Chapter 3

RANDOM VARIATE GENERATION

In order to do a Monte Carlo simulation either by hand or by computer, techniques must be

developed for generating values of random variables having known distributions. These values
are often called random variates. As was the case in the drive-in window example above, the
starting place for random variate generation is usually the generation of random numbers, which
are random variates that are uniformly distributed on the interval from 0 to 1 (uniform [0, 1]).

3.1 Inverse Transform Method

The “table look-up” technique that we used earlier may be used whenever the simulation is
being done by hand or by using a spreadsheet. It is also used in some simulation languages due
to the speed with which it can be implemented on a digital computer. One of its disadvantages is
that it takes a great deal of effort to implement on a computer, and thus is seldom used whenever
the simulation is written in a high level language such as C or FORTRAN. The procedure
basically uses the inverse of the cumulative distribution function F given in a table rather than a
formula. A random number, 𝑅, is generated and then the (inverse) value of 𝑋 that would give 𝑅
as the value of 𝐹(𝑋) is determined. Depending on the desired accuracy, linear interpolation may
be used.
As an example, suppose we wanted to generate values having a standard normal distribution
(mean 0 and standard deviation 1) using a spreadsheet. We first must decide the accuracy
desired. Suppose that 1 decimal place is good enough. Next we would generate the values of 𝑋 in
the proper range, say –3 to 3. We would want to put these in a column with a blank column to
the left. In the left hand column, we would then put the formula for the cumulative standard
normal distribution (changing the value in the first cell to zero and the last one to one). Since
spreadsheet has a built in function for this distribution, we would use the syntax
=NORMSDIST(B2), if the column is A and begins in the second row. The B2 refers to the cell
containing 𝑋, the value to be evaluated. For example, part of the table is given below in Table
3.1.

3-1
3-2 RANDOM VARIATE GENERATION

Table 3.1: Standard Normal Distribution Table

If a column of random numbers is generated, then the vertical look-up function can be used
to generate the values of a random variate having the standard normal distribution. This
technique was used to generate 100 values of this random variate. A histogram of the results is
given in Figure 3.1.

Figure 3. 1 Histogram of 100 Normally Distributed Random Variates

Values of a normally distributed random variate, 𝑋, having mean 𝜇 and standard deviation 𝜎
may be found by using the usual transformation
𝑋−𝜇
𝑍= , i.e.  𝑋 = 𝜇 + 𝜎𝑍 ,
𝜎
SIMULATION NOTES 3-3

where 𝑍 has a standard normal distribution or, in the case of a spreadsheet, by using the
cumulative normal distribution function (as above) with mean 𝜇 and standard deviation 𝜎.
Placing the column for 𝑋 to the right of the column for 𝑍 in the above procedure effectively
produces the inverse of the cumulative distribution function. Whenever the inverse of the
cumulative distribution function for the random variate to be generated can be found in closed
form, is available in the software used, or can be given in a table, the inverse function
technique may be used. This technique is based on the observation that if the cumulative
distribution function for 𝑋 is 𝐹, if 𝐹 is continuous and one-to-one, and if 𝑅 is a random variable
that has a uniform[0, 1] distribution, then 𝐹 !! (𝑅) has the same distribution as 𝑋. Figure 3.2
below illustrates the technique, which is also called the inverse transform method.
Suppose we want to generate random variates having a uniform[𝑎, 𝑏] distribution. The linear
function ℎ given by ℎ(𝑟) = 𝑎 + (𝑏 − 𝑎)𝑟 would map the interval [0, 1] onto the interval [𝑎, 𝑏].
A possible way to generate a value for 𝑋 would be to generate a random number 𝑅 then to use
ℎ(𝑅) = 𝑎 + (𝑏 − 𝑎)𝑅   as the value. The inverse transform technique, illustrated below, should
give the same generator.
Figure 3.2: Inverse Transform Method

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1

-6 -4 -2 0 2 4 6

The probability density function (pdf), 𝑓, and the cumulative distribution function (cdf), 𝐹,
for the uniform[𝑎, 𝑏] random variable, 𝑋, are given by
0 if  𝑥 < 𝑎,
1
if  𝑎 ≤ 𝑥 ≤ 𝑏, 𝑥−𝑎
𝑓 𝑥 = 𝑏−𝑎      and      𝐹 𝑥 = if  𝑎 ≤ 𝑥 ≤ 𝑏,
𝑏−𝑎
0 otherwise,
1 if  𝑥 > 𝑏.
Solving 𝑦 = (𝑥 − 𝑎)/(𝑏 − 𝑎) for 𝑥 in terms of 𝑦 yields

