I'm currently frequenting a course in Combinatorics and I've come to this problem where I'm supposed to compute the following limit:

$$\lim_{n \to \infty} \sqrt [n] {\sum_{k=0}^{n} {n\choose k} ^{t}}$$

where $t$ is a real number, using Stirling's formula:

$$n! \sim \sqrt{2\pi n} \left (\frac{n}{e} \right) ^{n} $$

I've already used Stirling's formula to simplify the binomial coefficient a little bit, but I'm still far away from achieving a reasonable answer.

Any help will be appreciated.

  • 1
    $\begingroup$ What did you do so far? $\endgroup$ May 8, 2021 at 17:11
  • $\begingroup$ I just applied Stirling's formula to the binomial coefficient. I don't know what to do with the sum. $\endgroup$
    – Albert
    May 8, 2021 at 17:13
  • $\begingroup$ Probably worth estimating the largest binomial coefficient, $\binom{n}{\lfloor n/2\rfloor}.$ $\endgroup$ May 8, 2021 at 17:17

1 Answer 1


First when $t\geq 0.$

Since: $$\binom{n}{\lfloor n/2\rfloor}^t\leq\sum_{k=0}^n\binom nk ^t\leq(n+1)\binom n{\lfloor n/2\rfloor}^t$$

and $(n+1)^{1/n}\to 1,$ we get, by the squeeze theorem, that your limit is the same as:

$$\left(\lim_{n\to\infty}\binom n{\lfloor n/2\rfloor}^{1/n}\right)^t$$

Use Stirling to approximate $\binom n{\lfloor n/2\rfloor}.$ You will get a slightly different estimate for when $n=2m$ and $n=2m-1,$ but the $n$th root will wash out that difference.

When $t<0,$ it is much easier, because $$1=\binom n0^t\leq \sum_{i=0}^n \binom{n}i^t\leq (n+1)$$

  • $\begingroup$ Thank you for your answer. I forgot that theorem from Calculus so I had to check it out! By the way, in the first case, where $t \geq 0$, I guess the final result is $2^{t}$ right? $\endgroup$
    – Albert
    May 8, 2021 at 18:13
  • 1
    $\begingroup$ Yes, that is correct. @Albert $\endgroup$ May 8, 2021 at 18:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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