# Finding b_n - Binomial theorem problem

#### Pranav

##### Well-known member
Question:
If $\displaystyle \sum_{r=0}^{2n} a_r(x-2)^r=\sum_{r=0}^{2n} b_r(x-3)^r$ and $a_k=1$ for all $k \geq n$, then show that $b_n={}^{2n+1}C_{n+1}$.

Attempt:
I haven't been able to make any useful attempt on this one. I could rewrite it to:

$$\sum_{r=0}^{n-1} a_r(x-2)^r + (x-2)^n\left(\frac{(x-2)^{n+1}-1}{(x-2)-1}\right)=\sum_{r=0}^{2n} b_r(x-3)^r$$

I am clueless about the next step.

Any help is appreciated. Thanks!

#### Klaas van Aarsen

##### MHB Seeker
Staff member
Question:
If $\displaystyle \sum_{r=0}^{2n} a_r(x-2)^r=\sum_{r=0}^{2n} b_r(x-3)^r$ and $a_k=1$ for all $k \geq n$, then show that $b_n={}^{2n+1}C_{n+1}$.

Attempt:
I haven't been able to make any useful attempt on this one. I could rewrite it to:

$$\sum_{r=0}^{n-1} a_r(x-2)^r + (x-2)^n\left(\frac{(x-2)^{n+1}-1}{(x-2)-1}\right)=\sum_{r=0}^{2n} b_r(x-3)^r$$

I am clueless about the next step.

Any help is appreciated. Thanks!
Hi Pranav!

Did you already expand it for $n=2$ and $n=3$?
Just to see if we can find a pattern?

I don't have a solution yet, but I can see that $b_n$ can be fully deduced from the terms starting from n up to 2n.
All terms below n are irrelevant.

Let us define $c_t$ such that the sum is equal to $$\displaystyle \sum_{t=0}^{2n} c_t x^t$$.
That is, we have that:
$$\sum_{t=0}^{2n} c_t x^t = \sum_{r=0}^{2n} (x-2)^r=\sum_{s=0}^{2n} b_s(x-3)^s$$

With $a_{2n}=1$, we get that highest power $x^{2n}$ has coefficient:
$$c_{2n}=b_{2n}=a_{2n}=1$$
For $n=0$ we would be done, since $b_n = b_0 = \binom{2\cdot 0 + 1}{0+1} = \binom{1}{1} = 1$.

Next is $c_{2n-1}$, which yields:
\begin{array}{lclcl}
c_{2n-1} &=& a_{2n-1} + \binom{2n}{1}a_{2n}(-2) &=& b_{2n-1} + \binom{2n}{1}b_{2n}(-3) \\
c_{2n-1} &=& 1 - 4n &=& b_{2n-1} - 6n \\
b_{2n-1} &=& 2n + 1
\end{array}
For $n=1$ we would be done, since $b_n = b_1 = 2\cdot 1 + 1 = \binom{2\cdot 1 + 1}{1+1} = \binom{3}{2} = 3$.

Next is $c_{2n-2}$...
Sorry, that's as far as I got at this time.

#### Pranav

##### Well-known member
Hi Pranav!

Did you already expand it for $n=2$ and $n=3$?
Just to see if we can find a pattern?

I don't have a solution yet, but I can see that $b_n$ can be fully deduced from the terms starting from n up to 2n.
All terms below n are irrelevant.

Let us define $c_t$ such that the sum is equal to $$\displaystyle \sum_{t=0}^{2n} c_t x^t$$.
That is, we have that:
$$\sum_{t=0}^{2n} c_t x^t = \sum_{r=0}^{2n} (x-2)^r=\sum_{s=0}^{2n} b_s(x-3)^s$$

With $a_{2n}=1$, we get that highest power $x^{2n}$ has coefficient:
$$c_{2n}=b_{2n}=a_{2n}=1$$
For $n=0$ we would be done, since $b_n = b_0 = \binom{2\cdot 0 + 1}{0+1} = \binom{1}{1} = 1$.

Next is $c_{2n-1}$, which yields:
\begin{array}{lclcl}
c_{2n-1} &=& a_{2n-1} + \binom{2n}{1}a_{2n}(-2) &=& b_{2n-1} + \binom{2n}{1}b_{2n}(-3) \\
c_{2n-1} &=& 1 - 4n &=& b_{2n-1} - 6n \\
b_{2n-1} &=& 2n + 1
\end{array}
For $n=1$ we would be done, since $b_n = b_1 = 2\cdot 1 + 1 = \binom{2\cdot 1 + 1}{1+1} = \binom{3}{2} = 3$.

