1

$X_n$ are r.v.s, is it true that $E[\sum_{n=1}^{\infty} X_n] = \sum_{n=1}^{\infty} E[X_n] $?

My feeling is that this is not necessarily true. But cannot come up with an example.

Can someone provide a counterexample or give a proof for this statement?

user184389
  • 11
  • 3

1 Answers1

-1

If $X_n$ are defined on the same probability space then the theorem is true and a proof is demonstrated in this question.

If $X_n$ are defined on different probability spaces then clearly $+$ must be defined on them for the question to have any meaning.

If $X$ is defined as the number of Farmer Brown's sheep in the paddock at 10am and $Y$ as the position of an electron in the CERN supercollider at the same time (for some frame of reference) then clearly it is not meaningful to speak of adding them.

Community
  • 1
Dale M
  • 3,104
  • Dude, it's countably infinite sum, not finite sum – user184389 Oct 20 '14 at 04:39
  • The link you gave is not relevant to this question. It states the result for finite sum of r.v.s – user184389 Oct 20 '14 at 04:39
  • So yea your answer is not correct – user184389 Oct 20 '14 at 04:59
  • Actually it states for 2 random variables, but the sum of 2 random variables is a random variable so it can be applied over and over: that is all you need for any countable set of random variables. The sum may $\to \pm\infty$ but that doesn't matter. – Dale M Oct 20 '14 at 05:05
  • 1
    "it can be applied over and over: that is all you need for any countable set of random variables." No, one needs something more, which is called Fubini theorem. – Did Oct 20 '14 at 23:03