1
$\begingroup$

Prove summation by parts:

$$ \sum_{n=1}^N a_n b_n = a_N B_N - \sum_{n=1}^{N-1}(a_{n+1} - a_n) B_n $$ where $B_N$ means $\sum_{n=1}^N b_n$ and $B_0 = 0$.


The proof is via induction. The base case is trivial. Then, given the identity up to $N$, we write:

$$ a_{N+1}B_{N+1} - a_{N+1} B_N = a_{N+1} b_{N+1} \\ a_{N+1}B_{N+1} - a_N B_N - a_{N+1} B_N + a_N B_N = a_{N+1} b_{N+1} \\ a_{N+1}B_{N+1} - a_N B_N - (a_{N+1} - a_N)B_N = a_{N+1} b_{N+1} \\ a_{N+1}B_{N+1} - a_N B_N - [\sum_{n=1}^{N}(a_{n+1} - a_n) B_n - \sum_{n=1}^{N-1}(a_{n+1}-a_n)B_n] = a_{N+1} b_{N+1} \\ a_{N+1}B_{N+1} - \sum_{n=1}^{N}(a_{n+1} - a_n) B_n - [a_N B_N - \sum_{n=1}^{N-1}(a_{n+1}-a_n)B_n] = a_{N+1} b_{N+1} \\ a_{N+1}B_{N+1} - \sum_{n=1}^{N}(a_{n+1} - a_n) B_n = a_{N+1} b_{N+1} + \sum_{n=1}^N a_n b_n\\ \\ $$

QED.


Questions:

A. Is my proof correct? Well-written? Please share your feedback and recommendations, both for the approach and the writing.

B. How would you set out to prove this? My sequence was:

  1. Think how I would express $\sum a_nb_n$ in terms of $B_N$
  2. Realize that $a_3B_3 = a_3(b_1 + b_2 + b3) + a correction factor$, and that correction factor would telescope
  3. This turned the identity to be proven into something that actually made sense intuitively
  4. I realized the proof would then follow from induction
  5. The induction was difficult, due to the complexity of the terms and subscripts; I eventually painstakingly worked it backwards (from the desired difference towards the original identity), and then rewrote it forwards. My writeup is a bit verbose, but I found that without that, I myself couldn't keep all the terms and subscripts straight

C. Finally: If you were, as a student, to read this proof, how would you reverse engineer the process used to produce it, so that you could learn to do so on your own? Too often, I read proofs, have no issue with them, but seem like they were pulled out of a hat.

$\endgroup$
2
  • $\begingroup$ Your definition of $B_n$ does not make sense because it uses $n$ in two different ways. Also, I think a simpler approach than induction would be to expand the double sums and simplify. $\endgroup$
    – RobPratt
    Commented Oct 14, 2022 at 4:04
  • 1
    $\begingroup$ As Rob points out, $B_n$ in your definition should be $B_N$. Also, you don't use $B_0$ anywhere, so there is no reason to define it. I suggest writing the identity as $$\sum_{n=1}^N a_n b_n = a_N B_N + \sum_{n=1}^{N-1}(a_n - a_{n+1}) B_n$$ and proving by rearranging the sum as $$\left[\sum_{n=2}^N a_n (B_n - B_{n-1})\right] + a_1B_1$$ $\endgroup$ Commented Oct 14, 2022 at 19:14

0

You must log in to answer this question.