《Repast智能体模拟软件基础资料》由会员分享，可在线阅读，更多相关《Repast智能体模拟软件基础资料（25页珍藏版）》请在金锄头文库上搜索。
1、RepastnThe Repast 3 family of tools is the precursor to Repast SimphonyqRepast for JavaqRepast for the Microsoft .NET framework, and qRepast for Python Scripting nRepast Simphony (Repast S) is in pure JavaqExtends the Repast portfolio by offering a new approach to simulation development and executio
2、nqA set of advanced computing technologies for applications such as social simulationqJava APIs as well as GUIs for running and developing models and Eclipse pluginnRepast home page: http:/ nSeveral links on the right of the page qPapersqDocumentationnOn-line introductory tutorial: shows you how to
3、use the interface to create modelsqComes with the download, installed as RepastSimphony-1.2.0docsGetting Started.pdf qThe HTML version is in the folder/RepastSimphony-1.2.0/docs/tutorial/SIM/nRepast Simphony Development Environment: How to set up an Eclipse development environment for Repast. qThis
4、comes with the download, and is installed as RepastSimphony-1.2.0eclipse and the fileRepastSimphony-1.2.0eclipsereadmereadme.html nRepast Simphony Reference: Coding the APIsqThis comes with the download, and is installed as RepastSimphony-1.2.0docsReference.pdf qThe HTML version is in the folderRepa
5、stSimphony-1.2.0docsreferenceSIMFeatures of Repast SimphonynFluid model component development using any mixture of Java, Groovy, and a flowchart interface qGroovy: a dynamic/scripting language for the Java Platform as an alternative to Java nFeatures similar to those of Python, Ruby, PerlnLibraries
6、for qgenetic algorithms qneural networks qregressionnEstimate parameters to give a best fit for the datae.g., least squares qrandom number generation qspecialized mathematicsnAn automated Monte Carlo simulation framework qRelies on repeated random sapling qUsed when its infeasible/impossible to comp
7、ute an exact result with a deterministic algorithmnAutomated connections to optional external tools, including plugins to qthe R statistics environment (alternative to commercial SPSS) q*ORA risk assessment toolnGiven social, knowledge and task network info, find those people, skills or knowledge an
8、d tasks critical for performance or info security qPajek network analysis programqJoSQL for performing SQL queries on collections of Java objects qthe VisAD (Visualization for Algorithm Development) scientific visualization package,nA Java component library for interactive and collaborative visualiz
9、ation and analysis of numerical data qthe Weka data mining platform, nMachine learning algorithms for data mining tasks that can be applied directly to a dataset or called from Java codeqmany popular spreadsheets qMATLABnFull integration withqthe JUNG (Java Universal Network/Graph Framework) network
10、 modeling library nAn extendible language for the modeling, analysis, and visualization of data that can be represented as a graph or networkqMicrosoft Excel spreadsheets qUCINET DL file importingnA comprehensive package for the analysis of social network data ($40 for students)qGeographical spaces
11、including 2D and 3D GIS supportInfo from the Repast 3 Websitehttp:/ Deschamp, Univ. of WellingtonnRepast to simulate a knowledge network involving heterogeneous agentse.g., small and medium enterprises, universities, public research institutes qHow co-operative research strategies can be implemented
12、 among agents with different objectives and rewardsqHow this affects the dynamics of innovation qHow agents evolve and convergeqThe impact at the system level Owen Densmore of Sun MicrosystemsnRepast to model self organizing networks within the context of the Local Knowledge Networking projectqSeeks
13、 to establish computing sessions without central control, thus more adaptive during failure and are more robustqPropagate requests via alternative techniques (e.g., rings of neighbors)TutorialsHowto DocumentationqThese mostly relate to using the APIs and are all briefThird Party TutorialsqOf variabl
14、e qualityqSeveral of the links are deadqSee especially the link to Tesfatsions tutorial (but RepastJ)ExamplesnJava Web Start (.jnlp) filesnClick the linknClick OK in the resulting window, asking whether to open it with the Java Web Start LaunchernGet a control bar and a windownWindow with tabs for q
15、Parameters: set the values for the model parameters (see examples, below)qCustom Actions: controls for the simulationqRepast Actions: built-in Repast functionalitye.g., making movies, edit chartsnControl barnThe control icons, left to right:qLoad ModelqMultiple Runs StartqStartqStepqInitializeqStopq
