Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Chapters 12.1 and 12.

2: Randomness
SIE 431/531
Young-Jun Son, PhD
Systems and Industrial Engineering
son@sie.arizona.edu
Randomness
• Data, parameters are not deterministic
• Example
• Inter-arrival time, service time
• Choices of hamburgers
• Balking behavior
• Machine break down frequency, machine fix frequency
• And many more
Representation of Randomness
• Random variables having statistical distributions (TRIA, UNIF,
NORMAL, EXP, etc)
• ( ) => derivation of statistical distribution =>
Random variate generation
• TRIA (4, 5, 6), UNIF (4, 6), EXP (10), DISC (0.3, 1, 1.0, 2) where
1: Combo#1, 2: Combo #2, etc

• From this statistical distribution (PDF or PMF), how do we get


different parameters (variates) each time?
Method to generate Random Number & Random Variate
• Random number: any number ~ U (0, 1)
• Random variate (observation): any number ~ statistical
distribution: TRIA (4, 5, 6), UNIF (4, 6), EXP (10), DISC (0.3, 1,
1.0, 2), etc
• To generate a (any) random variate, generating a random number
(U (0, 1)) is the first step. Why?
Inverse Transformation Method
• Note: f(X): pdf or pmf, F(X) is CDF
• First sample a random number from U (0, 1)
• Second: Transform: ( ) = F ( ) => Find the value X such that ( )
• Example: TRIA (4, 5, 6), UNIF (4, 6), DISC (0.3, 1, 1.0, 2)
Inverse Transformation Method (2)
• Example: TRIA (4, 5, 6), UNIF (4, 6), DISC (0.3, 1, 1.0, 2)
Random Number Generation
• Fair way to sample a random number from U (0, 1)
• Linear congruential method
• Matrix method
Linear Congruential Method
• Recursive formula

• This meets our objective!! You can come up with your own formul
a. To do so, what do we need to select?
Linear Congruential Method (2)
• Example: 5, 3, 16, 7 (seed)
Example 1
• The most common random number generator in use today is call
ed a linear congruential generator. The linear congruential gener
ator that we use is as follows:
i  (2i 1  3) mod 5
• Using this generator, manually generate 3 consecutive random va
riates (i.e. i=1, 2, and 3) following the Uniform distribution, Unif
(4, 6). Assume your seed value is 3. Show all the steps of your
answers.
Example 1 (2)
Example 2
• For the same random numbers, derive three random variates foll
owing a discrete distribution: 10% of combo 1, 80% of combo 2,
10% of combo 3
Linear Congruential Method
• LCM parameters used by Arena until version 3.0
• m = 231 – 1
• a = 75 = 16807
• c=0
• Z0 = 14561, ….., 32535

• Cycle, Full Cycle


• Period (p): the length of a cycle (p m)
• Generator is called “full period” if p == m
Matrix Method
• Purpose: random number generator ~ U(0, 1)
• Exactly same with LCM but more complex (matrix) calculation

• What is the seed required? ( )


• Once we obtain a number between 0~1, what do we do?
Desired Properties of RNG
• Reproducible: if ( ) is same, the sequence of random numb
ers will be same => any good?
• Obtaining ( ) random numbers in different runs
=> we can check this using a statistical test!!
Concept of Multiple Random Number Streams
• Each stream corresponds with a seed
• Sometimes, we assign diff. streams for diff. random sources for
comparing a system

You might also like