Download presentation
Presentation is loading. Please wait.
Published byMitchell Sparks Modified over 9 years ago
1
CS 351/ IT 351 Modeling and Simulation Technologies Review (2012-04-04) Dr. Jim Holten
2
CS 351/ IT 351 Overview Software Engineering Models Graphs Hybrid Dynamical System (HDS) Behavior Models Critical Infrastructure Models and Network Flow Models Errors High Performance Computing (HPC)
3
CS 351/ IT 351 Software Engineering Maintainability and readability Development and testing Top-down, bottom-up, etc.
4
CS 351/ IT 351
5
Models Overview of uses Meshes Regular Polyhedral Mesh General Polyhedral Mesh Finite Element Mesh Networks Particles Other?
6
CS 351/ IT 351 Models
7
CS 351/ IT 351 Meshes
8
CS 351/ IT 351 Networks
9
CS 351/ IT 351 Graphs Nodes and Links Paths, diameter Hypergraphs, Hypernodes, Hyperlinks
10
CS 351/ IT 351 Graphs and Hypernode
11
CS 351/ IT 351 HDS Behavior Models Dynamics Models Modes of behavior – separate behavior models (state variable values over time) Finite State Machine (FSM) – control which behavior model to use at any given time Inputs, state variables, and outputs Internal patterns – detect internal and external conditions for FSM input control
12
CS 351/ IT 351 Agent HDS Model
13
CS 351/ IT 351 Critical Infrastructure and Network Flow Models Interdependent networked resource flows Graph-based topological description Node behavior as agents with HDS behavior models Analysis tools?
14
CS 351/ IT 351 Errors In Models Sources of Errors Representing Errors Validation and Verification
15
CS 351/ IT 351 Sources of Errors Data Acquisition Accuracy Data Value Representations Real World versus Ideal Model Numerical Methods Accuracy Wrong Algorithm and Coding Errors
16
CS 351/ IT 351 Numerical Methods Accuracy
17
CS 351/ IT 351 Representing Errors Stochastic Distributions – probability distribution function (PDF) Normal (Gaussian) Distribution Uniform Other Bounded Errors Generating Random Variates – using the cumulative distribution function (CDF)
18
CS 351/ IT 351 Error Distribution 1.Histogram 2.Normalize to PDF 3.Integrate to CDF
19
CS 351/ IT 351 High Performance Computing (HPC) Why? Data too large for one system Takes too long to run to completion Runs too slow to keep up with data stream Break model into parallel components Data parallel partitions Task parallel operations Task parallel pipelines
20
CS 351/ IT 351 Data Parallel
21
CS 351/ IT 351 Process to Parallel Pipeline/Combination Process quisi Acquisition
22
CS 351/ IT 351 High Performance Computing (HPC) Issues? Hard to convert sequential or small parallel programs to large numbers of processes Hard to debug Number of processes, data elements, interrelationships – hard to keep track of the all Small number of knowledgeable programmers Application domain and parallel programming essential for these jobs Next step?
23
CS 351/ IT 351 Questions?
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.