7

I recently came across a question while studying for an exam. I haven't been able to solve it. We had to prove: $$\sum_{k=0}^{n}{2n+1\choose k} = 2^{2n}$$

We had to use counting techniques. This was my attempt

Let S be the set of all subsets of [1....2n]. We know that the size of S is $2^{2n}$ Another way of counting the subsets of [1....2n] is ?????

...

Therefore, since we've used two different methods to count the same thing, then $$\sum_{k=0}^{n}{2n+1\choose k} = 2^{2n}$$

My problem is, I can't think of a second way to count the subsets such that it equals the summation. Am I on the right track here, or is there another set of objects I can count to make the proof easier?

Thanks for the help.

  • 3
    I don't quite understand why you ask for a counting proof and then accept an answer that uses equations between binomial coefficients instead of counting things. – joriki Jun 23 '16 at 16:33
  • 1
    Sorry for the confusion. I figured that the equation describes a way that objects can be counted. The way I've learned to do these proofs is to consider some set, and show that the elements of the set can be counted in two different ways. These two different ways will generate two different expressions, which we can equate (since we're counting the same thing). Hope that makes sense – The_Questioner Jun 23 '16 at 17:08

4 Answers4

16

The subsets of $\{1,\ldots,2n+1\}$ come in pairs of complements. Exactly one member of each such pair has up to $n$ elements.

joriki
  • 238,052
7

We have to consider the set $\{1,2,3,...,2n+1\}$. There are $2^{2n+1}$ subsets here.

However, another way to look at this is that we can choose $k$ numbers from the set of $2n+1$ elements, which is ${2n+1 \choose k}$. We sum this from $k=0$ to $k=2n+1$, giving us: $$\sum_{k=0}^{2n+1} {2n+1 \choose k}=2^{2n+1}$$ Now, remember that: $${2n+1 \choose k}={2n+1 \choose 2n+1-k}$$ This means ${2n+1 \choose 2n+1}$ is a duplicate of ${2n+1 \choose 0}$, ${2n+1 \choose 2n}$ is a duplicate of ${2n+1 \choose 1}$, ${2n+1 \choose 2n-1}$ is a duplicate of ${2n+1 \choose 2}$, ..., and ${2n+1 \choose n+1}$ is a duplicate of ${2n+1 \choose n}$. Therefore, we can sum from $k=0$ to $k=n$ and then multiply that by $2$ to account for the duplicates: $$2\sum_{k=0}^{n} {2n+1 \choose k}=2^{2n+1}$$ Hopefully, you can take it from here. Good luck!

Noble Mushtak
  • 18,402
  • 28
  • 44
  • Thanks. It seems everyone who answered this question had the same idea to consider the set {1...2n+1]. How do you guys know to do that? I would have never thought of doing it like this. – The_Questioner Jun 23 '16 at 16:41
  • 2
    It said ${2n+1 \choose k}$. That means you choose $k$ elements from a set of $2n+1$ elements, so we knew it had something to do with a set of $2n+1$ elements. Then, everyone realized that we can split the combinations into pairs of duplicates and thus we only need to sum from $k=0$ to $k=n$. Personally, I had never seen this before, so it took me a while to realize it, but the important part is that you look at the binomial coefficient, not the power of $2$, to realize that this is a set of $2n+1$ and not $2n$. – Noble Mushtak Jun 23 '16 at 16:44
3

$$\sum_{k=0}^{n}{2n+1\choose k} =\sum_{k=0}^{n}{2n+1\choose {2n+1-k}} = \sum_{k=n+1}^{2n+1}{2n+1\choose k}$$

$$2^{2n+1}=\sum_{k=0}^{2n+1}{2n+1\choose k}=\sum_{k=0}^{n}{2n+1\choose k}+\sum_{k=n+1}^{2n+1}{2n+1\choose k}=2\sum_{k=0}^{n}{2n+1\choose k}$$

$$\implies 2^{2n+1}=2\sum_{k=0}^{n}{2n+1\choose k}$$

which implies the result given.

An interpretation of this is that a randomly selected subset of $\{1,\cdots,2n+1\}$ is equally likely to contain $\le n$ or $>n$ elements.

πr8
  • 10,800
  • Agree that it's perhaps not clear, though it's really just letting $j=2n+1-k$, adjusting the boundaries of the sum appropriately, and then renaming the dummy variable $j$ as $k$ for consistency. – πr8 Jun 25 '16 at 06:04
0

Here is a way to see that $$ \sum_{k=0}^{n} {2n+1 \choose k} = 2^{2n}, $$ by a counting argument which counts the number of subsets of $\{1, \ldots, 2n\}$ rather than half the number of subsets of $\{1, \ldots, 2n, 2n+1\}$. Essentially to count the subsets of $\{1, \ldots, 2n\}$, we break the subsets into whether or not their size is at most $n$.

Now to choose a subset, $B$, of $\{1, \ldots, n\}$, for some $0 \leq k \leq n$, we choose $k$ elements of $\{\star, 1, 2, \ldots, 2n\}$ getting a subset $A$. If $\star$ is not in $A$, then set $B=A$. If $\star$ is in $A$, then set $B=\{1, \ldots, 2n\} \setminus A$. Now the number of ways to choose our set $A$ is precisely $$ \sum_{k=0}^{n} {2n+1 \choose k}. $$ What remains is to show a 1-1 correspondence between the sets $A$ and subsets of $\{1, 2, \ldots, 2n\}$.

First, let us show that the correspondence between sets $A$ and subsets $B$ is onto. If $B$ is a subset of $\{1, \ldots, 2n\}$, then if $|B| \leq n$, we can let $A=B$; but if $|B| \geq n+1$, then we set $A=\{\star\} \cup \left( \{1, \ldots, 2n\}\setminus B\right)$ [now $|A| = 1+2n-|B| \leq n.$] Either way, this $A$ would give rise to our subset $B$.

Conversely, let us show that this correspondence is injective. Suppose $A$ and $A'$ both correspond to subset $B=B'$. Now if either $A$ and $A'$ both contain $\star$ or $A$ and $A'$ both do not contain $\star$, then clearly we must have $A=A'$. Suppose $A$ contains $\star$ but $A'$ does not. However, \begin{align*} |B| = 2n-(|A|-1) = n+1 +(n-|A|) &\geq n+1 \\&> n \geq |A'|=|B'|. \end{align*}

D Poole
  • 2,762