Stochastic Simulation

Performance and Dependability

Paulo Maciel

Centro de Informática - UFPE

 Objective

▪ To study the fundaments of stochastic

simulation, its methods, and applying
simulation for solving performance and
dependability problems.
 Program

▪ Introduction
▪ Problems and Mistakes in Simulation
▪ System and Model
▪ Classification of Models
▪ Types of Simulation
▪ Discrete Event Simulation: an Overview
▪ Random Number Generation
▪ Random Variates Generation
▪ Output Analysis – Steady State Simulation
▪ Output Analysis – Transient Simulation
▪ Verification and Validation
 Methodology

▪ Expositive classes

▪ Practical classes
 Evaluation

▪ List resolutions

▪ Write a draft paper

 Basic References

Art of Computer Systems Performance Analysis Techniques For

Experimental Design Measurements Simulation And Modelingm, Raj
Jain, Wiley Computer Publishing, John Wiley & Sons, Inc,1991.

Simulation modeling handbook : a practical approach, Christopher A.

Chung. 2003.

Stochastic Simulation: Algorithms and Analysis, Søren Asmussen,

Peter W. Glynn. Springer, 2007.

Introduction to Discrete Event Systems, Christos G. Cassandras,

Stéphane Lafortune. Springer. 2008
 Introduction

▪ Simulation

▪ Analysis

▪ Evaluation
Introduction

▪ When is simulation
an interesting option?
 Introduction

▪ Common mistakes in simulation projects:

▪ Not considering numerical or analytical methods
as possible alternative
▪ Inappropriate level of detail
▪ Not taking into account the modeling and coding
costs
▪ Unverified and invalid models
▪ Improperly handled initial conditions
▪ Results without confidence level and error margin
▪ Poor random-number generators
▪ Poor interpretation of results
 System and Model

The Concept of System

An aggregation or assemblage of things so combined by nature or
man as to form an integral or complex whole (Encyclopedia
Americana).
An interacting or interdependent group of items forming a unified
whole (Webster’s Dictionary)
A combination of components that act together to perform a function
(IEEE Standard Dictionary).

There are two features in these definitions:

1. a system consists of interacting “components”, and
2. a system is associated with a “function”.
 System and Model

Model

▪ Abstraction level

▪ Perspective
 System and Model

Model
 System and Model

State and state variable

 System and Model

Event

An event may be identified with a specific action

taken by a person or system or
it may be viewed as a spontaneous occurrence
dictated by nature or
it may be the result of several conditions which are
suddenly all met.
System classification
 Types of Simulation

Static vs Dynamic Simulation

▪ Static simulation is adopted for representing a system

in which time does not play a role or to describe a
system at a particular time.

▪ Dynamic simulation is used for depicting a system

that evolves over time.
 Types of Simulation

Deterministic vs Stochastic Simulation

▪ Deterministic simulation considers models that do not

contain any random variable, that is, the output is
determined once the input quantities and the model
relations are defined.

▪ Stochastic simulation adopts models that have at

least one random variable.
 Types of Simulation

Continuous vs Discrete Model Simulation

▪ Continuous model simulation relies on continuous

models for representing the system whereas discrete
model simulation is supported by discrete models.
 Types of Simulation

▪ Trace driven (Deterministic simulation)

▪ Monte Carlo (Stochastic static model simulation)

▪ Discrete event (Stochastic dynamic discrete model

simulation)
 Trace-Driven Simulation

▪ Trace-driven simulation uses traces (time-ordered

record of events on a real system) as its input
applied to the simulation model which depicts the
system behavior. This sort of simulation strategy
has been applied for evaluating resource
management algorithms, such as paging algorithms,
cache analysis, CPU scheduling algorithms, and
algorithms for dynamic allocation of storage.
 Monte Carlo Simulation

▪ Monte Carlo simulations are used to model

probabilistic phenomena that do not change
characteristics over time.

▪ Monte Carlo simulations are also used for

evaluating non-probabilistic expressions using
probabilistic methods.
Monte Carlo Simulation
 Excel
Monte Carlo Simulation
Mathematica

Integration
 Minitab Mathematica
Monte Carlo Simulation
Excel Mathematica
V2
▪ Estimating  - Model

1.0

0.8

0.6

0.4

0.2

Input parameters
0.2 0.4 0.6 0.8 1.0

Model
 Monte Carlo Simulation

Piston pump
 Monte Carlo Simulation

Cont.
 Monte Carlo Simulation

Model
 Monte Carlo Simulation

Input data parameters

 Monte Carlo Simulation

▪ Create random input data

▪ Apply the input data to the model
▪ Analyze the output

Minitab Excel Mathematica

 Monte Carlo Simulation

Analyze the output

 Monte Carlo Simulation

Analyze the output

The average flow rate is pretty close to the specified value, but
the standard deviation far exceed 0.2 ml/min.
Discrete Event Simulation: an Overview
Discrete Event Simulation: an Overview
Discrete Event Simulation: an Overview
Discrete Event Simulation: an Overview
Discrete Event Simulation: an Overview
Discrete Event Simulation: an Overview
Discrete Event Simulation: an Overview
Discrete Event Simulation: an Overview
Discrete Event Simulation: an Overview
 Excel
N/N/1

