In number theory, a positive integer $n$ is called highly composite if it has more divisors than any smaller positive integer. This notion has been studied by several notable mathematicians; for instance, Ramanujan and Erdos.

We can define a group-theoretical analogue of this. Let's say a positive integer $n$ is groupy if there are more groups of order $n$ (up to isomorphism) than any smaller positive integer. Denote the $k$th groupy number by $g(k)$.

The first few groupy numbers are $1, 4, 8, 16, 24, 32, 48, 64, 128, 256, 512, 1024, 2048$ and indeed we have an OEIS sequence, namely A046059. Many powers of $2$. We have the natural conjecture that for $k \geq 8$, we have $g(k) = 2^{k-2}$. However, we do have "outliers" $24$ and $48$, and it's possible that there might be infinitely many such outliers. A weaker conjecture is that the set of indices $k$ for which $g(k)$ is not a power of $2$ has zero asymptotic density.

Has this sequence been studied in any depth before?

  • 1
    $\begingroup$ Between the idea that "most" groups are nilpotent and the Higman-Sims estimate for the number of groups of order $p^n$, I'd be surprised if $g(k)$ isn't just $2^{k-2}$ for sufficiently large $k$ (though not necessarily $k\geq 8$). $\endgroup$ – anomaly Dec 28 '18 at 21:20
  • $\begingroup$ @anomaly [nitpick] I think you mean to say it would be $g(k) = 2^{k-M}$ for sufficiently large $k$ and some fixed $M$. If there are more outliers (beyond $k=8$), then we can't be sure it'll be $2^{k-2}$ anymore. $\endgroup$ – MathematicsStudent1122 Dec 28 '18 at 21:26
  • 1
    $\begingroup$ It's a folklore conjecture that the vast majority of groups are nilpotent of order a power of $2$; if this is the case then there aren't any more outliers. $\endgroup$ – Qiaochu Yuan Dec 28 '18 at 21:37
  • $\begingroup$ Related: math.stackexchange.com/q/241369/407165 $\endgroup$ – Yanior Weg Apr 11 at 16:25

Your Answer

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

Browse other questions tagged or ask your own question.