Jonathan Ozik

Complex adaptive systems modeling with Repast Simphony

PurposeThis paper is to describe development of the features and functions of Repast Simphony, the widely used, free, and open source agent-based modeling environment that builds on the Repast 3 library. Repast Simphony was designed from the ground up with a focus on well-factored abstractions. The resulting code has a modular architecture that allows individual components such as networks, logging, and time scheduling to be replaced as needed. The Repast family of agent-based modeling software has collectively been under continuous development for more than 10 years.MethodIncludes reviewing other free and open-source modeling libraries and environments as well as describing the architecture of Repast Simphony. The architectural description includes a discussion of the Simphony application framework, the core module, ReLogo, data collection, the geographical information system, visualization, freeze drying, and third party application integration.ResultsInclude a review of several Repast Simphony applications and brief tutorial on how to use Repast Simphony to model a simple complex adaptive system.ConclusionsWe discuss opportunities for future work, including plans to provide support for increasingly large-scale modeling efforts.

Modeling Dynamic Multiscale Social Processes in Agent-Based Models

Identity-related issues play central roles in many current events, including those involving factional politics, sectarianism, and tribal conflicts. Two popular models from the computational-social-science (CSS) literature - the threat anticipation program and SharedID models - incorporate notions of identity (individual and collective) and processes of identity formation. A multiscale conceptual framework that extends some ideas presented in these models and draws other capabilities from the broader CSS literature is useful in modeling the formation of political identities. The dynamic, multiscale processes that constitute and transform social identities can be mapped to expressive structures of the framework.

Large-Scale Agent-Based Modeling with Repast HPC: A Case Study in Parallelizing an Agent-Based Model

We present a case study for parallelizing a large-scale epidemiologic ABM developed with Repast HPC, the Chicago Social Interaction Model (chiSIM). The original serial model is a CA-MRSA model which tracks CA-MRSA transmission dynamics and infection in Chicago, and represents the spread of CA-MRSA in the population of Chicago. We utilize both within compute node parallelization using the OpenMP toolkit and distributed parallelization across multiple processes using MPI. The combined approach yields a 1350 % increase in run time performance utilizing 128 compute nodes.

Test-driven agent-based simulation development

Developing a useful agent-based model and simulation typically involves acquiring knowledge of the models domain, developing the model itself, and then translating the model into software. This process can be complex and is an iterative one where changes in domain knowledge and model requirements or specifications can cause changes in the software that in turn may require additional modeling and domain knowledge. Test-driven development is a software development technique that can help ameliorate this complexity by evolving a loosely coupled flexible design, driven by the creation of many small, automated unit tests. When the focus shifts to writing small tests that exercise the simulations behavior, the larger problem of translating a conceptual model into working code is decomposed into a series of much smaller, more manageable and highly focused translations. This paper explores the application of this technique to agent-based simulation development with examples from Repast Simphony, ReLogo and Repast HPC.

From desktop to large-scale model exploration with Swift/T

As high-performance computing resources have become increasingly available, new modes of computational processing and experimentation have become possible. This tutorial presents the Extreme-scale Model Exploration with Swift/T (EMEWS) framework for combining existing capabilities for model exploration approaches (e.g., model calibration, metaheuristics, data assimilation) and simulations (or any “black box” application code) with the Swift/T parallel scripting language to run scientific workflows on a variety of computing resources, from desktop to academic clusters to Top 500 level supercomputers. We will present a number of use-cases, starting with a simple agent-based model parameter sweep, and ending with a complex adaptive parameter space exploration workflow coordinating ensembles of distributed simulations. The use-cases are published on a public repository for interested parties to download and run on their own.

The relogo agent-based modeling language

ReLogo is a new agent-based modeling (ABM) domain specific language (DSL) for developing agent-based models in the free and open source Repast Suite of ABM tools; the Java based Repast Simphony ABM toolkit and the C++ high performance computing Repast HPC toolkit both incorporate ReLogo. The language is geared towards a wide range of modeling and programming expertise, combining the sophisticated and powerful ABM infrastructure and capabilities in the Repast Suite with the ease of use of the Logo programming language and its associated programming idioms. This paper will present how ReLogo combines a number of concepts, including object-oriented programming, simple integration of existing code libraries, statically and dynamically typed languages, domain specific languages, and the use of integrated development environments, to create an ABM tool that is easy to learn yet is also capable of creating large scale ABMs of real world complex systems.

Simulating water, individuals, and management using a coupled and distributed approach

Water is a key issue in sustainable urban development. SWIM (Simulating Water, Individuals and Management) is an agent-based model of water supply, management structure, and residential water consumer perception and behavior. Initial work applied data mining on newspaper articles to map networks of water management institutions and structures. SWIM extends this by linking an agent-based model of residential water consumption connected via networks of water managers to a global-scale hydrological model. In our case study, we focus on Tucson, Arizona, where management and social behaviors are well documented. Census data are used to create synthetic populations of consumers endowed with price sensitivity and behaviors impacting water use. Social networks, including those based on geographic proximity, allow water use behaviors to spread to others. We examine possible factors leading to recent attested declines in per-capita water use, leveraging ensemble runs on high-performance computing resources using the Swift parallel scripting language to strategically explore complex parameter spaces.

High-throughput cancer hypothesis testing with an integrated PhysiCell-EMEWS workflow

Cancer is a complex, multiscale dynamical system, with interactions between tumor cells and non-cancerous systems. Therapies act on this cancer-host system, sometimes with unexpected results. Systematic investigation of mechanistic models could help identify the factors driving a treatment’s success or failure, but exploring mechanistic models over high-dimensional parameter spaces is computationally challenging. In this paper, we introduce a high throughput computing (HTC) framework that integrates a mechanistic 3-D multicellular simulator (PhysiCell) with an extreme-scale model exploration platform (EMEWS) to investigate high-dimensional parameter spaces. We show early results in adapting PhysiCell-EMEWS to 3-D cancer immunotherapy and show insights on therapeutic failure. We describe a PhysiCell-EMEWS work-flow for high-throughput cancer hypothesis testing, where thousands of mechanistic simulations are compared against data-driven error metrics to perform hypothesis optimization. We close by discussing novel applications to synthetic multicellular systems for cancer therapy.

Endogenous Social Networks from Large-Scale Agent-Based Models

We present a parallel computational method for generating endogenous social networks from large-scale simulation data from the Chicago Social Interaction Model (chiSIM). The model scope aims to simulate the population of the entire city of Chicago which includes approximately 2.9 million discrete individuals. Generated person collocation networks contain more than 106 person nodes and more than 109 collocation edges. Analysis of such network structure can be challenging when applied to urban scale population data due to their size. A parallel logging implementation is described that records person activity data via an agent-based model. The social network generation method is demonstrated on a distributed compute cluster. The person collocation network analysis and visualization obtained via the parallel methodology provides previously unreported characterizations of simulated large-scale urban social network structure.