0
$\begingroup$

Consider $n$ integer-valued and independent random variables $e_1, e_2, \dots, e_n$ with known distribution functions $m_1, m_2, \dots, m_n$. Let's denote with $E^{1..n}$ the random variable given by the sum of $e_1, e_2, \dots e_n$, that is $E^{1..n} = \displaystyle\sum_i{e_i}$. The PDF of $E^{1..n}$, $m^{1..n}$, is given by the convolution of $m_1, m_2, \dots, m_n$.

I would like to calculate $P(E^{1..n} = 1 \,| \,e_1 = 1)$.

My intuition tells that that this is equal to $P(E^{2..n} = 0) P(e_1=1)$ but I just can't figure out the math behind it.

$\endgroup$
  • $\begingroup$ This would be a better place for your question: math.stackexchange.com $\endgroup$ – Dr_Sam Feb 26 '13 at 7:24
  • $\begingroup$ But the answer is no: $P(E^{1..n} = 1 | e_1 = 1)$ = $P(E^{2..n} + e_1 = 1 | e_1 = 1)$ = $P(E^{2..n} = 0 | e_1 = 1)$ = $P(E^{2..n} = 0)$ because of the independance. $\endgroup$ – Dr_Sam Feb 26 '13 at 8:06
1
$\begingroup$

As said in my comments of the question, a better place for asking this kind of question would be on the math part of stackexchange.

However, we can find

$P(E^{1..n}=1|e_1=1) = P(E^{2..n}+e_1=1|e_1=1) = P(E^{2..n}=0|e_1=1) = P(E^{2..n}=0)$,

the last equality being the result of the independance of the $e_i$.

$\endgroup$

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.