Is $f(x_1, x_2, x_3) = f(x_1| x_2, x_3)f(x_2 |x_3)f(x_3)$ equivalent to the standard chain rule of probability $f(x_1, x_2, x_3) = f(x_3 |x_2, x_1)f(x_2|x_1)f(x_1)$?

I ask because $f(x_1, f_2) = f(x_1|x_2)f(x_2) = f(x_2|x_1)f(x_1)$ and thought this may apply to high order cases.

  • $\begingroup$ can you correct your first equation? $\endgroup$ – gunes Feb 15 at 22:51

Based on your second paragraph, I assume you ask if the following is true or not: $$p(x_1|x_2,x_3)p(x_2|x_3)p(x_3)=p(x_3|x_2,x_1)p(x_2|x_1)p(x_1)$$

Yes, it holds. The indices doesn't matter. Call them $x_1=a,x_2=b,x_3=c$ and you won't have a standard form. To be specific, you'll have $n!$ different factorizations if you have $n$ RVs.

The intuition is you can always group RVs as if they're single, e.g. $x_1=a, (x_2,x_3)=b$. When you apply your standard chain rule formula you can end up in any possible factorization.

Note: $p(x),f(x)$ are all abuse of notations. A better notation is to use something like $p_X(x)$, to indicate both the RV and the specific value.

  • $\begingroup$ Thank you very much. I have also corrected the question so it is clearer. $\endgroup$ – InvestingScientist Feb 16 at 9:12

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.