𝑥 = 𝑎 + 𝑏 − 𝑎 𝑦.
3-4 RANDOM VARIATE GENERATION

Thus the generator given by the inverse function technique would be

𝑋 = 𝑎 + 𝑏 − 𝑎 𝑅,

where 𝑅 is a random number, and is the same as was obtained above using a linear mapping of
the interval [0, 1] onto the interval [𝑎, 𝑏].
Another useful random variable generator that can be obtained using the inverse transform
method is the one for exponentially distributed random variables. One is needed whenever a
simulation of a Poisson process is to be done, since the time between occurrences of a Poisson
process has an exponential distribution. Let 𝑋 be a random variable that has an exponential
distribution with mean 𝛼 = 1/𝜆 (𝜆 is called the rate parameter). Then the cdf of 𝑋 is given by
!! !
𝐹 𝑥 = 1−𝑒 if  𝑥 ≥ 0, = 1 − 𝑒 !!" if  𝑥 ≥ 0,
0 otherwise 0 otherwise.

Solving 𝑦 = 1 − 𝑒 !" for 𝑥 in terms of 𝑦 yields

ln 1 − 𝑦
𝑥=− .
𝜆
Thus a random variable generator for 𝑋 is
ln 1 − 𝑅
𝑋=− = −𝛼 ln 1 − 𝑅 ,
𝜆
whenever 𝑋 is distributed exponentially with mean 𝛼 = 1/𝜆. Note that since 𝑅 ≤ 1, ln(𝑅) ≤ 0.
Thus – ln(𝑅) ≥ 0, and 𝑋 ≥ 0 as it should be.
Consider the random variable with pdf given by
𝑥
if  0 ≤ 𝑥 ≤ 1,
2
𝑓 𝑥 = 3 if  1 < 𝑥 ≤ 2,
4

0 otherwise.
! ! !
This is a pdf since 𝑓 𝑥 ≥ 0 for all 𝑥 and !! 𝑓(𝑥) 𝑑𝑥 = ! 𝑥/2   𝑑𝑥 + !
3/4 𝑑𝑥 = 1. First we
must determine the cdf 𝐹. If 𝑥 < 0, then 𝐹(𝑥) = 0. If 0 ≤ 𝑥 ≤ 1, then
! ! ! !
𝑡 𝑡! 𝑥!
𝐹 𝑥 = 𝑓 𝑡 𝑑𝑡 = 0  𝑑𝑡 +  𝑑𝑡 = 0 + = .
!! !! ! 2 4 !
4
SIMULATION NOTES 3-5

If 1 ≤ 𝑥 ≤ 2, then
! ! !!
𝑡 3
𝐹 𝑥 = 𝑓 𝑡 𝑑𝑡 = 0  𝑑𝑡 + 𝑑𝑡 + 𝑑𝑡
!! !! ! 2 ! 4
1 3 ! 1 3 3𝑥 − 2
=0+ + 𝑡 = + 𝑥−1 =  
4 4 ! 4 4 4

Check that 𝐹 1 = (3 1 − 2)/4 = 1/4 and 𝐹 2 = (3 2 − 2)/4 = 1, and so the values at the
endpoints match. In summary,

0 if  𝑥 < 0,

𝑥!
if  0 ≤ 𝑥 ≤ 1,
4
𝐹 𝑥 =
3𝑥 − 2
if  1 < 𝑥 ≤ 2,
4

1 if  𝑥 > 2.
We restrict the domain of 𝐹 to [0, 2] to obtain a one-to-one, and therefore, invertible, function.
Let 𝑅 = 𝐹(𝑥). If 0 ≤ 𝑅 ≤ 1/4, then it must be the image of some 𝑥 between 0 and 1. If
1/4 < 𝑅 ≤ 1, then it must be the image of some x between 1 and 2. Hence, if 0 ≤ 𝑅 ≤ 1/4, then
𝑅 = 𝑥 ! /4. Solving for 𝑥 gives 𝑥 = ± 4𝑅 = ±2 𝑅 and we take the + since we know 𝑥 lies
between 0 and 1. If 1/4 < 𝑅 ≤ 1, then 𝑅 = (3𝑥 − 2)/4. Solving for 𝑥 gives 𝑥 = (4𝑅 + 2)/3.
Thus our random variate generator is

