1
Random Variate
Generation
Presented by – Ritu Raj
2
1. Radom Variate
2. Inverse-transform Technique
Overview 3. Acceptance-rejection Technique
4. Composition Method
5. Convolution Method
6. Summary
SSE Random Variate Generation Ritu Raj
3
Random Variate
A random variate is a variable generated from uniformly distributed
pseudorandom numbers. Depending on how they are generated, a
random variate can be uniformly or nonuniformly distributed.
“In the mathematical fields of probability and statistics, a random variate is
a particular outcome of a random variable”.
Random variates are frequently used as the input to simulation models.
SSE Random Variate Generation Ritu Raj
4
Preparation
It is assumed that a source of uniform [0,1] random numbers exists.
• Linear Congruential Method (LCM)
Random numbers 𝑅, 𝑅1 , 𝑅2 , … with
• PDF
0 0≤𝑥≤1
𝑓𝑅 𝑥 = ቊ
1 otherwise
• CDF
0 x<1
𝐹𝑅 𝑥 = ቐ x 0≤𝑥≤1 fig.1 pdf fig.2 cdf
1 𝑥>1
SSE Random Variate Generation Ritu Raj
5
Inverse-transform Technique
The concept:
• For CDF function: r = F(x)
• Generate r from uniform (0,1), a.k.a U(0,1)
• Find x, x = 𝐹 −1 (r)
F(x) F(x)
Figure3 random variate [1]
SSE Random Variate Generation Ritu Raj
6
Inverse-transform Technique
The inverse-transform technique can be used in principle for any
distribution.
Most useful when the CDF F(x) has an inverse 𝐹 −1 (x), which is easy to
compute.
Required steps:
1. Compute the CDF of the desired random variable X
2. Set F(X) = R on the range of X
3. Solve the equation F(X) = R for X in terms of R
4. Generate uniform random numbers 𝑅1 , 𝑅2 , 𝑅3 , ... and compute the
desired random variate by 𝑥𝑖 = 𝐹 −1 𝑅𝑖
SSE Random Variate Generation Ritu Raj
7
Applications of the Inverse
Transformation Technique
Table 1 Application of Inverse Transformation Technique[2]
SSE Random Variate Generation Ritu Raj
8
Acceptance-Rejection Technique
Useful particularly when inverse CDF does not exist in closed form.
Can be used if a pdf g(x) exists such that c g(x) majorizes the
pdf f(x) ⇒ c g(x) > f(x) ∀ x
Required steps :
1. Generate x with pdf g(x).
2. Generate y uniform on [0, cg(x)].
3. If y < f(x), then output x and return.
Otherwise, repeat from step 1.
4. Continue rejecting the random variates x and y until y > f(x)
SSE Random Variate Generation Ritu Raj
9
g is a Nice" Majorizing function,
can say Uniform.
Throw darts at rectangle under g
until hit f
Prob {Accept X} is proportional
to height of f (X) - called trial
ratio
Figure4 Accept-Rejection
SSE Random Variate Generation Ritu Raj
10
Composition Method
Can be used if CDF F(x) = Weighted sum of n other CDFs.
𝑛
𝐹 𝑥 = 𝒑𝑖 𝐹𝑖 𝑥
𝑖=1
Here, 𝒑𝑖 ≥ 𝟎, σ𝑛𝑖=1 𝒑𝑖 = 𝟏 𝒂𝒏𝒅 𝐹𝑖′𝑠 are distribution functions.
n CDFs are composed together to form the desired CDF
Hence, the name of the technique.
The desired CDF is decomposed into several other CDFs
⇒ Also called decomposition.
Can also be used if the pdf f(x) is a weighted sum of n other
pdfs: f 𝑥 = σ𝑛𝑖=1 𝒑𝑖 𝒇𝑖 𝑥
SSE Random Variate Generation Ritu Raj
11
Composition Method
Required steps :
1. Generate a random integer I such that:
𝑃 𝐼 = i = 𝑝𝑖
2. This can easily be done using the inverse-transformation method.
3. Generate x with the 𝑖 𝑡ℎ pdf 𝒇𝑖 𝑥 and return.
SSE Random Variate Generation Ritu Raj
12
Convolution Method
Convolution
• The sum of independent random variables
o Sum of n variables: 𝑥 = 𝑦1 + 𝑦2 + 𝑦3 + ⋯ + 𝑦𝑛
o Generate n random variate 𝒚𝑖 's and sum
o For sums of two variables, pdf of x = convolution of
pdfs of 𝑦1 and 𝑦2 . Hence the name
o Although no convolution in generation
SSE Random Variate Generation Ritu Raj
13
Convolution Method
o If, pdf or CDF = Sum ⇒ Composition
o Variable x = Sum ⇒ Convolution
𝒇 ∗ 𝒈 𝒕 = න 𝒇 𝝉 𝒈 𝒕 − 𝝉 𝒅𝝉
−∞
Can be applied to obtain
• Erlang variates
• Binomial variates
SSE Random Variate Generation Ritu Raj
14
Summary
Is CDF Yes
invertible?
Use inversion
Is CDF a Yes
sum of other Use composition
CDFs?
SSE Random Variate Generation Ritu Raj
15
Is pdf a sum
Yes
of other Use Composition
pdfs?
Is the
variate a Yes
sum of Use convolution
other
Variates?
Is the
Variate Yes
related to Use characterization
other
variates?
SSE Random Variate Generation Ritu Raj
16
Continue...
Does
a majorizing Yes
function
Use rejection
exist?
NO
Use empirical inversion
SSE Random Variate Generation Ritu Raj
17
References
[1] Modeling and Performance Analysis with Simulation
Available: https://www.mi.fu-berlin.de/inf/groups/ag-
tech/teaching/2012_SS/L_19540_Modeling_and_Performance_Analysis_with_Simulation/index.html
[2] Random Variate Generation
Available: https://www.cse.wustl.edu/~jain/cse567-08/k_28rvg.htm
Reading References:
1. Introduction to Discrete Event Systems by Christos G. Cassandras and Stéphane Lafortune
2. Wikipedia: https://en.wikipedia.org/wiki/Random_variate
3. Simulation CSCI 6637: https://www.eg.bucknell.edu/~xmeng/Course/CS6337/Note/master/
18
