-1

$$\sum_{k=0}^{\left \lceil \frac{n}{2} \right \rceil}\binom{n}{2k}=2^{n-1}$$

I have no idea how to start, can i get hints?

RobPratt
  • 45,619

6 Answers6

3

$$\binom{n}{0}+\binom{n}{2}+\binom{n}{4} \ldots$$

counts the number of even-sized subsets of $\{1 \ldots n\}$. I assume that you know the total number of subsets of $\{1\ldots n\}$ is $2^n$. To show the even-sized subsets number $2^{n-1}$, a direct way would be to show that it's the same as the number of subsets of $\{1 \ldots n-1\}$.

So: lay out all of your subsets of even size. For each set $S$, consider whether it contains $n$ or not. If it does not contain $n$, then it is an even sized subset of $\{1 \ldots n-1\}$. If it does contain $n$, then $S -\{n\}$ is an odd-sized subset of $\{1 \ldots n-1\}$. Clearly this is a reversible function - given any $T \subset \{1 \ldots n-1\}$, I can tell you the unique even subset of $\{1 \ldots n\}$ that maps to it (depending on whether it has even or odd size itself). It follows that the two sets have the same size.

1

Hint: Consider the following bijection $$\varphi:\mathcal{P}([n])\longrightarrow \mathcal{P}([n]),$$ given by $$\varphi (X)=\begin{cases}X\cup \{n\}&\text{if }n\not \in X\\ X\setminus \{n\}& \text{Otherwise}\end{cases}.$$ Show that this is a bijection that maps even sized to odd sized sets and viceversa. Use binomial theorem to conclude(or involution principle if you know it).

Phicar
  • 14,722
-1

Maybe you can try by induction?

As a starting point, compute it for $n=0, 1, 2$, just to verify than the relation is true for some base cases.

Than, you can assume the relation to be true for $n$ and show that, with this assumption, then, the relation holds also for $n+1$.

And you're done!

I would try this way! Let me know ;)

Marco
  • 49
-1

Hint: Using the binomial theorem, what is $ ( 1 + 1) ^n$?
What is $ ( 1 - 1) ^n$?

Calvin Lin
  • 68,864
-1

The LHS gives the number of subsets of even size of a set with cardinality $n$.

Now, there are a total of $2^n$ subsets, and clearly exactly half of them will have even cardinality, i.e. $2^n/2 = 2^{n-1}$

Since both sides count the same thing, we are done.

Vishu
  • 14,469
  • @CalvinLin Alright, but I stated ‘by symmetry’ to convey that it’s intuitively obvious. – Vishu Jan 03 '21 at 18:00
  • 1
    If so, please say "it's obvious that" instead of the loaded term "by symmetry". I don't think many other people equate "by symmetry" with "it's obvious". thanks for removing it. – Calvin Lin Jan 03 '21 at 18:02
  • 1
    Separately, I don't think it's "clear" or "obvious", especially for someone who's seeing it for the first time (IE OP). Maybe it is for someone who's already seen it, but most people that I know who see it for the first time, get stumped by it. In fact, the more useful hint that I've seen, is asking them to apply ${ n \choose k } = {n-1 \choose k-1 } + { n -1 \choose k}$, and then sum it up. – Calvin Lin Jan 03 '21 at 18:04
  • @CalvinLin Shouldn’t it be relatively clearer? As in, why would the odd-sized subsets be favoured over the even-sized ones or vice-versa? Besides, the OP was asking for a purely combinatorial proof. – Vishu Jan 03 '21 at 18:07
  • 2
    By your logic, would the "sets with 3n", "sets with 3n+1", "sets with 3n+2" elements all have the same size, since why should one of them be favored over the other and vice versa? However, we know the the conclusion is not true. – Calvin Lin Jan 03 '21 at 18:08
  • @CalvinLin I never claimed that my logic extends beyond this. I just think it’s intuitively clear in this case. – Vishu Jan 03 '21 at 18:10
  • And I'm claiming that it's not intuitive nor obvious, esp to someone who saw that equation for the first time. We can disagree on that if you want. I've nothing more to add to this conversation. – Calvin Lin Jan 03 '21 at 18:11
  • @CalvinLin Neither you nor I have the right to say what someone else is thinking. Ask them. One can only say with certainty what he/she thinks. And I said ‘clearly’ because it is clear to me. – Vishu Jan 03 '21 at 18:12
-1

\begin{align} \sum_{k=0}^{\left \lceil \frac{n}{2} \right \rceil}\binom{n}{2k} &=\sum_{k=0, k\text{ even}}^{n}\binom{n}{k} =\sum_{k=0}^{n}\binom{n}{k}-\sum_{k=0, k\text{ odd}}^{n}\binom{n}{k}\\ &=2^n+\sum_{k=0, k\text{ odd}}^{n}(-1)^k\binom{n}{k}\\ &=2^n+\sum_{k=0}^{n}(-1)^k\binom{n}{k}-\sum_{k=0, k\text{ even}}^{n}(-1)^k\binom{n}{k}\\ &=2^n+(1-1)^n-\sum_{k=0}^{\left \lceil \frac{n}{2} \right \rceil}\binom{n}{2k}\\ \implies 2\sum_{k=0}^{\left \lceil \frac{n}{2} \right \rceil}\binom{n}{2k} &= 2^n \implies \sum_{k=0}^{\left \lceil \frac{n}{2} \right \rceil}\binom{n}{2k} = 2^{n-1} \end{align}

Explorer
  • 3,147
  • 1
    IMO It's easier to see this identity by writing out $(1+1)^n, (1-1)^n$ separately, which was what I was going for. – Calvin Lin Jan 03 '21 at 18:16