2 𝑅 if  0 ≤ 𝑅 ≤ 1/4,
𝑋 = 4𝑅 + 2
if  1/4 < 𝑅 ≤ 1.
3
3.2 Using Excel’s Functions
There are several probability functions whose inverses are built into Excel. Thus they can be
used with the inverse transform method to generate random variates. Two of the most used are
the inverse of the Normal distribution and the inverse of the Gamma distribution. Care must be
taken when using an inverse function in Excel because the function is not always the
(mathematical) inverse of the cumulative distribution function.
The Normal distribution takes two parameters, the mean and the standard deviation of the
random variable. The Excel Help display for the function is shown in Figure 3.3. The format for
the function call is NORMINV(random_number, mean, standard_deviation). Thus to
generate a random variate having mean 2 and standard deviation 0.5, we would enter
=NORMINV(RAND(),2,0.5)
3-6 RANDOM VARIATE GENERATION

Figure 3.3: Excel Help on NORMINV

in the Excel cell where we wish the value to appear. It is good spreadsheet practice to never use
specific parameters in formulas, but to give each parameter its own cell and use cell references.
Thus we should enter the formula as shown in Figure 3.4. Notice that the $ in the formulas that
fixes the references to the parameter cells.

Figure 3.4: Generating Normal Random Variates

SIMULATION NOTES 3-7

The Gamma distribution also requires two parameters. These two parameters are not as well
known as the ones for the Normal distribution. The two parameters are usually denoted by 𝛼 and
𝛽. They are related to the mean 𝜇 and the standard deviation 𝜎 of the distribution by the
following formulas:
𝜇 = 𝛼𝛽        and        𝜎 ! = 𝛼𝛽 ! .

If 𝛼 = 1, we have the exponential distribution with 𝜆 = 1/𝛽. If 𝛼 is a positive integer, the

distribution is called an Erlang distribution. The Gamma distribution is often used to model
waiting time distributions. See Appendix B for a discussion of the relationship between the
exponential distribution and the time between occurrences of a random phenomenon.
The Excel help screen for the GAMMAINV function is shown in Figure 3.5. Thus to
generate a random variate having parameters 𝛼 and 𝛽, we enter =GAMMAINV(RAND(), α, β)
in the Excel cell where we wish the value to appear. An example is shown in Figure 3.6.

Figure 3.5: Excel Help on GAMMAINV

3.3 Chi-square Goodness-of-fit

As with a random number generator, a random variate generator should produce values that
satisfy statistical tests that indicate whether or not the values generated are from the desired
distribution. The first test we will use is a Chi-square test, similar to the one used in Chapter 2.
Differences arise because the expected number of occurrences in a subinterval may be so small
that the resulting calculations are skewed. For this reason, we will require that the expected
number of occurrences in a subinterval be at least 5 (or close to it).
To illustrate, 100 values were generated from a Normal random variate using the technique
illustrated in Figure 3.4. The histogram tool was used to count the number of occurrences in each
of 10 subintervals which were determined by the tool. The results are shown in Figure 3.7.
3-8 RANDOM VARIATE GENERATION

Figure 3.6: Generating Gamma Random Variates

Figure 3.7: Frequency Observed Normal Distribution

We need to determine the probability of a value lying in each of the intervals. To do this, we
make use of Excel’s NORMDIST function, as shown in Figure 3.8, to find the cumulative
probability of the random variable being less than the bin value. These values are in Column G in
Figure 3.8. The probability of the random variable being in the subinterval is then found by
subtracting the cumulative value of the left end point from the cumulative value of the right end
point. These values are in column H in Figure 3.8. The expected number of values in a
subinterval may then be calculated by multiplying the probability by the total number of
observations (column I).
SIMULATION NOTES 3-9

