Scribd
Upload
79 views

3 - Notes Triangular Distribution

The document discusses the triangular distribution, including its density function, cumulative distribution function, and how to sample from it using a uniform random variable. It describes how the height of the triangular distribution does not depend on the most likely value, and relates the height to the relative distances from the lower and upper bounds. Homework is assigned to derive the inverse cumulative distribution function and develop a triangular distribution sampler in Excel. The document also generalizes the triangular distribution to have more than three parameters.

Uploaded by

jesus garcia
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
79 views

3 - Notes Triangular Distribution

The document discusses the triangular distribution, including its density function, cumulative distribution function, and how to sample from it using a uniform random variable. It describes how the height of the triangular distribution does not depend on the most likely value, and relates the height to the relative distances from the lower and upper bounds. Homework is assigned to derive the inverse cumulative distribution function and develop a triangular distribution sampler in Excel. The document also generalizes the triangular distribution to have more than three parameters.

Uploaded by

jesus garcia
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

EMGT 273 Discrete System Simulation

Lecture Notes by Instructor: Dr. J. Rene van Dorp Page 1


INTERMEZZO TRIANGULAR DISTRIBUTION

Three Parameters:
Lower Bound: L
Upper Bound: U
Most Likely (or Best Guess): M

Density Function of Triangular Distribution
L M U L
H?
A
1
A
2


QUESTION:
What the value of the height h?

EMGT 273 Discrete System Simulation
Lecture Notes by Instructor: Dr. J. Rene van Dorp Page 2
ANSWER:
Total Area under Density function should be one:

= + 1 ) (
2
1
) (
2
1
H M U H L M


=
)
`

+ 1 ) (
2
1
) (
2
1
M U L M H


L U
H L U H

= =
)
`


2
1 ) (
2
1





EMGT 273 Discrete System Simulation
Lecture Notes by Instructor: Dr. J. Rene van Dorp Page 3
INTERESTING OBSERVATIONS:

H does not depend on M


L U
L M
L U
L M A

=
2
) (
2
1
1
which equals the relative distance
from the lower bound.


L U
M U
L U
M U A

=
2
) (
2
1
2
which equals the relative distance
from the upper bound.

EMGT 273 Discrete System Simulation
Lecture Notes by Instructor: Dr. J. Rene van Dorp Page 4
CUMULATIVE DISTRIBUTION FUNCTION OF TRIANG(L,M,U)

) Pr( ) ( x X x F =

1. Suppose
M x L


L M U L
A
1
A
2
x


2 2
1
( )
x L x L M L
F x A
M L M L U L

| | | |
= =
| |

\ . \ .

EMGT 273 Discrete System Simulation
Lecture Notes by Instructor: Dr. J. Rene van Dorp Page 5
2. Suppose
U x M



L M U L
A
1
A
2
x


L U
M U
M U
x U
A
M U
x U
x F

|
.
|

\
|

=
|
.
|

\
|

=
2
2
2
1 1 ) (



Hence
EMGT 273 Discrete System Simulation
Lecture Notes by Instructor: Dr. J. Rene van Dorp Page 6
2
2
( )
1
x L M L
L x M
M L U L
F x
U x U M
M x U
U M U L


| |

|

\ .
=


| |

|


\ .


From the form of F(x), L X U we can derive the expression for F
-1
(y), 0 X
1.
( )
( )
1
)( 0
( )
(1 ) )( 1
M L
L y U L M L y
U L
F y
M L
U y U L U M y
U L




HOMEWORK: Derive the expression for F
-1
(y). With the form of F
-1
(y) we can
sample from a Triang(L,M,U) distribution via a Uniform Random Variable on [0,1].

EMGT 273 Discrete System Simulation
Lecture Notes by Instructor: Dr. J. Rene van Dorp Page 7
HOMEWORK: Develop a Triangular Distribution Sampler by completing the
template: TriangularTemplate.xls using the RAND function in EXCEL for
Uniform[0,1] Samples.

GENERALIZATION OF TRIANGULAR DISTRIBUTION
( )
1
n
n
x L M L
L x M
M L U L
F x
U x U M
M x L
U M U L


| |

|

\ .
=


| |

|


\ .



( )
( )
1
1
1
)( 0
( )
(1 ) )( 1
n
n
n
n
M L
L y U L M L y
U L
F y
M L
U y U L U M y
U L




EMGT 273 Discrete System Simulation
Lecture Notes by Instructor: Dr. J. Rene van Dorp Page 8
where n > 0. Is a viable alternative to the Beta Distribution (which does not have
a closed form for its cdf!). See, Kotz and Van Dorp (2004) Beyond Beta, Other
Continuous Distributions with Bounded Support and Applications, World
Scientific Publishing Company.

You might also like