News Category: ‘Past Friday Seminars’

Friday, 11th September: 3.00pm

A Relationship Between Growth and Size

Rob Axtell, Department of Computational Social Science

Rob will present a relationship between growth rates and sizes of subgroups in heterogeneous populations. Skew size distributions are shown to require heavy-tailed growth rates for stability. Alternatively, heavy-tailed growth processes induce skew sizes. Applications to firms, mutual funds, and bird populations have been investigated. Some abstract computational examples will be presented. The relationship of these new results to the related Gibrat and Kesten processes will be explored.

Friday, 18th September: 3.00pm

Agent Modeling Using GPUs

Matthew Hendry, Department of Computational Social Science

The advances in the computational power of Graphic Processing Units (GPUs) have created supercomputer capabilities at the cost of a cheap PC.  Many fields have successfully adapted their algorithms to the highly parallel GPU architecture with significant increase in computational performance.  This past summer we explored whether agent-based models could add their name to the growing list of problems amenable to GPUs.  I’ll discuss our preliminary results coding two different ABMs and compare them to the MASON implementations.

Friday, 25th September: 3.00pm

Innovation in Software Development as an Evolutionary Process: Comparison Between Software Engineers and Genetic Programming Agents

Randy Casstevens, Department of Computational Social Science

Genetic programming was developed with the goal of having computers automatically develop code without explicitly being told what to do. This resembles many tasks that software engineers face.  In the simplest case, a potential user tells the developer an incomplete list of inputs and expected outputs.  It is then the software developerâ??s job to convert these software requirements into a design and later into an implementation.  This presentation will explore whether patterns seen by software engineers can be recreated using an agent-based model.  This model consists of agents that are using genetic programming to automatically generate software solutions.

Data from software engineers competing in a Matlab programming contest will be compared with results gathered from the genetic programming agents.  The genetic programming agents will make random changes to the programs and the programs with the best performance are most likely to survive to the next generation.  Conversely the software engineers make purposeful, non-random changes to the code.  It is not
expected that the genetic programming agents will exhibit all of the same patterns seen by software engineers, but some of the patterns persist.  Therefore, these patterns may not be a result of higher cognitive functions of humans, but rather a more fundamental part of evolutionary problem solving.

Much of the power from genetic programming comes from its crossover operator where a section of code is replaced by another individual’s code.  In the Matlab contest, developers were also allowed to borrow and modify code from fellow programmers.  Furthermore, the code in the Matlab contest evolved over time with changes being contributed by many different programmers.  This presentation will compare the products of these two similar, but very different types of software development.

Friday, 9th October: 3.00pm

Introduction to NetLogo

Holly Russo, Department of Computational Social Science

NetLogo is a Java-based modeling environment and one of the easiest environments in which to construct an agent-based model with a rich visualization quickly.  It is also free and has extensive resources such as a huge model library full of example code and a pretty good userâ??s manual.  Many of the Computational Social Science classes you will take will require you to build an agent-based model, and usually you will have the choice of modeling environments.  NetLogo is a great option for those of you not comfortable with another language yet or even as a programming â??white boardâ?? for rapid prototyping and testing of algorithms.  However, NetLogo is also capable of some pretty sophisticated programs and provides Systems Dynamics and Social Network modeling, too.

This seminar is an introduction to NetLogo.  During the seminar, we will cover all of the basics you need to get started programming as well as tips on where to find more help and information.  We will go over the basic components of NetLogo, the format for a NetLogo program, how to set up and conduct experiments, and how to import and export files.  We will also go over the model library and userâ??s manual, both of which are priceless tools even for the experienced NetLogo programmer.  We will talk a little about different versions of NetLogo.  Finally, we will briefly discuss other extensions and advanced features of NetLogo; however,  we will not cover them in any great depth.

This seminar is for those who are completely new to NetLogo or who have only built a couple of programs.  Other more experienced students, although welcome to attend, will not likely benefit much from this seminar but may appreciate the refresher.  During the seminar, I will collect the email addresses of attendees, in order to send out soft copies of the slides used in the presentation.  We will look mostly at models from the model library so you â?? armed with the slides from our discussion â?? can explore the models on your own time.   I will also show you one or two of the more sophisticated models that I have built in the past, so you get a feel for what is possible in NetLogo, and email the models (with code) to you after the seminar.

Friday, 16th October: 3.00pm

Some Cognitive Foundations of Computational Social Science

Bill Kennedy, Department of Computational Social Science

This talk will be a small sampling of the topics to be addressed in a course offered next spring with nearly the same title.  Officially, the course “examines cognitive foundations and information processing in computational social agents and compares to comparable human cognitive phenomena, including emotions, trust, reciprocity [with] an emphasis on modeling project”.  The class will discuss basic human needs, basic and cognitive emotions, and drivers of individual social behavior within groups and society.  The talk this Friday will address an aspect of individual decision-making and an aspect of group performance, both with audience participation.

Friday, 23rd October: 3.00pm

Friday, 30th November: 3.00pm

Friday, 6th November

Modeling the Fairfax County Alcohol System

Yasmin Said, Computational and Data Sciences, George Mason University

Users of alcohol are incorporated into a societal system, which for many purposes resembles an ecological system. An understanding of how this ecological alcohol system works provides an opportunity to evaluate effectiveness of interventions. I use a hybrid directed graph social network model calibrated with conditional probabilities derived from actual data with the idea of reproducing the experience of acute outcomes reflecting undesirable individual and societal outcomes. In the present model, I also approximate geospatial effects related to transportation as well as temporal effects. Drinking behaviors among underage users can be particularly harmful from both a societal and individual perspective. Using the model based on data from experiences in Fairfax County, Virginia, I am able to reproduce the multinomial probability distribution of acute outcomes with high accuracy using a microsimulation of all residents of Fairfax, approximately 1,000,000 agents simulated. By adjusting conditional probabilities corresponding to interventions, I am able to simulate the effects of those interventions. This methodology provides an effective tool for investigating the impact of interventions and thus provides guidance for public policy related to alcohol use.

Friday, 13th November: 3.00pm

In Search of the Roots of Social Complexity

Chris Rouly, Department of Computational Social Science

This is a report on results from research that tested a hypothesis stating the aggregate phenomenon known to social science as clan-level social complexity can be explained computationally using the following theoretic components: human metabolic and bio-reproductive theory, nutrient seeking (foraging) behaviors, evolutionary theory, and a spatial ecology. To test this hypothesis an agent-based model was created wherein each agent had an artificial chromosome containing eleven genes (8-bits per gene) ten of which had independently inheritable, graded, and expressible traits like draught tolerance, temperature sensitivity, a robust metabolism, and improved fecundity in small-group settings, for example. The agents could move freely within a diverse 2D ecology, enjoyed caloric and water metabolic costs, and had human-like 28-day reproductive cycles with gestation and nursing metabolic adjustments. One simulation in particular ran for more than 200 agent generations of 6,421 simulated years. Reported here are results of that simulation, the changes that occurred in three of the eleven genes in the agent population under test as a consequence of artificial evolution, and how “sociality” may have begun to emerge in the population.