I have

$$\sum_k^n P_k x_k$$

Am I allowed to split it up into two sums so I have it like

$$\sum_k^n P_k \sum_k^nx_k$$

  • 2
    $\begingroup$ No. But since $P_{n}$ doesn't depend on the sum (of $k$), you can take it out, and it becomes $P_{n}\sum_{k=1}^{n}x_{k}$. $\endgroup$ – Vinícius Novelli Oct 5 '14 at 3:41
  • 1
    $\begingroup$ In $\sum_k^nP_nx_k$, $P_n$ doesn't depend on $k$ so can be factored out. $\endgroup$ – Kim Jong Un Oct 5 '14 at 3:41
  • $\begingroup$ Assuming that $P_n$ is a constant with respect to $k$, the above comments are true. $\endgroup$ – Squirtle Oct 5 '14 at 3:48
  • 3
    $\begingroup$ $2 = 1\cdot 1 + 1\cdot 1 \neq (1+1)\cdot(1+1) = 4$. $\endgroup$ – Slade Oct 5 '14 at 6:18


$$x_1\qquad\qquad\qquad\qquad x_2\qquad\qquad\qquad\qquad x_3\qquad\qquad\qquad\qquad x_4$$

Imagine that the vertical segments are your $P$'s. Then the product of sums is the area of the entire rectangle, whereas the sum of products is only the red area.


Assuming P depends on k and not n then the summation $$ \sum_k P_k x_k $$ Creates terms like $$ P_1 x_1 + P_2 x_2 + \ldots + P_n x_n $$ But, if you took $P_k$ out then you would have two independent sums: $$ \sum_k P_k \sum_k x_k $$ Producing two separate sums multiplying each other like $$ (P_1 + P_2 + \ldots + P_n) (x_1 + x_2 + \ldots + x_n) $$ And this obviously produces terms like $P_1 x_1, P_1 x_2,\ldots$ or as produced by the double sum written like: $$ \sum_i \sum_j P_i x_j $$ So the original summation is definitely not the same as this double sum.



^^^That offers a nice summary of summation notation and its properties.

To answer your question specifically, you should not split this into two different summations. Rather, notice that P does not depend on k. You can treat P as a constant and simply move it outside the summation so that it multiples the result of your summation of $x_k$, which does depend on your variable k.

  • $\begingroup$ actually I made a typo P depends on k and n. I ask because I know that the the sum with the $P_nk is always equal to 1 for all n and k. But thanks for the response, it does make sense with what I wrote! $\endgroup$ – Jeff Oct 5 '14 at 4:09
  • $\begingroup$ That link answers my question though, I appreciate it thanks $\endgroup$ – Jeff Oct 5 '14 at 4:10

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.