16、PauseqSetup ModelqView ParametersqExitnFirst step a couple of times, then start, and pause to inspect a given state, then step or start again Basic Grid Examples nHeatbugs qSee the Swarm description: http:/www.swarm.org/examples-heatbugs.htmlParameter tabs selectedSimulation after a whileBasic Grid
17、Examples (cont.)qHeatbugs on Hexagons (Same reference as previous)qHypercycles (See the reference)qGame of Life (See Wikipedia, “Conways Game of Life”)qMousetrap nSee the Swarm description: http:/www.swarm.org/examples-mousetrap.htmlqSugarScape (The reference is given, but also Google it.)Network Ex
18、amples qJiggle Toy (Explanation given)qFriendship Network (Link to reference)Geographic Information Systems (GIS) Examples qGIS Agents (Blank screen) System Dynamics Examples qRabbit Population (Brief explanation and link to reference) Installing Repast SimphonynGo to http:/ nUnder heading “The Late
19、st Repast Simphony Release” at the top of the page,qclick on “Windows Installer”nIn the short wizard-based installation process, accept all the defaults Running a Repast Simphony ExamplenAll programs Repast Simphony Repast Simphony IDEnAccept this default nIn the Package Explorer panel (left side of
20、 window), expand the repast.simphony.demos foldernThe Model: simple variation of predator-prey behavior using 3 agent types: wolves, sheep, and grass nBoth the wolves and sheep move randomly on a gridqMovement has a cost: lost energy. nWolves and sheep need to eat food to replenish their energyqThey
21、 die once their energy level reaches 0nWolves prey on sheep qMay eat them if the 2 are located in the same spatial positionnSheep may eat grass if located on a patch with living grassnOnce eaten, grass needs to re-grow before being eaten again qRe-grow rate: counting down after the grass has been ea
22、ten nReproduction is modeled by a random process qCreates a child from the parentqDivides the energy of the parent agent in halfqAssigns the energy equally to the parent and child nRun Run Configuration nIn the panel on left, click on PredatorPreynDont modify the defaults nClick RunnSee the 4 tabs b
23、elow the panel on the leftnThe Scenario Tree is currently selected and is displayed in the panel nClick the Run Options tab qMoves above the Scenario Tree tabqThe Run Options panel now occupies the left side of the windowqThe Schedule Tick Delay slider initially indicates 0 msec.qMove it to, say, 6
24、msec. nNow click the Parameters tabqParameters Panel to appears qFor now, just accept initial valuesqCan change such parameters as the initial number of sheep, what a sheep gains from a given amount of food, and the reproduction rate of the wolves nBack in the Scenario Tree, click A DisplaynThe Gene
25、ral interface is initially selected from among the 4 options available in the left panelqNo need to change any of these values nThe Agent Style and Grid Style interfaces let you qspecify the order in which the agent icons are overlaid in the display (and some more advanced things)qmodify the size of
26、 the grid units qspecify whether grid lines are to be displayedqNo need to modify any of this.nThe Schedule interface lest you modify how the display is updated relative to the simulation.qAgain, just accept defaultsnNow click OK nIn the model window, click the Step Run icon at the top (double arrow
27、 pointing right)qThe scenario is displayed in the large panel. nClick the Step Run icon several times to step through the simulation. nNow stop the simulation by clicking the Stop Simulation icon, a square within a circle, just right of the Step Run iconnIn the Scenario Tree again, double click A Ch
28、artqInitially, the Chart Data Mappings interface is shown.qGet the Visual Setting interface by clicking the option in the left panelqNo need to change any setting in either interfacenClick OK.nBelow the display, there are now two tabsA Chart Predator Prey Model A Display. nClick the Reset Run icon,
29、shown with 2 arrows and the right of the Stop Simulation iconqThe display and the tabs below it disappear. nClick the Step Simulation icon againqThe display and the tabs reappearnClick the Step Simulation button several more times qThen click the A Chart tabqPlots the sheep and wolf populations over
30、 the simulation steps through which you have progressed.nNow show the display (by clicking the tab)qWatch the patches of grass grow and shrink and the sheep and wolf figures increase ad decrease in numberqControl how quickly the simulation runs with the slider in the Run Options panelnClick the pause icon, to the left of the Run Simulation icon, has 2 barsqDisplay the chartnTo dismiss the simulation window, select File ExitnTo terminate Repast, select File Exit in the main Repast window
