0

What is the product of the following summation with itself:

$$\left(\sum_{k=2}^{n}k(k-1)a_{k}t^{k-2} \right) \left(\sum_{k=2}^{n}k(k-1)a_{k}t^{k-2}\right) $$

Is the above equal to the double summation below and if not, what is the correct proof or derivation? $$\sum_{k=2}^{2n}\left(\sum_{i=2}^{k}k(k-1)a_{i}a_{k-i}\right)t^{k-2}$$

I based my answer on the following rule for the product of two summations. However, it is also not clear how they derived this (i.e. where does the upper bound $2n$ come from in the first summation and the total second summation). What is the exact derivation of this step by step? $$\left(\sum_{k=0}^{n}a_{k} \right) \left(\sum_{k=0}^{n}b_{k}\right) = \sum_{k=0}^{2n}\left(\sum_{i=\max(0,k-n)}^{\min(n,k)}a_{i}b_{k-i}\right)$$

Math98
  • 85
  • 8
  • 1
    Do NOT change your OP. You changed your OP in this post https://math.stackexchange.com/questions/4717312/second-order-partial-derivative-of-s-sum-k-02n-left-sum-i-max0-k-n – MathFail Jun 14 '23 at 10:59

2 Answers2

2

We consider the Cauchy product of two polynomials in $z$.

We obtain \begin{align*} \color{blue}{\left(\sum_{k=0}^n\right.}&\color{blue}{a_kz^k\Bigg)\left(\sum_{l=0}^nb_lz^l\right)}\\ &=\sum_{q=0}^{2n}\left(\sum_{{k+l=q}\atop{0\leq k,l\leq n}}a_kb_l\right)z^q\tag{1}\\ &=\sum_{q=0}^{2n}\left(\sum_{{{k=0}\atop{0\leq k\leq n}}\atop{0\leq q-k\leq n}}^qa_kb_{q-k}\right)z^q\tag{2}\\ &=\sum_{q=0}^{n}\left(\sum_{k=0}^qa_kb_{q-k}\right)z^q +\sum_{q=n+1}^{2n}\left(\sum_{k=q-n}^na_kb_{q-k}\right)z^q\tag{3}\\ &\,\,\color{blue}{=\sum_{q=0}^{2n}\left(\sum_{k=\max\{0,q-n\}}^{\min\{q,n\}}a_kb_{q-k}\right)z^q}\tag{4} \end{align*}

Comment:

  • In (1) we see the greatest power of $z$ is $2n$ explaining the upper limit of the outer sum.

  • In (2) we eliminate the index $l$ using $l=q-k$, but we still have to respect the conditions on $k$ and $l=q-k$.

  • In (3) we split the sum and respect explicitly the conditions $0\leq k\leq n$ and $0\leq q-k\leq n$.

  • In (4) we merge the sums from (3).

    • Note the outer sum in (4) has $2n+1$ summands with index $q$ ranging from $0$ to $2n$. This coincides with the $n+1$ summands from the left-hand outer sum of (3) with index $q$ ranging from $0\leq q\leq n$ and the $n$ summands from $n+1$ to $2n$ from the right-hand outer sum of (3) with index $q$ ranging from $n+1$ to $2n$.

    • The summand $a_kb_{q-k}$ is the same in both inner sums of (3) which is also taken in (4).

    • Comparing lower and upper limit of the inner sums in (3) with $\max\{0,q-n\}\leq k\leq \min\{q,n\}$ and respecting the lower and upper limits of the outer index $q$ shows equality.

Evaluating (4) at $z=1$ we obtain the relation \begin{align*} &\color{blue}{\left(\sum_{k=0}^{n}a_k\right) \left(\sum_{k=0}^{n}b_k\right)}\color{blue}{= \sum_{q=0}^{2n}\left(\sum_{k=\max\{0,q-n\}}^{\min\{q,n\}}a_kb_{q-k}\right)} \end{align*} in accordance with OPs claim.

