11
$\begingroup$

I am a probability theory beginner. The expression for the variance of a random variable $x$ (of a random process is

$$\sigma^2 = E(x^2) - (\mu_{x})^2$$

If $E(x^2) = (\mu_{x})^2$, then $\sigma^2 = 0$. Can this happen ? Can a random variable have a density function whose variance (the second central moment alone) is $0$ (other than the dirac delta function).

$\endgroup$
  • $\begingroup$ Variance is $0$ if and only if the random variable is constant. $\endgroup$ – voldemort May 8 '14 at 5:27
  • 4
    $\begingroup$ @voldemort more precisely: almost surely constant. $\endgroup$ – Hagen von Eitzen May 8 '14 at 5:53
15
$\begingroup$

The variance $$ E(X^2)-E(X)^2=E(X-E(X))^2 $$ is equal to $0$ if and only if $X$ is equal to $E(X)$ in all of its support. This can only happen if $X$ is equal to some constant with probability $1$ (known as a degenerate distribution).

$\endgroup$
  • $\begingroup$ The information 'in all of its support' was somehow a misleading information to me. Consider $E[X]\ne 0$, both $supp(X)$ and $supp(X)^c$ are not null subsets and define $X=E[X]$ on $supp(X)$, $X=0$ on $supp(X)^c$. Then $Var[X]=E[X-E[X]]^2=E[X]^2\cdot P[supp(X)^c]^2\ne 0$. So it has to hold $X=E[X]$ almost surely or equivalently that $supp(X)^c$ is a null subset. $\endgroup$ – user408858 Nov 4 '18 at 15:40
  • $\begingroup$ It is, of course, $Var[X]=E[(X-E[X])^2]$. So in the formula there should be $Var[X]=E[X]^2\cdot P[supp(X)^c]$. $\endgroup$ – user408858 Nov 4 '18 at 16:49

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.