Download presentation
Presentation is loading. Please wait.
Published byNelson Burke Modified over 8 years ago
1
April 23, 2013Research in Progress Seminar MASS: A Multi-Agent Spatial Simulation Library Munehiro Fukuda, Ph.D. School of Science, Technology, Engineering, and Mathematics 0
2
Table of Contents 1. Software agents 2. MASS: multi-agent spatial simulation library 3. MASS execution performance 4. Practical applications 5. Research issues and plans April 23, 2013Research in Progress Seminar1
3
Software Agents Software agents software that acts on behalf of a user or provides a particular service Cognitive agents coarse-grain execution entities that achieve network- administrative and computation-intensive task, based on their behavioral intelligence Reactive agents fine-grain entities, each reacting to its environment with simple rules. April 23, 2013Research in Progress Seminar2
4
Cognitive Agents eCommerce: Dispatch agents to remote servers to plan an trip and to make reservations accordingly. April 23, 2013Research in Progress Seminar3 Note: a picture from Internet search
5
Reactive Agents Simulations: Have many reactive agents interact each other and observe their group behavior. April 23, 2013Research in Progress Seminar4 Today’s focus Note: a picture from Internet search
6
ABM: Agent-Based Modeling Computational model View computation as interaction of reactive agents or individuals and obtain outputs as their emergent collective behavior. Describe simulations that are difficult to model with mathematical formulas. Examples AntFarm: ants’ food collecting simulation Wa-Tor: a predator prey simulation MatSim: a multi-agent transport simulation FluTe: an influenza epidemic simulation April 23, 2013Research in Progress Seminar5
7
Related Work Swarm: Santa Fe Institute The first execution ABM platform for scientific computing Emphasis on ABM programming Parallelized with multithreading NetLogo: Northwestern Univ. An extension of Logo (in education) Menu-based ABM programming Graphical outputs Interpretive environments April 23, 2013Research in Progress Seminar6 Paramount focus on model design Note: pictures from Wiki
8
Scalability Challenge in ABM Scalability: The more accuracy we pursue, the more agents we need in a simulation Examples: MatSim: 20 minutes to simulate 200K cars driving through Bellevue on I-405 FluTe: 2 hours to simulate epidemic in 10M individuals April 23, 2013Research in Progress Seminar7 Parallelization: simulation with many computers
9
Parallelization Issues Parallel-computing problems 1. Problem partitioning and mapping to computing node 2. Inter-node communication Descriptions 1. Agent collision detection 2. Distributed termination 3. Ghost space management April 23, 2013Research in Progress Seminar8 Our motivation: a design of a parallel ABM execution platoform
10
Multi-Agent Spatial Simulation April 23, 2013Research in Progress Seminar9
11
MASS Specification Public static void main( String[ ] args ) { MASS.init( args ); Places space = new Places( handle, “MySpace”, params, xSize, ySize); Agents agents = new Agents( handle, “MyAgents”, params, space, population ); space.callAll( MySpace.func1, params ); space.exchangeAll( MySpace.func2, neighbors ); agents.exchangeAll( MyAgents.func3 ); agents.manageAll( ); MASS.finish( ); } func2( ) func1( ) …… func3( ) April 23, 2013Research in Progress Seminar The Key is “all in parallel”. 10
12
Implementation Status Java Multithreaded version: John Spiger (BSCSS) Multi-process version: John Emanu (BSCSS) Multi-threaded multi-process version: Tim Chuang (MSCSS) GPU Prototype: Tosa Ojiru and Robert Jordan (MSCSS) C++ in progress: Narayania Chandrasekaran and Cheri Wasous (MSCSS) Tools Sensor to MASS data streaming: Jose Melchor Parallel file reader/writers: Kelsey Weingartner and Sanjoy Bappudi (BSCSS) Trial uses CSS534/490: Parallel Programming in Grid and Cloud (25 students) April 23, 2013Research in Progress Seminar11
13
Performance Evaluation Environments Giga-Ethernet of 24 Linux machines (512MB) Test programs 2D wave dissemination simulation 2D random walk program Conway’s game of life Image steganography April 23, 2013Research in Progress Seminar12
14
2D Wave Simulation April 23, 2013Research in Progress Seminar13 places z[t][i][j] = 2.0 z[t-1][i][j] – z[t-2][i][j] + c2(dt/dd)2 (z[t-1][i+1][j] + z[t-1][i-1][j] + z[t-1][i][j+1] + z[t-1][i][j-1] – 4.0 z[t-1][i][j])
15
Random Walk April 23, 2013Research in Progress Seminar14 Agent communication Agent migration
16
Conway’s Game of Life by Daniel Lewis, a MSCSSE student Simulation consists of grid of cells Live cells with <2 live neighbors die (underpopulation) Live cells with 2–3 live neighbors live on Live cells with >3 live neighbors die (overcrowding) Dead cells with 3 live neighbors come alive (reproduction) #1234 1464.2432.1442.1436.7 274.8655.8556.6056.42 363.0447.1945.8644.16 464.8243.9443.4740.92 Threads Procs April 23, 2013Research in Progress Seminar15 Execution in seconds
17
Image Steganography by Preethi Rajaram, an MSCSS student April 23, 2013Research in Progress Seminar16
18
Summery of MASS Execution Pros Handling quite scalable, (i.e., memory- intensive) applications Cons Not yet best fitted to fine-grain computation, (i.e., a lot of communication among too small agents and cells) April 23, 2013Research in Progress Seminar17
19
Potential Practical Applications 1. Climate analysis 2. Brain Grid: Neural network simulation 3. Protein network motif search 4. FluTe: Influenza epidemic simulation 5. MatSim: Multi-agent transport simulation April 23, 2013Research in Progress Seminar18
20
Climate Analysis with Profs. Eric Salathe and Hazel Asuncion April 23, 2013Research in Progress Seminar19
21
Neural Network Simulation with Prof. Mike Stiber April 23, 2013Research in Progress Seminar20
22
Protein Network Motif Search with Prof. Wooyoung kim April 23, 2013Research in Progress Seminar21
23
Influenza Epidemic Simulation from Univ. New Mexico April 23, 2013Research in Progress Seminar22 communities person Infected Contagious http://www.cs.unm.edu/~dlchao/flute/
24
Multi-Agent Transport Simulation from http://www.matsim.org/ April 23, 2013Research in Progress Seminar23 start goal start goal 45 minutes
25
Research Items 1. A cluster of GPU machines 2. Data streaming 3. Agent diffusion, merger, and distributed termination 4. Ghost space management 5. Guarded agent migration 6. Load balancing April 23, 2013Research in Progress Seminar24
26
Data Streaming with CSS undergraduate students: Kelsey Weingartner and Sanjay Bappudy April 23, 2013Research in Progress Seminar25
27
Conclusions 1. ABM parallelization with the MASS library 2. MASS scalable execution performance with test applications 3. Parallelization of practical applications with UWB colleagues 4. More R & D items to go April 23, 2013Research in Progress Seminar26
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.