I am working out of Mathematical Statistics and Data Analysis by John Rice and ran into the following interesting problem I'm having trouble figuring out.

Ch 2 (#65)

How could random variables with the following density function be generated from a uniform random number generator?

$$f(x) = \frac{1 + \alpha x}{2}, \quad -1 \leq x \leq 1,\quad -1 \leq \alpha \leq 1$$

So I believe I'm suppose to use the following fact to solve the problem

Proposition D

Let U be uniform on [0, 1], and let X = $F^{-1}$(U). Then the cdf of X is F.


$$P(X \leq x) = P(F^{-1}(U) \leq x) = P(U \leq F(x)) = F(x)$$

That is, we can use uniform random variables to generate other random variables that will have cdf F

So my goal should then be to find a cdf and it's inverse then give as input to the inverse the uniform random variable. I've included my attempt.

Given $f(x) = \frac{1 + \alpha x}{2}$

$$F(X) = \int_{-1}^{x} \frac{1 + \alpha t}{2} dt \; = \; \frac{x}{2} + \frac{\alpha x}{4} + \frac{1}{2} - \frac{\alpha}{4}$$

$$4 \cdot F(X) - 2 + \alpha = 2x + \alpha x$$

$$F^{-1}(X) = \frac{4X - 2 + \alpha}{2 + \alpha}$$

So our random variable is, for example, T where

$$T = F^{-1}(U) = \frac{4U - 2 + \alpha}{2 + \alpha}$$

The answer in the back of the book is

$$X = [-1 + 2 \sqrt{1/4 - \alpha(1/2 - \alpha / 4 - U)}]/ \alpha$$

I'm not really sure where I went wrong. Any help?

  • $\begingroup$ I think I may have spotted a problem with my integration, I'll continue to try and work it out, but any other input is still welcome $\endgroup$ – ApprenticeOfMathematics Jan 4 '16 at 17:17
  • 1
    $\begingroup$ Yes, the antiderivative should have a $t^2$ term. $\endgroup$ – André Nicolas Jan 4 '16 at 17:19
  • $\begingroup$ Yes that was it, I worked it out. The devil's always in the details I guess $\endgroup$ – ApprenticeOfMathematics Jan 4 '16 at 17:32
  • 1
    $\begingroup$ Slips of this kind are universal. The unfortunate thing is that sometimes they lead students who understand something perfectly well to doubt their understanding. $\endgroup$ – André Nicolas Jan 4 '16 at 17:36
  • $\begingroup$ This is known as Inverse Transform Sampling. Its a good technique to generate random numbers from a given density, however its naive in a sense that the CDF must be calculated (which is not always possible or is too hard). The proof to why this works is here:en.wikipedia.org/wiki/Inverse_transform_sampling You should look into rejection sampling techniques to see an alternative way to generate random numbers. $\endgroup$ – Tyler Hilton Jan 4 '16 at 23:37

The cdf appears to be wrong. When $-1\leq x\leq 1$, \begin{align*} F_X(x) &= \int_{-1}^{x} \frac{1 + \alpha t}{2} dt\\ &=\int_{-1}^x \frac{1}{2}+\frac{\alpha}{2}t\,dt\\ &=\frac{1}{2}[x+1]+\frac{\alpha}{4}[x^2-1]\\ \end{align*} Other than that, your approach seems fine.

  • $\begingroup$ Thanks for your help! After completing the square I was able to solve it. I would upvote your answer, but I don't have enough reputation yet. $\endgroup$ – ApprenticeOfMathematics Jan 4 '16 at 17:33
  • $\begingroup$ @ApprenticeOfMathematics Don't worry. It was pointed out in the comments while I was writing the answer, but I didn't see it. I'm glad you got it. $\endgroup$ – Em. Jan 4 '16 at 17:37

Comment: Demonstration in R with $\alpha = .2$ of answerbook result.

 alpha = .2;  m = 10^5;  u = runif(m)
 x = (-1 + 2*sqrt(1/4 + alpha*(1/2 + alpha/4 - u)))/alpha
 hist(x, col="wheat", prob=T)
 curve((1 + alpha*x)/2, -1, 1, lwd=2, col="blue", add=T)

enter image description here

  • 1
    $\begingroup$ Very nice! It's awesome seeing simulations supporting the textbook result $\endgroup$ – ApprenticeOfMathematics Jan 4 '16 at 18:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.