A), Highfield Campus, 2pm. All welcome. Refreshments served after the talk.
Abstract: Space weather – the natural variability of near-Earth space – is a hazard to modern society, with the potential to affect satellite services and electricity supply. It is the fourth highest natural hazard risk on the Government’s National Risk Register, alongside heatwaves and low temperatures. Arguably the greatest source of uncertainty is the substorm – an earthquake-like disruption of near-Earth space that eludes deterministic prediction. In this talk, I will show how substorm occurrence may be understood as a random walk towards an absorbing barrier, similar to the Gerstein-Mandelbrot model of neuron firing in the brain. I will discuss the limits of prediction and similarities to the space weather of Jupiter and Saturn.
On Wednesday 5th October, Alex Kalloniatis, from Defence Science and Technology Organisation, Canberra, Australia, will give the talk:
Networks and synchronisation: mathematical modelling of socio-technical decision making systems
Life Sciences Building (85), room 2207, Highfield Campus, 4pm. All welcome. Refreshments served after the talk.
Abstract: Much of the decision making in structured organisations such as military, business and administrative units is distributed across many individuals, each contributing information that spans the spectrum from simple ‘facts’ about who and what is ‘out there’, to an integrated understanding of what is happening and why, and what will happen in the future – so that future actions can be determined. They also involve a variety of technological systems and displays to facilitate such Situation Awareness and Sense-Making. Such decision making systems may be represented as networks, and the process of making decisions may be represented as the evolution of individual and collective states in time. In this talk I will present my own efforts at encoding such a model in terms of a system of differential equations based on the well known system of synchronising oscillators, the Kuramoto model. I present two adaptations of this model. The first represents two adversary organisations engaging in a competitive process, each seeking to outpace the speed of decision making of the other. The second focuses on a single organisation but structured in the manner of a typical Divisional construct with separate branches performing different organisational functions, operating under different time pressures and cycles. In both cases, rich seemingly unexpected behaviour arises from numerical solution or simulation – there is ’emergence’ – and dynamical regimes lying between order and chaos. I discuss the value of such models in their tractability to explore, for example, optimal network structures, and the scope for expanding such models to find a balance between Levins’ dimensions of Realism, Precision and Generality.
On Wednesday 29th October, Doyne Farmer, from the University of Oxford, will give the talk:
A long-term vision for computational economics
Nightingale Building (67) room 1003, Highfield Campus, 4pm. All welcome. Refreshments served after the talk.
Abstract: The ability to simulate phenomena is probably the biggest driver of theoretical progress in physical science during the last 50 years. The same is not true in economics and social science in general. Why is this so? I will argue that the time is ripe for this to happen and present a plan for how it could be done and what breakthroughs are required. I will review the accomplishments of agent-based models in economics so far, discuss the key theoretical and practical challenges for creating the next generation of models, and present key lessons from the CRISIS project (for which I am scientific coordinator). I will particularly focus on the need for large scale simulation models, analogous to the global circulation models used in meteorology and climate, and discuss the similarities and differences with meteorology. Finally I will present a vision of what such large scale models might be like and what they would enable us to do ten or twenty years from now.
On Wednesday 24th September, Chris Gordon-Smith, from the SimSoup project, will give the talk
“Molecules Designed for Chemical Network Memory and Non-Genetic Inheritance”
Please note the temporary change of venue.
Nuffield Theatre (Building 6) Room 1081 (Nuffield Room B), Highfield Campus, 4pm. All welcome. Refreshments served after the talk.
In this talk I will present a memory system based on an artificial chemistry. This is relevant as a ‘proof of concept’ for metabolism based Origin of Life theories, and in the field of biological and chemical computing. Each memory unit can be switched between three alternative active states. A unit maintains itself in a particular state using an autocatalytic reaction process. Switching between states occurs when an external stimulus triggers the autocatalytic process for the new state, along with an associated process that inhibits autocatalytic activity for the old state. I will show artificial molecular species with structures that support the autocatalytic and inhibiting processes. I will also present results from the SimSoup artificial chemistry simulator showing the operation of a 5-unit memory system with 243 alternative states (equivalent to just under 8 bits of memory). The design supports systems with more units, but computational requirements to run the simulator increase substantially. I will conclude the talk with a short review of some alternative network architectures for chemical memory and inheritance.