Just a simple notational question: I want to concisely denote sampling $n$ times from a distribution and summing over a function applied to the samples. I am not sure how to do this however. I cannot define a set of $X_i$ as global variables ahead, because the summation occurs in the scope of an outer summation. And each ieration the inner sum samples anew. Currently, I have

$$\sum_k^N\sum_{x \sim X}^n f(x)$$

But this annoys me. It is abuse of notation and makes no sense, as the end index $n$ has no associated running index like $i = 1$ defined below the summation symbol. Is there an exact way to denote this or can I rely on the reader to understand what this notational hack is supposed to mean from the context, where I talk about the formula?


First use some notation to define your sample. Let $X_1, X_2,\ldots,X_n$ be the $n$ members in your sample. If $f$ is the function you're applying to each member, then the sum you are looking for can be written $ \sum_{i=1}^n f(X_i) $.

If you have a different set of $X$'s within each outer iteration, then you could modify your notation: $X_{1,1}, X_{1,2},\ldots,X_{1,n}$ are the items in iteration 1, then $X_{2,1}, X_{2,2},\ldots,X_{2,n}$ are the items on iteration 2, and so on. The overall sum would then be written $\sum_{k=1}^N\sum_{i=1}^n f(X_{k,i})$.

  • $\begingroup$ Hm, that would work for how I described the problem, yes. But actually that is not flexible enough. I cannot define $X_i$ as global variables, as the sum actually occurs in the scope of another sum, and each iteration of the outer sum the inner sum samples again. I did not think of mentioning that complication, sorry. $\endgroup$ – user3578468 Aug 5 at 22:59
  • $\begingroup$ @user3578468 Please see edit. Is this helpful? $\endgroup$ – grand_chat Aug 5 at 23:07
  • $\begingroup$ Yes, that works, thanks! $\endgroup$ – user3578468 Aug 5 at 23:09

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.