4
$\begingroup$

Everything is in the title. I wonder if there is known good asymptotic (when $n\to+\infty$) for the quantity $LCM(\binom{2k}k)_{1\le k\le n}$

Thanks in advance

$\endgroup$
1
  • $\begingroup$ It's tabulated at oeis.org/A025540 but there's nothing there about asymptotics. $\endgroup$ Apr 13, 2016 at 3:11

2 Answers 2

5
$\begingroup$

Put $A_n$ to be the lcm of $\binom{2k}{k}$ for $1\le k\le n$, and let $B_n$ denote the lcm of the numbers at most $2n$ (thus $B_n = \exp(\psi(2n))$). We claim that $A_n = B_n/2$ unless $n=2^r-1$ in which case $A_n =B_n$. Since asymptotics for the $\psi(2n) = \sum_{r\le 2n} \Lambda(r)$ are well known (the prime number theorem), this solves the problem.

Consider an odd prime $p$. The power of $p$ dividing $B_n$ is the largest power $p^r$ lying below $2n$. Now, if $k\le n$, then the power of $p$ dividing $\binom{2k}{k}$ is $$ \sum_{\ell=1}^{\infty}\Big( \Big\lfloor \frac{2k}{p^\ell} \Big\rfloor -2 \Big\lfloor \frac{k}{p^\ell} \Big\rfloor\Big). $$ Since only terms with $p^\ell \le 2k$ are relevant, and each term in parentheses is at most $1$, this exponent is at most $r$. Also if we take $2k=p^r+1$ (which is below $2n$, since $p^r$ is odd and at most $2n$) then $$ \Big\lfloor \frac{p^r+1}{p^\ell} \Big\rfloor -2 \Big\lfloor \frac{p^{r}+1}{2p^\ell} \Big\rfloor =p^{r-\ell} - 2\Big \lfloor \frac{p^{r}+1}{2p^{\ell}}\Big\rfloor = 1 $$ for all $1\le \ell \le r$ (it must be odd, and therefore equal one). So the power of $p$ dividing $A_n$ matches that dividing $B_n$.

The case for $2$ is similar. We check that the power of $2$ dividing $\binom{2k}{k}$ is strictly less than the largest power of $2$ at most $2k$, unless $k$ is one less than a power of $2$ in which case it equals the largest power of $2$ at most $2k$.

$\endgroup$
0
2
$\begingroup$

Kummer's theorem implies the formula: $$\mathrm{LCM}(\binom{2}{1},\binom{4}{2},\dots,\binom{2n}{n}) = 2^{\lfloor \log_2(n+1)\rfloor} \cdot \prod_{3\leq p\leq 2n} p^{\lfloor \log_p(2n-1)\rfloor},$$ where $p$ runs over primes.

In particular, $$\mathrm{LCM}(\binom{2}{1},\binom{4}{2},\dots,\binom{2n}{n}) \leq (n+1)\cdot (2n-1)^{\pi(2n)-1}$$ where the r.h.s. is larger than LCM in at most $(2n)\#$ (primorial) times.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.