1
$\begingroup$

I have a known distribution $f(x)$ (in fact, I can safely assume that $f(x)$ is the Maxwell-Boltzmann distribution, i.e. $f(x)\propto x^2 \exp(-x^2)$). I take $N$ samples from the distribution, but am only interested in a subset of these samples: samples that are larger than $x_0$ (a fixed and known value). I'll denote the number of samples that satisfy this condition $M$, and the subset itself I'll denote $X_{x_0}$. For each of the $M$ samples in $X_{x_0}$, I perform a Bernoulli trial with a constant probability $p$. What I'm interested to know is the variance of the number of successful trials. So, my two questions are:

  1. Am I correct in thinking that the distribution of the number $M$ (number of samples larger than $x_0$) is binomial: $f(M) = B(N,p_{x_0})$, where $p_{x_0} = \int_{x_0}^{\infty} f(x)dx$? (I assume I'm correct, but I'm never sure when it comes to probability theory and statistics).
  2. Second, and more important question: is there an expression of the variance of the number of successful trials? Or perhaps it's possible to derive a closed-form expression for the distribution of the number of successful trials? So I assume it would depend on three parameters: number of initial samples $N$, probability $p_{x_0}$ of a sample exceeding the treshold value $x_0$, and probability $p$ of a successful Bernoulli trial.

Thanks!

$\endgroup$
1
$\begingroup$

For each $x_i$ of the $N$ samples you perform simultaneously two checks: Is $x_i$ larger than $x_0$ (with probability $p_{x_0}$ and then your Bernoulli trial with probability $p$. Assuming independence of this trial and the check $x_i > x_0$ the probability is $p \cdot p_{x_0}$ that you have a successful trial. Thus (assuming independence of the particles) $B(N,p \cdot p_{x_0})$ is the distribution of successful trials with expectation $N \cdot p \cdot p_{x_0}$ and variance $N \cdot p \cdot p_{x_0} \cdot (1- p \cdot p_{x_0})$.

$\endgroup$
1
  • $\begingroup$ Wow, I didn't think of the "just multiply the probabilities" thing! Thank you so much! $\endgroup$
    – Kunstmord
    Apr 3 at 23:36

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.