Next is $c_{2n-2}$...
Sorry, that's as far as I got at this time.

But is there any proper way of doing it, I mean by some algebraic manipulation? I am thinking of differentiating both the sides wrt r number of times. I will see if I reach somewhere using that.

#### Pranav

##### Well-known member
I have reached the right answer but I cannot perform a summation I come across midway in the solution.

Differentiating wrt r n times as I said before, I get:
$$\sum_{r=0}^{2n} r(r-1)...(r-n+1)a_r(x-2)^{r-n}=\sum_{r=0}^{2n} r(r-1)...(r-n+1)b_r(x-3)^{r-n}$$
$$\Rightarrow \sum_{r=0}^{2n} \frac{r!}{(r-n)!}a_r(x-2)^{r-n}=\sum_{r=0}^{2n} \frac{r!}{(r-n)!}b_r(x-3)^{r-n}$$
Substituting x=3 and using the fact that $a_k=1$ for all $k\geq n$,
$$\sum_{r=n}^{2n} \frac{r!}{(r-n)!}=n! \cdot b_n$$

I did the summation using Wolfram Alpha but how do I solve the summation manually?

#### Klaas van Aarsen

##### MHB Seeker
Staff member
I have reached the right answer but I cannot perform a summation I come across midway in the solution.

Differentiating wrt r n times as I said before, I get:
$$\sum_{r=0}^{2n} r(r-1)...(r-n+1)a_r(x-2)^{r-n}=\sum_{r=0}^{2n} r(r-1)...(r-n+1)b_r(x-3)^{r-n}$$
$$\Rightarrow \sum_{r=0}^{2n} \frac{r!}{(r-n)!}a_r(x-2)^{r-n}=\sum_{r=0}^{2n} \frac{r!}{(r-n)!}b_r(x-3)^{r-n}$$
Substituting x=3 and using the fact that $a_k=1$ for all $k\geq n$,
$$\sum_{r=n}^{2n} \frac{r!}{(r-n)!}=n! \cdot b_n$$

I did the summation using Wolfram Alpha but how do I solve the summation manually?
Nope. I have no clue.
Looks like you're nicely going in the right direction, seeing how everything collapses.

Btw, your summations should start at n instead of 0 after taking the nth derivative.

Anyway, I can see that the summation can be reduced to a summation of combinations that add up to another combination.
And I can see that it holds true for a number of examples.
But as yet I have no clue how to prove it.

#### Pranav

##### Well-known member
Nope. I have no clue.
Looks like you're nicely going in the right direction, seeing how everything collapses.

Btw, your summations should start at n instead of 0 after taking the nth derivative.

Anyway, I can see that the summation can be reduced to a summation of combinations that add up to another combination.
And I can see that it holds true for a number of examples.
But as yet I have no clue how to prove it.

Is there any alternative method if there is no way to solve the summation?

Thanks!

#### ZaidAlyafey

##### Well-known member
MHB Math Helper
Differentiating wrt r n times as I said before, I get:
$$\sum_{r=0}^{2n} r(r-1)...(r-n+1)a_r(x-2)^{r-n}=\sum_{r=0}^{2n} r(r-1)...(r-n+1)b_r(x-3)^{r-n}$$
You have some typos

You are differentiating with respect to $x$ .

The nth derivative with respect to $x$

$$\sum_{r=n}^{2n} r(r-1)...(r-n+1)a_r(x-2)^{r-n}=\sum_{r=n}^{2n} r(r-1)...(r-n+1)b_r(x-3)^{r-n}$$

#### ZaidAlyafey

##### Well-known member
MHB Math Helper
$$\displaystyle \sum_{r=n}^{2n} \frac{r!}{(r-n)!}=n! \cdot b_n$$

$$\displaystyle b_n =\sum_{r=n}^{2n} \frac{r!}{(r-n)!\, n! }$$

$$\displaystyle b_n =\sum_{r=n}^{2n} {n \choose r }$$

we need to prove that

$$\displaystyle \sum_{r=n}^{2n} {r \choose n } = {2n+1 \choose n+1}={2n+1 \choose n}$$

By induction on $n$

for $n = 1$ we have

$$\displaystyle \sum_{r=1}^{2} {r\choose 1 } = {3 \choose 1}$$

Next assume

$$\displaystyle \sum_{r=k}^{2k} {r \choose k } = {2k+1 \choose k}$$

and we need to prove that

$$\displaystyle \sum_{r=k+1}^{2k+2} {r \choose k+1 } = {2k+3 \choose k+1}$$

I will leave the rest for you.

