Scribd
Upload
29 views

Random Experiments With R

This document provides examples of using the R programming language to simulate random experiments and generate distributions. It demonstrates how to [1] generate lists of random variables like standard normals, [2] create histograms of the results, [3] set the random number seed for reproducibility, [4] simulate the maximum of random variables, and [5] generate exponential distributions using the inverse transform method. The examples illustrate basic probability concepts and how to estimate probabilities from simulated data.

Uploaded by

Siddhartha Siddu
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
29 views

Random Experiments With R

This document provides examples of using the R programming language to simulate random experiments and generate distributions. It demonstrates how to [1] generate lists of random variables like standard normals, [2] create histograms of the results, [3] set the random number seed for reproducibility, [4] simulate the maximum of random variables, and [5] generate exponential distributions using the inverse transform method. The examples illustrate basic probability concepts and how to estimate probabilities from simulated data.

Uploaded by

Siddhartha Siddu
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Mathematics of Finance, Courant Institute, Fall 2015

http://www.math.nyu.edu/faculty/goodman/teaching/mathFin/index.html
Always check the class message board before doing any work on the assignment.

Random Experiments with R


Corrections: (check the class message board)

This exercise uses R to simulate a random experiment and make a histogram


of the results. The sample code ListAndHistogram.R illustrates the features of
R you need for this.

Example 1: The R function rnorm(n) produces a list of n independent stan-


dard normal random variables. Here x is a list and x[k] is element k of the list.
Now we know several kinds of objects in R. If x is an object in R, then x could
be a number, or x could be a string (character string), or x could be a list.
The x here is a list of number, but it is possible to make a list of strings or even
a list of lists. The for loop prints the numbers x. Then, a single cat statement
prints the whole object x, which is a list of n numbers.

Example 2: This creates a list of n standard normals in a more complicated


way. You don’t have to understand this, but if you want to ... . The statement
rnorm(1) first creates a list of one number. Then x k = rnorm(1)[1] gets the
first (and only) number in this list and gives the variable x k that value. The
variable name x k is supposed to remind you of the mathematical notation xk .
The R function c makes a list out of the previous list x and the new element
x k by putting x k on the end. So, if x is the list (4, 6, 9), and xk is 7, then c(
x, x k) is the list (4, 6, 9, 7). The statement x = c(x,x k) results in x being
a list object with value (4, 6, 9, 7). It appends the value of x k to the end of
the list x. The code in Example 2 first creates a list of length 1, then appends
n − 1 random numbers to the list to make a list of length n. It seems more
complicated in this case, but sometimes you don’t know how long a list will be
when you start, so you have to build the list by appending.

Example 3: This is the same as Example 1, except that the seed has been set
to 1. Note that these are not the numbers you got the first two times.

Example 4: This is the same as Example 3, with the same seed. The numbers
should be the same as Example 3.

Example 5: This is the same as Examples 3 and 4, but with a different seed.
The numbers should be different from the Example 3 and Example 4 numbers.

Example 6: Here we create a big list with a lot of independent standard nor-
mal random variables. The R command hist creates a histogram plot. The

1
code gives several arguments to this function, each on its own line for clarity.
The first argument is the numbers, which are in the list x. The second ar-
gument is the R variable breaks, which is the number of bins to use. Using
more bins makes each bin smaller. Experiment with different numbers of bins,
say 10 and 100. The R variable main is what you want printed at the top of
the histogram plot. You can see that we created a string called Title for this
purpose. The R command that creates Title embeds the number of samples
into the text. That’s what sprintf is for. The last argument, probability,
tells R to plot the estimated probability density instead of the bin counts. The
1 2
picture should be approximately f (x) = √12π e− 2 x . To check this, the code
computes and prints f (0) = √12π . Check whether the printed number matches
the plotted value for x = 0. Note that the hist statement appears twice in
(almost) exactly the same form. The first time it makes a popup window on
your screen with the histogram plot. The second time it makes a .pdf file called
NormalHistogram.pdf with the same plot. The titles of the popup and the .pdf
are slightly different. Look for the difference in the R code.

Example 7: This is a probability experiment in R. Suppose Yj are independent


standard normals, and X is the max of L of them. What is the PDF of X?
This code does the probability experiment n times to create n samples of the
X distribution. These numbers Xk are recorded in the R list x. Experiment
with L. For large L the distribution is more narrow, as you can see by looking
at the labels on the x axis. Try, for example, L = 2, L = 10, and L = 100.
For L = 2, it isn’t so unlikely for the maximum to be negative. That’s very
unlikely if L = 10, and nearly impossible for L = 100 (but not mathematically
impossible). The bigger L is, the longer it takes the code to run. If the code is
too slow with your desired L, reduce the number of samples, which is n.

Example 8: This shows how to generate exponential random variables. A


random variable T is exponential with rate constant equal to λ if it has the PDF
λe−λt

if t > 0
f (t) =
0 if t < 0 .
The code generates n independent exponential times, Tk , using the formula
Tk = − λ1 log(Uk ), where the Uk are independent and uniformly distributed in
[0, 1]. You might know why this works, but you can see that it does work by
looking at the histogram. The histogram will become more clear if you increase
n. Try it. The sample code also uses the Tk to estimate Pr(T > t) for two
values of t. The theoretical formula is
Z ∞ Z ∞
0
0 0
Pr(T > t) = f (t )dt = λ e−λt dt0 = e−λt .
t t

The empirical estimate from the data is


# {Tk > t}
Pr(T > t) ≈
n

2
On top on the right is the number of samples Tk so that Tk > t. The code prints
the theoretical and empirical probability for two t values. The agreement is not
great for n = 1000, but it gets better for larger n values.

You might also like