Figure 3.8: Expected Calculations

Since the expected numbers of values in the first three cells are less than 5, pooling is
required. In order to get the required minimum of 5, we pool the first three cells with the fourth.
For the same reasons, we pool the last four cells. The same cells are pooled in the observed
(frequency) data. The results are shown in Figure 3.9. Observe that all the expected values are
now greater than 5.
The calculation of the Chi-square statistic proceeds as in Chapter 2. The degrees of freedom
parameter is calculated as the number of cells minus one minus the number of parameters
estimated. In this case, we estimated no parameters, since we knew the mean and standard
deviation of the population at the outset. Since we ended with 6 cells, the degree of freedom is 5.
The results are shown in Figure 3.10. We accept the null hypothesis that the data was drawn
from a Normal distribution with a p-value of 0.896.
In a more general setting, when the mean and standard deviation are estimated from the data,
the degrees of freedom would be 6 – 1 – 2 = 3. This would make our critical value at 90%
confidence (𝛼 = 0.10) equal to 6.25 and our p-value equal to 0.65. Thus we would still have
accepted the null hypothesis that the data came from the Normal distribution with mean 2 and
standard deviation 1/2.
3-10 RANDOM VARIATE GENERATION

Figure 3.9: Pooling Results

Figure 3.10: Chi-square Calculations

SIMULATION NOTES 3-11

3.4 Using Random Numbers to Estimate Area

One of the applications of random variate generation is the estimation of areas and using
!
these to obtain estimates of irrational numbers. For example, recall that ln 2 = ! 1/𝑥  𝑑𝑥 is the
area under the curve 𝑦 = 1/𝑥 between 𝑥 = 1 and 𝑥 = 2. If we generate a large number of pairs
of random numbers (𝑋, 𝑌), where 𝑋 is uniformly distributed on [1, 2] and 𝑌 is uniformly
distributed on [0, 1], then the ratio of the number of pairs lying under the curve, 𝑚, to the total
number of pairs generated, 𝑛, should be approximately the same as the ratio of the area under the
Area !
curve to the total area in the rectangle [1, 2]×[0, 1]. That is, Total  Area ≈ ! . In this example, the
area of the enclosing rectangle, Total Area, is equal to one.
To illustrate, use Excel to generate 30 values of 𝑋 using the generator 𝑋 = 1 + RAND() and
30 values of 𝑌 using the generator 𝑌 = RAND(). See Figure 3.11. The ordered pair (𝑋, 𝑌) will
lie under the curve whenever 𝑌   <  1/𝑋. This is the same as 𝑋𝑌 < 1. Thus we use the formula
=IF(D2<1,1,0) to record 1 whenever the point lies under the curve and 0 whenever the point lies
on or above the curve. The number of points in the desired area is then found by summing the
values in column E.

Figure 3.11: Estimation of 𝐥𝐧 𝟐

With only 30 data points, the estimate of ln 2 may not be accurate enough. Even with 1000
pairs, the accuracy may be only one decimal (see Table 3.1).
3-12 RANDOM VARIATE GENERATION

Table 3.1: Estimation of 𝐥𝐧 𝟐 with 1000 pairs

m= 697
n= 1000
Estimate= 0.697
ln(2)= 0.693147
error= 0.003853

Problems
1. Use a spreadsheet or a computer program to generate 100 values of an exponentially
distributed random variate having mean 0.5. Count the number of values of the random
variate in each subinterval of length 0.25 and obtain a histogram of the results.
2. Develop a random variate generator for a random variable 𝑋 with the pdf
𝑒 !! if  𝑥 < 0,
𝑓 𝑥 =
𝑒 !!! if  𝑥 ≥ 0.
3. Consider a random variable 𝑋 which has pdf

𝑥 if  0 ≤ 𝑥 ≤ 1,
𝑓 𝑥 = 2−𝑥 if  1 ≤ 𝑥 ≤ 2,
0 otherwise.
This distribution is called a triangular distribution with endpoints 0 and 2 and mode at 1.
Develop a random variate generator for this random variable.

4. A squadron of bombers is attempting to destroy an ammunition depot that has a rectangular

shape and is 1000 meters by 500 meters. The bombing run will proceed down the long center
axis of the depot. If a bomb lands anywhere on the depot, a hit is scored. Otherwise, the
bomb is a miss. There are ten bombers in each squadron. A bombing run consists of each of
the 10 bombers dropping their bomb on the depot. The aim point is the center of the depot.
The point of impact is assumed to have a bivariate normal distribution around the aim point
with a standard deviation of 600 meters in the down-range direction and 200 meters in the
cross-range direction. Simulate 100 bombing runs and obtain an estimate of the expected
number of hits on a run. How many bombing runs must be simulated to ensure the estimate is
accurate to two decimals with 90% confidence?

