12

I would like to prove inductively that $${2n\choose n}=\sum_{i=0}^n{n\choose i}^2.$$

I know a couple of non-inductive proofs, but I can't do it this way. The inductive step eludes me. I tried naively things like $${2n+2\choose n+1}={2n+2\over n+1}{2n+1\choose n}=2\cdot {2n+1\over n+1}{2n\choose n},$$

But I don't think it can lead me anywhere. I would like the proof to be as simple as possible.

Martin Sleziak
  • 53,687
Bartek
  • 6,265
  • 2
    This is a special case of Chu-Vandermonde identity. Sometimes a proof by induction might be easier if you prove a more general result. Although I am not sure whether it helps in this case. – Martin Sleziak Mar 04 '13 at 11:44
  • I'll add a link to another question, where a combinatorial proof is given. – Martin Sleziak Mar 04 '13 at 11:48
  • This result is formulated too narrowly to have much chance of a inductive proof: knowing something about just the central binomial coefficients is insufficient in the induction because you don't have a useful recurrence realtion for just the central binomial coefficients. However, proving as Martin Sleziak suggested Chu-Vandermonde by induction (using Pascal's recurence) is a piece of cake. – Marc van Leeuwen Mar 05 '13 at 11:00

2 Answers2

7

Split the $2n$ elements into two groups of size $n$ Then the no. of ways of choosing $n$ from the $2n$ is the no. of ways of choosing $i$ from the 1st and $n-i$ from the 2nd and letting $i$ vary.

Cooperation
  • 616
Ishan Banerjee
  • 6,459
1

After Martin Sleziak and Marc van Leeuwen's comments, I've found this inductive proof of Vandermonde's identity. (On this very site.)

Community
  • 1
Bartek
  • 6,265