Discrete Event Simulation: an example

 C:\Users\Paulo Maciel\Dropbox\Models\Mercury\Mercury 4.4.3\Mercury 4.4.3.2\Examples

Discrete Event Simulation: an example

 C:\Users\Paulo Maciel\Dropbox\Models\Mercury\Mercury 4.4.3\Mercury 4.4.3.2\Examples\StudyOnSimulation.xml

Discrete Event Simulation: a SPN example

 C:\Users\Paulo Maciel\Dropbox\Models\Mercury\Mercury 4.4.3\Mercury 4.4.3.2\Examples\StudyOnSimulation.xml

Discrete Event Simulation: a SPN example

 C:\Users\Paulo Maciel\Dropbox\Models\Mercury\Mercury 4.4.3\Mercury 4.4.3.2\Examples\StudyOnSimulation.xml

Discrete Event Simulation: a SPN example

Random Number Generator
 Excel
Random Number Generator

Generation of Uniform Random Numbers

 Excel
Random Number Generator

Generation of Uniform Random Numbers

 Excel
Random Number Generator

Generation of Uniform Random Numbers

 Excel
Random Number Generator

Generation of Uniform Random Numbers

 Excel
Random Number Generator

Generation of Uniform Random Numbers

 Excel
Random Number Generator

Generation of Uniform Random Numbers

 Random Variate Generator

A random variable is a function.

A random variate is an outcome provided by

a random variable.
 Random Variate Generator

Basic methods

– Inversion transform
– Convolution
– Composition
– Characterization
– Acceptance-Rejection
Random Variate Generator
Inversion transform
 Excel
Random Variate Generator Exponential

Inversion transform Mathematica

Exponential
The sequence of random numbers will not be identical,
but will be statistically equivalent.
Generation of Exponential Variate
 Random Variate Generator Excel
Exponential

Inversion transform Mathematica

Exponential

Generation of Exponential Variate

 Random Variate Generator Excel
Exponential

Inversion transform Mathematica

Exponential

Generation of Exponential Variate

Random Variate Generator Excel
Discrete
Inversion transform Mathematica
Discrete
Random Variate Generator Excel
Discrete
Inversion transform Mathematica
Discrete
Random Variate Generator Excel
Discrete
Inversion transform Mathematica
Discrete
Random Variate Generator Excel
Discrete
Inversion transform Mathematica
Discrete
Random Variate Generator
Inversion transform
Random Variate Generator Excel
Erlang
Convolution Mathematica
Erlang
Random Variate Generator Excel
Erlang
Convolution Mathematica
Erlang
Random Variate Generator Excel
Hyper-Exponential
Composition Mathematica
Hyper-Exponential
Random Variate Generator Excel
Hyper-Exponential
Composition Mathematica
Hyper-Exponential
Random Variate Generator Excel
Hyper-Exponential
Composition Mathematica
Hyper-Exponential
Random Variate Generator Excel
Normal
Characterization Mathematica
Normal
Random Variate Generator Excel
Normal
Characterization Mathematica
Normal
Random Variate Generator Excel
Acceptance-Rejection
Random Variate Generator Excel
Acceptance-Rejection
Random Variate Generator Excel
Acceptance-Rejection
Random Variate Generator Excel
Acceptance-Rejection
Random Variate Generator Excel
Acceptance-Rejection
Random Variate Generator Excel

Deciding which method to adopt

 Excel
Output Analysis
 Excel
Output Analysis
 Excel
Output Analysis
 Excel
Output Analysis
Output Analysis Excel
Output Analysis Excel
Output Analysis Excel
Output Analysis
Output Analysis
Output Analysis
Output Analysis Excel
 Excel
Output Analysis
Output Analysis Excel
Output Analysis Excel
Some Details
Output Analysis Excel
Output Analysis
Output Analysis
Output Analysis
Output Analysis
Output Analysis
Output Analysis Excel
Output Analysis Excel

<
Output Analysis Excel
Output Analysis Excel
Output Analysis
Excel
Output Analysis Excel

Correlation
Output Analysis Excel
Correlation
Output Analysis Excel
Correlation
Output Analysis Excel
Correlation
Output Analysis Excel
Mathematica
Correlation
Excel
Minitab
Correlation
Correlation
Output Analysis Excel
Mathematica
Correlation
Excel
Minitab
Correlation
Correlation