2

I am quite certain that $2\binom{d}{k} \le \binom{2d}{k}$ holds for every positive integers $k,d$, where $1 \le k \le d$.

Is there a simple proof? A combinatorial proof ?

An immediate combinatorial argument gives $2\binom{d}{k} \le \binom{2d}{2k}$, as the LHS corresponds to choosing $k$ elements out of $d$, and then choosing additional $k$ elements out of a distinct new set of $d$ elements. The RHS correspond to an "unconstrained" choice of $2k$ out of $2d$.

(I checked the cases $k=1,d$, and various other examples with "small numbers". For "large" $d$ the asymptotics of the binomial coefficients seems compatible with this inequality).

Asaf Shachar
  • 25,967
  • 4
    When choosing $k$ out of $2d$ elements, you can choose $k$ among the first $d$ or $k$ among the last $d$ (or some mix). This gives $2\binom{d}{k}$ as a lower bound. – Wojowu Feb 04 '19 at 11:11

2 Answers2

2

Since this turned out to be quite trivial, I will just make Wojowu's comment into an answer:

When choosing $k$ out of $2d$ elements, we can choose $k$ among the first $d$ or $k$ among the last $d$ (or some mix). This gives $2\binom{d}{k}$ as a lower bound.

Asaf Shachar
  • 25,967
1

A "combinatorial proof":

When chosing $k \ge 1$ elements from a $2d$ set ($k \le d$), we can chose $j$ ($0 \le j \le k$) elements from the first $d$ elements and $k-j$ from the last $d$ elements. The number of possibilities is therefore

$$ \binom{2d}{k} = \sum_{j=0}^k \binom{d}{j} \binom{d}{k-j} = 2\binom{d}{k} + \sum_{j=1}^{k-1} \binom{d}{j} \binom{d}{k-j} $$

The "remaining" sum might be empty (if $k=1$) but is clearly not negative.

Wolfgang Kais
  • 1,430
  • 8
  • 7