#### jacks

##### Well-known member
Given $\displaystyle \sum_{r=0}^{2n}a_{r}\left(x-2\right)^r = \sum_{r=0}^{2n}b_{r}\left(x-3\right)^r$ and $a_{k} = 1$ forall $k\geq n$

Let $(x-3) = t$ Then $(x-2) = (t+1)$

$\displaystyle \sum_{r=0}^{2n}a_{r}(1+t)^r = \sum_{r=0}^{2n}b_{r}t^r$

$\displaystyle a_{0}+a_{1}(1+t)+a_{2}(1+t)^2+............+a_{n}(1+t)^n+a_{n+1}(1+t)^{n+1}+.......+a_{2n}(1+t)^{2n} = b_{0}+b_{1}t+b_{2}t^2+.......+b_{n}t^n+b_{n+1}t^{n+1}+......+b_{2n}t^{2n}$

Now camparing Coefficient of $t^n$ on both side and using $a_{k}=1$ forall $k\geq n$

$\displaystyle \binom{n}{n}+\binom{n+1}{n}+\binom{n+2}{n}+........+\binom{2n}{n} = b_{n}$

$\displaystyle \binom{n+1}{n+1}+\binom{n+1}{n}+\binom{n+2}{n}+........+\binom{2n}{n} = b_{n}$

Now using $\displaystyle \binom{n}{r}+\binom{n}{r-1} = \binom{n+1}{r}$ succesively

$\displaystyle \binom{2n+1}{n+1} = b_{n}$

Here $\displaystyle b_{n} =$ Coefficient of $t^n$ on $\bf{R.H.S}$

#### Klaas van Aarsen

##### MHB Seeker
Staff member
Next assume

$$\displaystyle \sum_{r=k}^{2k} {r \choose k } = {2k+1 \choose k}$$

and we need to prove that

$$\displaystyle \sum_{r=k+1}^{2k+2} {r \choose k+1 } = {2k+3 \choose k+1}$$

I will leave the rest for you.
This is where I got stuck...

However, the approach of jacks looks perfect!

#### ZaidAlyafey

##### Well-known member
MHB Math Helper
This is where I got stuck...
Agreed , I should have tried it before posting , sorry .

#### Pranav

##### Well-known member
Given $\displaystyle \sum_{r=0}^{2n}a_{r}\left(x-2\right)^r = \sum_{r=0}^{2n}b_{r}\left(x-3\right)^r$ and $a_{k} = 1$ forall $k\geq n$

Let $(x-3) = t$ Then $(x-2) = (t+1)$

$\displaystyle \sum_{r=0}^{2n}a_{r}(1+t)^r = \sum_{r=0}^{2n}b_{r}t^r$

$\displaystyle a_{0}+a_{1}(1+t)+a_{2}(1+t)^2+............+a_{n}(1+t)^n+a_{n+1}(1+t)^{n+1}+.......+a_{2n}(1+t)^{2n} = b_{0}+b_{1}t+b_{2}t^2+.......+b_{n}t^n+b_{n+1}t^{n+1}+......+b_{2n}t^{2n}$

Now camparing Coefficient of $t^n$ on both side and using $a_{k}=1$ forall $k\geq n$

$\displaystyle \binom{n}{n}+\binom{n+1}{n}+\binom{n+2}{n}+........+\binom{2n}{n} = b_{n}$

$\displaystyle \binom{n+1}{n+1}+\binom{n+1}{n}+\binom{n+2}{n}+........+\binom{2n}{n} = b_{n}$

Now using $\displaystyle \binom{n}{r}+\binom{n}{r-1} = \binom{n+1}{r}$ succesively

$\displaystyle \binom{2n+1}{n+1} = b_{n}$

Here $\displaystyle b_{n} =$ Coefficient of $t^n$ on $\bf{R.H.S}$
Thanks a lot jacks!

BTW, my friend came up with exactly the same procedure this morning and I was going to post it here as soon as I reach home.

EDIT: Looks like you made it a bit lengthy. Rewrite the summation as:

$$\sum_{r=0}^{n-1} a_r(1+t)^{r}+ (1+t)^n\frac{(1+t)^{n+1}-1}{t}=\sum_{r=0}^{2n}b_rt^r$$
$$\sum_{r=0}^{n-1} a_r(1+t)^{r}+ \frac{(1+t)^{2n+1}-(1+t)^n}{t}=\sum_{r=0}^{2n}b_rt^r$$
Now we need the coefficient of $t^n$ on both the sides.

On the left hand side, we find the coefficient of t^(n+1) from (1+t)^(2n+1) as there is a $t$ in the denominator.

Hence, the coefficient of t^n on LHS is $\binom{2n+1}{n+1}$.

Last edited: