2
$\begingroup$

I call a set X of positive integers strongly lcm-closed if a,b ∈ X if and only if lcm(a,b) ∈ X. In the finite case X is the set of divisors of lcmx ∈ Xx. But in the infinite case it is more interesting, for example, $\{a \geq 1: a \not\equiv 0 \pmod p\}$ and $\{p^a:a \geq 0\}$ for any prime p, are strongly lcm-closed sets.

Which sets are strongly lcm-closed sets?

This question arose in my Ph.D. thesis (p.107) where strongly lcm-closed sets describe where autotopisms of Latin squares give rise to subsquares.

As a side question:

Is there a common name for strongly lcm-closed sets?

$\endgroup$
1
  • $\begingroup$ Excellent! This is exactly what I'm after. By the way, I would accept both answers below, if I could. Thanks! $\endgroup$ Feb 21, 2010 at 21:51

2 Answers 2

5
$\begingroup$

Given a supernatural number $N$, the set of positive integer divisors of $N$ is a strongly lcm-closed set. And any nonempty strongly lcm-closed set $X$ arises in this way, with $N$ equal to the supernatural lcm of the $x$ in $X$. (See Serre, Galois cohomology for the notion of supernatural number: it is a formal product over primes, $\prod_p p^{n_p}$, where each $n_p$ is in $\lbrace 0,1,2, \dots,\infty \rbrace$.)

$\endgroup$
3
  • $\begingroup$ I do not see why every strongly-lcm set could be obtained this way, how about {p^2a} for p fix? $\endgroup$
    – domotorp
    Feb 21, 2010 at 14:43
  • $\begingroup$ Your example doesn't satisfy the definition: we have lcm(p,p^2) in X, so p and p^2 should be in X too. $\endgroup$ Feb 21, 2010 at 15:26
  • $\begingroup$ Oh, you are right. $\endgroup$
    – domotorp
    Feb 21, 2010 at 20:27
3
$\begingroup$

Those would be the ideals in the lattice of positive integers ordered by divisibility.

$\endgroup$

Your Answer

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

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