5. Generate 100 random variates having a Normal distribution with mean 5 and standard
deviation 2. Obtain a histogram of the values generated and compare its shape to that
expected of a Normal distribution. Be sure to include the symmetry of the histogram in your
discussion.

6. Generate 100 random variates having a Gamma distribution with mean 5 and standard
SIMULATION NOTES 3-13

deviation 2. Obtain a histogram of the values generated and compare its shape to that
expected of a Gamma distribution. Be sure to include the symmetry of the histogram in your
discussion.

7. Use a Chi-square test to test the goodness-of-fit of the values generated in Problem 1 to an
exponential distribution.

8. Use a Chi-square test to test the goodness-of-fit of the values generated in Problem 5 to a
Normal distribution.

9. Use a Chi-square test to test the goodness-of-fit of the values generated in Problem 6 to a
Gamma distribution.
!
10. Use simulation to approximate ln 3 = !
1/𝑥  𝑑𝑥 . Use 1,000 points in your simulation.

11. Use simulation to approximate the area in the first quadrant that lies inside the circle with
radius 1 and center at the origin. Use your estimate to find an estimate of 𝜋. Use 1,000 points
in your simulation.

12. A large catalog merchandiser is planning to have a special furniture promotion a year from
now. To do this, the company must place its order for the furniture now. It plans to sign a
contract with the manufacturer for 3,000 chairs at a cost of $175 per unit, which the company
plans to offer initially for $250 per unit. The promotion will last for eight weeks, after which
all remaining units will be offered for sale at half the initial price, or $125 per unit. The
company believes that 2,000 units will be sold during the first eight weeks.

a. Based on these assumptions, setup an Excel spreadsheet model and determine the profit
that the company expects from the promotion.

Research suggests that the demand for the chair during the first eight weeks has a normal
(Gaussian) distribution with a mean of 2,000 chairs and standard deviation of 500 chairs. The
number of chairs ordered and the price per unit are known, but the actual selling price of the
chairs can be easily changed. The initial selling price will definitely be between $200 and
$300, but we have no reason to believe that any one value is more likely than another in this
range. Thus we assume the initial selling price of the chairs has a uniform distribution
between $200 and $300.

b. Simulate the special furniture promotion and obtain an estimate of the net profit the
company might receive.

c. Obtain both point and confidence interval estimates of the average profit that the
company might expect from the promotion. Discuss any differences between the
estimates using the continuous random variables (this exercise) and the estimates
obtained earlier using discrete random variables.
3-14 RANDOM VARIATE GENERATION

d. Next we wish to give a graphical (histogram) presentation of the estimated probabilities

for possible net profits the company may derive from the special furniture promotion.
First, determine the largest and smallest values that your simulation produced. Select a
nice round number smaller and one larger than your values and generate 10-15 values
between these numbers for use in the graph.

e. Finally we want to estimate the probability 𝑝, that the company will lose money on the
promotion. To obtain the point estimate, we simply divide the number of simulated
values that are negative by the total number. This gives us one estimate of the probability.
Now use Excel’s Table to generate at least 30 estimates of 𝑝. The average of these values
is 𝑝̂, the point estimate of  𝑝. The formula for a confidence interval estimate of a
proportion can be found in any introductory statistics text or on line at Wikipedia under
“Binominal proportion confidence interval” as

𝑝 1−𝑝
𝑝 ± 𝑧!!! !
𝑛

where 𝑝 is the proportion of successes in a Bernoulli trial process estimated from the
!
statistical sample, 𝑧!!! ! is the (1 − ! ) percentile of a standard normal distribution, and
𝑛 is the sample size.

Partial Solutions
ln(2𝑅) /2 if  0 ≤ 𝑅 ≤ 1/2,
2. 𝑋 =
−ln  (2 1 − 𝑅 )/2 if  1/2 < 𝑅 ≤ 1.

2𝑅 if  0 ≤ 𝑅 ≤ 1,
3. 𝑋 =
2− 2 1−𝑅 if  1 < 𝑅 ≤ 2.