While playing around with random binomial coefficients , I observed that the following identity seems to hold for all positive integers $n$:
$$ \sum_{k=0}^{2n} (-1)^k \binom{4n}{2k}\cdot\binom{2n}{k}^{-1}=\frac{1}{1-2n}.$$
However, I am unable to furnish a proof for it ( though this result is just a conjecture ).
Any ideas/suggestions/solutions are welcome.
When going through Jack's nice answer I did some intermediate steps to better see what's going on. Here is a somewhat more elaborated version, which might also be convenient for other readers.
Comment:
In (1) we write the reciprocal of a binomial coefficient using the beta function \begin{align*} \binom{n}{k}^{-1}=(n+1)\int_0^1z^k(1-z)^{n-k}\,dz \end{align*}
In (2) we do some rearrangements in order to apply the binomial theorem.
In (3) we consider the even function derived from $(1+z)^{2n}$ \begin{align*} \sum_{k=0}^{n}\binom{2n}{2k}z^{2k}=\frac{1}{2}\left((1+z)^{2n}+(1-z)^{2n}\right) \end{align*} Replacing $n$ with $2n$ and $z$ with $i\sqrt{\frac{1-x}{x}}$ and the application of de Moivre's theorem in (6) becomes plausible.
In (4) we do some simplifications.
In (5) we substitute $x=\cos ^2\theta, dx=-2\cos\theta\sin\theta\,d\theta$.
In (6) we apply De Moivre's theorem and in (7) and (8) trigonometric sum formulas.