Markus Scheuer
  • 108,315
  • Can you write the above more compact and is this equal to $\left(\sum_{k=0}^{n}a_{k} \right) \left(\sum_{k=0}^{n}b_{k}\right) = \sum_{k=0}^{2n}\left(\sum_{i=\max(0,k-n)}^{min(n,k)}a_{i}b_{k-i}\right)$? Can you show me this with a proper mathematical proof like you did above? – Math98 Jun 12 '23 at 09:58
  • 1
    @Math98: This more compact notation is sound. You might observe, it is different to your expression in the title. I can provide some helpful info next weekend, if you like. – Markus Scheuer Jun 12 '23 at 19:43
  • 1
    I changed the OP, the correct expression is there now. If you can show me how you go from your provided expression to the expression I showed and some intuitive explanation that would be appreciated. – Math98 Jun 13 '23 at 10:45
  • 1
    @Math98: Very good. I will add some information at the weekend, if no other proper answer will be provided meanwhile. – Markus Scheuer Jun 13 '23 at 11:04
  • @Math98: I've added some details which might help to clarify the situation. – Markus Scheuer Jun 17 '23 at 19:28
1

$$\left(\sum_{k=0}^{n}a_{k} \right) \left(\sum_{k=0}^{n}b_{k}\right) = \sum_{k=0}^{2n}\left(\sum_{i=0}^{k}a_{i}b_{k-i}\right)$$

You can think in the following way, first we list the LHS

$$(a_0+a_1+\dots+a_n)(b_0+b_1+\dots+b_n)$$

If we expand it, we will get terms like $a_i b_j$, now we can group these terms by the sum of index $i+j$, for example,

$$\begin{align}i+j=0&\Longleftrightarrow i=0, j=0\\ \\ i+j=1&\Longleftrightarrow (i=0,j=1)\lor (i=1,j=0)\\ \\ i+j=2&\Longleftrightarrow (i=0,j=2)\lor (i=2,j=0)\lor (i=1,j=1)\\ \\ \dots\end{align}$$

If we set the index sum be $i+j=k$, we get the group for product with the sum of index equals $k$, which is

$$\sum_{i=0}^{k}a_{i}b_{k-i}$$

Next, we can see the $k$ goes from $0+0=0$ until $n+n=2n$, therefore

$$\left(\sum_{k=0}^{n}a_{k} \right) \left(\sum_{k=0}^{n}b_{k}\right) = \sum_{k=0}^{2n}\left(\sum_{i=0}^{k}a_{i}b_{k-i}\right)$$

Can you proceed to verify your result? (your result in the OP is incorrect.)


Remarks:

For those terms $a_k, b_k$, where $k>n$, we set them to be zero.

MathFail
  • 21,128
  • Well intuitive explanation. I get the idea and the following result now, however it is based on a pattern I see and it seems to be correct. Should the answer be as follow: $\left(\sum_{k=0}^{n}a_{k} \right) \left(\sum_{k=0}^{n}b_{k}\right) = \sum_{k=0}^{2n}\left(\sum_{i=\max(0,k-n)}^{min(n,k)}a_{i}b_{k-i}\right)$? Do you know the correct intuitive and mathematical derivation for this? – Math98 Jun 12 '23 at 09:51
  • What is the point to write it in this way? $k\le n$ hence $\min(n,k)=k, \max(0, k-n)=0$ – MathFail Jun 12 '23 at 11:00
  • $k$ is not always smaller than $n$, as $k$ ranges from $0$ to $2n$. The point to write it in that way, is that otherwise the expression is incorrect or less compact. – Math98 Jun 12 '23 at 12:25
  • Sorry I didn't read it carefully, yes, it is right, you can list a few terms and can see the intuition, which immediately leads to the math derivation. But this OP is not asking for this. I suggest to close this OP first, and then put you new question in a new post. @Math98 – MathFail Jun 12 '23 at 18:48
  • I will just change the OP appropriately, as the step towards the answer I provided is already in line to what you showed. – Math98 Jun 13 '23 at 10:16
  • If you can show me how you can get to the answer in OP (changed now correctly) and some intuitive explanation like you did before, that would be nice. So an extension to what you already wrote. @MathFail – Math98 Jun 13 '23 at 10:49
  • @MathFail: It seems the last identity in your answer is not correct for $n\geq 1$. The right sum gives for $k=2n, i=0$ the summand $a_0b_{2n}$ which is not defined for the left sum. – Markus Scheuer Jun 17 '23 at 19:16
  • We set those undefined terms $a_{k}=0, b_k=0$ if $k> n$ @epi163sqrt – MathFail Jun 17 '23 at 19:18
  • @Mathfail: This convention needs to be explicitly stated since it is not common usage. Otherwise it is as you said in the comment simply undefined. – Markus Scheuer Jun 17 '23 at 19:21
  • 1
    Ok, I have added it @epi163sqrt – MathFail Jun 17 '23 at 19:25