Presentation is loading. Please wait.

Presentation is loading. Please wait.

Distributed Multi-Agent Management in a parallel-programming simulation and analysis environment: diffusion, guarded migration, merger and termination.

Similar presentations


Presentation on theme: "Distributed Multi-Agent Management in a parallel-programming simulation and analysis environment: diffusion, guarded migration, merger and termination."— Presentation transcript:

1 Distributed Multi-Agent Management in a parallel-programming simulation and analysis environment: diffusion, guarded migration, merger and termination Cherie Wasous CSS_700 Thesis Research – Autumn 2013

2 “parallel-programming simulation and analysis environment”

3 MASS (Multi-Agent Spatial Simulation ) parallel computing software library from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” “parallel-programming simulation and analysis environment”

4 MASS (Multi-Agent Spatial Simulation ) parallel computing software library Simplifies the task of creating and running parallel applications across multiple computers and CPU cores. from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” “parallel-programming simulation and analysis environment”

5 MASS (Multi-Agent Spatial Simulation ) parallel computing software library Simplifies the task of creating and running parallel applications across multiple computers and CPU cores. from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” “parallel-programming simulation and analysis environment”

6 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” MASS v.1 enables two primary types of simulations: 1.A Stand-alone Grid of Stationary Locations Example: A map and its grid of sub-locations

7 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” MASS v.1 enables two primary types of simulations: 1.A Stand-alone Grid of Stationary Locations Example: A map and its grid of sub-locations 2.A Grid of Stationary Locations with Mobile Units Example: A map, its grid of sub-locations, and mobile units

8 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” MASS Library 1. Stand-alone Grid of Stationary Locations Composed of: The Grid of Locations referred to as Places in MASS contains the grid of stationary locations

9 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” MASS Library 1. Stand-alone Grid of Stationary Locations Composed of: The Grid of Locations referred to as Places in MASS contains the grid of stationary locations Single Stationary Location referred to as a Place object in MASS contains basic information about the local place Example: local Temperature

10 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” MASS Library 2. Grid of Stationary Locations with Mobile Units Composed of: Collection of Mobile Units referred to as “Bag of” Agents in MASS each Place location has a “Bag of” Agents contains the mobile units for the Place location

11 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” MASS Library 2. Grid of Stationary Locations with Mobile Units Composed of: Collection of Mobile Units referred to as “Bag of” Agents in MASS each Place location has a “Bag of” Agents contains the mobile units for the Place location Mobile Unit referred to as Agent in MASS contains basic information about the mobile unit Example: Amount of sugar an ant (agent) has consumed

12 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” Overall MASS Framework Places Maintain & manages the Place locations Manages exchange between the Place locations callAll( ) callSome( ) exchangeAll( ) exchangeBoundary( )

13 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” Overall MASS Framework Places Maintain & manages the Place locations Manages exchange between the Place locations Place Maintains Place location data Provides a user software interface callAll( ) callSome( ) exchangeAll( ) exchangeBoundary( ) callMethod( ) {User created functions}

14 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” Overall MASS Framework Places Maintain & Manages the Place locations Manages exchange between the Place locations Place Maintains Place location data Provides a user software interface Agents Maintains & Manages the Agent units Manages the exchange and migration of Agent units callAll( ) callSome( ) exchangeAll( ) exchangeBoundary( ) callMethod( ) {User created functions} callAll( ) manageAll( ) migrate( ) spawn( ) kill( )

15 from: Romanus, css497 summer2013, “Developing and Extending the MASS Library (Java) Places.exchangeBoundary( )” Overall MASS Framework Places Maintain & Manages the Place locations Manages exchange between the Place locations Place Maintains Place location data Provides a user software interface Agents Maintains & Manages the Agent units Manages the exchange and migration of Agent units Agent Maintains the Agent data Provides a user software interface callAll( ) callSome( ) exchangeAll( ) exchangeBoundary( ) callMethod( ) {User created functions} callAll( ) manageAll( ) migrate( ) spawn( ) kill( ) callMethod( ) {User created functions}

16 MASS execution model from: Chuang, MS Thesis, “Design and Qualitative/Quantative Analysis of Multi-Agent Spatial Simulation Library”

17 from: Fukuda, et al, NSF proposal Fall 2013, “Multi-Agent-Based Parallelization of Scientific Data Analysis and Simulation” MASS v.2 Supports NetCDF in Parallel. Facilitates big-data analysis.

18 MASS v.future Enhance Multi-Agent Management. Eases adoption and improves performance for scientific data analysis and simulation.

19 “diffusion, guarded migration, merger, and termination” Diffusion MASS currently can: Create an agent at every place element Create an agent at specified place elements Create a large number of agents and spread across the place elements

20 “diffusion, guarded migration, merger, and termination” Diffusion MASS currently can: Create an agent at every place element Create an agent at specified place elements Create a large number of agents and spread across the place elements MASS future enhancements: Improve efficiency of creation more local ; reducing communication between nodes and between threads “Diffusion” option – where a few agents are spread across the place elements at each thread, and they efficiently clone themselves locally to all adjacent place elements ; leaving a footprint where visited

21 “diffusion, guarded migration, merger, and termination” Diffusion MASS currently can: Create an agent at every place element Create an agent at specified place elements Create a large number of agents and spread across the place elements MASS future enhancements: Improve efficiency of creation more local ; reducing communication between nodes and between threads “Diffusion” option – where a few agents are spread across the place elements at each thread, and they efficiently clone themselves locally to all adjacent place elements ; leaving a footprint where visited Continue Focus on User Interface: Keep it Simple, Powerful and Efficient.

22 “diffusion, guarded migration, merger, and termination” Guarded Migration MASS currently does not limit the number of agents migrating into a single place element. However, Traffic Simulation – limit just one car in a space Epidemic Simulations – limit each household to X number of people

23 “diffusion, guarded migration, merger, and termination” Guarded Migration MASS currently does not limit the number of agents migrating into a single place element. However, Traffic Simulation – limit just one car in a space Epidemic Simulations – limit each household to X number of people MASS future enhancements: Allow user to specify maximum agents per single place element Guarded Migration – using a fair, deterministic, distributed arbitration technique

24 “diffusion, guarded migration, merger, and termination” Guarded Migration MASS currently does not limit the number of agents migrating into a single place element. However, Traffic Simulation – limit just one car in a space Epidemic Simulations – limit each household to X number of people MASS future enhancements: Allow user to specify maximum agents per single place element Guarded Migration – using a fair, deterministic, distributed arbitration technique Continue Focus on User Interface: Keep it Simple, Powerful and Efficient.

25 “diffusion, guarded migration, merger, and termination” Merger, and Termination MASS currently does not support agent merge command. Cumbersome for user to write code for this. MASS currently only supports kill command for single agent. Cumbersome and inefficient for user to kill each agent.

26 “diffusion, guarded migration, merger, and termination” Merger, and Termination MASS currently does not support agent merge command. Cumbersome for user to write code for this. MASS currently only supports kill command for single agent. Cumbersome and inefficient for user to kill each agent. Continue Focus on User Interface: Keep it Simple, Powerful and Efficient.

27 Questions ???


Download ppt "Distributed Multi-Agent Management in a parallel-programming simulation and analysis environment: diffusion, guarded migration, merger and termination."

Similar presentations


Ads by Google