Top Banner
ETM 607 – Random Number and Random Variates • Define random numbers and .pseudo-random numbers • Generation of random numbers • Test for random numbers - Frequency tests - Autocorrelation • Random-Variate generation - Inverse-transform technique
24

ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Dec 25, 2015

Download

Documents

Ethan Bond
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

ETM 607 – Random Number and Random Variates

• Define random numbers and .pseudo-random numbers• Generation of random numbers• Test for random numbers

- Frequency tests- Autocorrelation

• Random-Variate generation- Inverse-transform technique

Page 2: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Definitions:Random number (Ri) – a value between 0 and 1.0, ~ U[0,1).

Random Variable – a variable with an associated probability distribution.Random Variable – a function that assigns a real number to each outcome in the sample space (Feldman and Valdez-Flores).

ex. X ~ U[0,1) Y ~ Exp(5.75)

Z ~ Normal(8.0,1.0)

ETM 607 – Random Number and Random Variates

Page 3: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Random Number (Ri):

12

1

12

)(][

2

1

2][

,0

10,1)(

2

abXV

baXE

otherwise

xxf

ETM 607 – Random Number and Random Variates

Page 4: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Random Number (Ri) – statistical properties:

Uniformity – if divided into n intervals of equal length, then the expected number of observations in each interval is n/N, where N is the total number of observations.

Independence – the probability of a value in a particular interval is independent of the previously generated value.

ETM 607 – Random Number and Random Variates

Page 5: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Generation of Pseudo-Random Number:

Pseudo – false, or “not quite”.

Random numbers generated in a computer. Not exactly random, but generated from an algorithm and are in fact repeatable given same starting position (good for debugging).

ETM 607 – Random Number and Random Variates

Page 6: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Generation of Pseudo-Random Number:

Goal – develop generation method such that output most closely imitates ideal properties of uniformity and independence.

Considerations1.fast and efficient (in code)2.portable to different computers3.should have long cycles (before number pattern repeats)4.Should be repeatable (for debugging)5.Closely approximate true ~U[0,1)

ETM 607 – Random Number and Random Variates

Page 7: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Linear Congruential Method:

Sequence of intergers: X1, X2, X3,…. Xn between 0 and m-1.

Xi+1 = (a Xi+ c) mod mWhere, X0 - initial seeda - multiplierc – incrementm – modulus

Then,

ETM 607 – Random Number and Random Variates

m

XR i

i

Page 8: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Linear Congruential Method:

In class exercise:

Xi+1 = (a Xi+ c) mod mGiven, X0 - 5376a - 13c – 0m – 10000

Find,

ETM 607 – Random Number and Random Variates

4321 ,, RandRRR

Page 9: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Linear Congruential Method:

Why was m = 10000 effective (from a computational perspective) in the example.

In computer algorithm, m is usually a function of 2b , producing same effect in binary terms.

Much research done to determine effective values of a and c to produce long cycles, uniformity and independence.

ETM 607 – Random Number and Random Variates

Page 10: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Combined Linear Congruential:

See book for combining linear congruential methods to produce random number streams with large cycles / periods.

ETM 607 – Random Number and Random Variates

Page 11: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Uniformity Tests :

Null hypothesis, H0: Ri ~ U[0,1) and H1: Ri ~ U[0,1)

Level of significance, = P(reject H0 | H0 true) probability of rejecting the null hypothesis when in null hypothesis is true.

Usually set = .05 or .01, or probability is 5% or 1% of rejecting null hypothesis when performing the test.

ETM 607 – Random Number and Random Variates

/

Page 12: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Uniformity Tests – Kolmogorov-Smirnov:Step 1 – rank data from smallest to largest Ri :

Step 2 - compute

Step 3 – compute D

Step 4 – Use Kolmogorov-Smirnov table A.8, selecting column associated with significance level, and row where N is the number of observations.

Step 5 – If sample statistic D from step 3 is greater than Dthe null hypothesis is rejected. If D <= Dcannot detect difference between random numbers and the uniform distribution.

