Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $X$ and $Y\,$ be independent normal random variables with zero mean. What are the differences between the distributions of

$$A\triangleq \frac{X+Y}{X-Y} \quad\text{ and }\quad B\triangleq \frac{X+Y}{|X-Y\,|}\quad?$$

share|cite|improve this question
Are you assuming equal variances for $X$ and $Y$? If so, $A$ is a Cauchy random variable as discussed here and I strongly suspect that so is $B$. – Dilip Sarwate Nov 15 '11 at 3:27
up vote 2 down vote accepted

1. If the variances of $X$ and $Y$ differ, the distributions of $A$ and $B$ differ

To see this, note that $[B\gt0]=[X+Y\gt0]$ has probability $\frac12$ since the distribution of $X+Y$ is centered normal, hence symmetric. On the other hand, $$ [A\gt0]=[X^2\gt Y^2]=[\sigma^2U^2\gt \tau^2V^2], $$ where $\sigma^2$ is the variance of $X$, $\tau^2$ is the variance of $Y$ and $U$ and $V$ are i.i.d. standard normal random variables. Hence $[A\gt0]=[U^2\gt (\tau^2/\sigma^2)V^2]$ has probability $\lt\frac12$ if $\tau^2\gt \sigma^2$ and $\gt\frac12$ if $\tau^2\lt \sigma^2$, but never exactly $\frac12$.

2. If the variances of $X$ and $Y$ coincide, the distributions of $A$ and $B$ coincide

To see this, note that, when the variances of $X$ and $Y$ coincide, $X+Y$ and $X-Y$ are i.i.d. hence $A$ is distributed as $U/V$ and $B$ as $U/|V|$, where $U$ and $V$ are i.i.d. standard normal random variables. Thus $A=SB$ where $S=\mathrm{sign}(V)$. Now, $S$ is independent on $(U,|V|)$ while $B$ is $(U,|V|)$-measurable, hence $S$ and $B$ are independent. Since the distribution of $B$ is symmetric, this implies that the distributions of $A=SB$ and $B$ coincide.

3. The distributions of $A$ and $B$

In the second case, the distribution of $A$ and $B$ is standard Cauchy. In the first case, introduce the parameter $$ \varrho=\frac{\sigma^2-\tau^2}{\sigma^2+\tau^2}. $$ Then, the same arguments as above show that $A$ and $B$ can be realized as $$ A=\varrho+\sqrt{1-\varrho^2}C,\qquad B=SA, $$ where $S$ and $C$ are independent, $S=\pm1$ is a symmetric sign and $C$ is standard Cauchy. This explains why $A$ and $B$ are i.i.d. if and only if the distribution of $A$ is symmetric if and only if the distribution of $A$ is centered if and only if $\varrho=0$ if and only if $X$ and $Y$ have the same variance.

share|cite|improve this answer

They have the same distribution if $X$ and $Y$ have the same variance. It is not too difficult to see that $X>Y$ implies $A(X,Y) = A(-X,-Y) = B(X,Y) = B(Y,X)$ and $A(Y,X) = A(-Y,-X) = B(-X,-Y) = B(-Y,-X)$ and conversely $X<Y$ implies $A(X,Y) = A(-X,-Y) = B(-X,-Y) = B(-Y,-X)$ and $A(Y,X) = A(-Y,-X) = B(X,Y) = B(Y,X)$. Since $(X,Y), (Y,X), (-X,-Y),$ and $(-Y,-X)$ all have the same probability distribution, $A$ and $B$ have the same distribution.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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