ETM 607 – Random Number and Random Variates

][]3[]2[]1[ ..... NRRRR

N

iRD

RN

iD

iNi

iNi

1max

max

)(1

)(1

},max{ DD

Page 13: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Uniformity Tests – Kolmogorov-Smirnov:Excellent example in book, Ex. 7.6

insert Ex 7.6

ETM 607 – Random Number and Random Variates

Page 14: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Uniformity Tests – Kolmogorov-Smirnov:Excellent example in book, Ex. 7.6

insert Fig 7.2

ETM 607 – Random Number and Random Variates

Page 15: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Uniformity Tests – Chi-Square Test:

Where Oi is the number of observations within a segment/range/cell, Ei is the expected number of observation in the segment/range, and n is the number of segments/ranges/cells.

For a uniform distribution that has n segments or ranges,

ETM 607 – Random Number and Random Variates

n

NEi

n

ii

ii

E

EO

0

220

)(

Page 16: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Uniformity Tests – Chi-Square Test:

Insert ex 7.7

ETM 607 – Random Number and Random Variates

Page 17: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Independence Tests – Autocorrelation:

Recall correlation (r)

r = .334437

r = -1.0

r = 1.0

ETM 607 – Random Number and Random Variates

Page 18: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Independence Tests – Autocorrelation:Autocorrelation is correlation of a series of data to help identify repeated patterns. Objective is to have autocorrelation values near 0 for all lags.

ETM 607 – Random Number and Random Variates

Lag = 1, r = .171282

Lag = 2, r = -.16459

Lag = 3, r = -.51783

Lag = 5, r = .337725

Time series plot of 7.4.2 data

Lag is the interval betweenplotted vales.

Page 19: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Independence Tests – Autocorrelation:See book for statistical method of applying hypothesis testing for the objective of 0 correlation at various time lags.

ETM 607 – Random Number and Random Variates

Page 20: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Random-Variate Generation – Chapter 8:

Random-Variate generation is converting from a random number (Ri) to a Random Variable, Xi ~ some distribution.

Inverse transform method:Step 1 – compute cdf of the desired random variable XStep 2 – Set F(X) = R where R is a random number ~U[0,1)Step 3 – Solve F(X) = R for X in terms of R. X = F-1(R).Step 4 – Generate random numbers Ri and compute desired random variates:

Xi = F-1(Ri)

ETM 607 – Random Number and Random Variates

Page 21: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Inverse transform method – Uniform Distibution Example:Step 1 – compute cdf of the desired random variable X

Step 2 – Set F(X) = R where R is a random number ~U[0,1)

Step 3 – Solve F(X) = R for X in terms of R. X = F-1(R).

Step 4 – Generate random numbers Ri and compute desired random variates:

Xi = Ri(b-a) + a

ETM 607 – Random Number and Random Variates

bx

bxaab

axax

xF

,1

,

,0

)(

ab

axRxF

)(

aabRXaXabR )(,)(

Page 22: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Inverse transform method – Uniform Distibution Example:

Xi = F-1 (R) = Ri(b-a) + a

If Xi ~ U[5,10)a = 5b =10

Ri Xi

.5 .5(10 - 5)+5 = 7.5

.7 .7(10 – 5) + 5 = 8.5

.1 .1(10 – 5) + 5 = 5.5

ETM 607 – Random Number and Random Variates

Page 23: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

Inverse transform method – General Idea:

Mapping (or transforming) from cdf to Random Variable

Insert fig 8.2

ETM 607 – Random Number and Random Variates

Page 24: ETM 607 – Random Number and Random Variates Define random numbers and.pseudo-random numbers Generation of random numbers Test for random numbers - Frequency.

In class exercise:

Determine the inverse function F-1 for the triangular distribution.

If, a = 5b = 7c = 10

Find X when R = .75.

ETM 607 – Random Number and Random Variates

cx

cxbacbc

xc

bxaacab

axax

xF

otherwise

cxbacbc

xc

bxaacab

ax

xf

,1

,))((

)(1

,))((

)(,0

)(

,0

,))((

)(2

,))((

)(2

)(

2

2