7
$\begingroup$

Given two finite posets $P$ and $Q$, we can form the direct product poset $P \times Q$ whose elements are pairs $(p,q) \in P \times Q$ with $(p,q) \leq (p',q')$ if $p \leq p'$ and $q \leq q'$. Let us say a finite poset $P$ on $\geq 2$ elements is indecomposable if $P=P_1 \times P_2$ implies that either $P_1$ or $P_2$ is equal to $P$.

Is it true that a finite poset $P$ on $\geq 2$ elements can be written as a product $P = P_1 \times \cdots \times P_n$ of indecomposable posets in a unique way up to permutation of the factors?

Surely this is a classical question/result. I am apparently having trouble coming up with the right terms to google for, however. I would definitely appreciate any pointer to the literature. If the general result fails, I would also be interested in what mild conditions we can put on $P$ (e.g. gradedness) to guarantee a positive result.

$\endgroup$
4
  • 2
    $\begingroup$ If the posets in question all have a global minimum and a global maximum, then this follows from Lemma 6.1 in: William R. Schmitt, Incidence Hopf algebras, home.gwu.edu/~wschmitt/papers/iha.pdf . $\endgroup$ Apr 28, 2016 at 0:16
  • 1
    $\begingroup$ Actually, a positive answer to your question in the case where $P$ is connected is claimed in: Junji Hashimoto, On Direct Product Decomposition of Partially Ordered Sets, Annals of Mathematics, Second Series, Vol. 54, No. 2 (Sep., 1951), pp. 315--318, sci-hub.io/10.2307/1969532# . $\endgroup$ Apr 28, 2016 at 0:58
  • 1
    $\begingroup$ In full generality, your question has a negative answer: See Tadasi Nakayama and Junji Hashimoto, On a problem of G. Birkhoff, Proc. Amer. Math. Soc. 1 (1950), pp. 141--142, ams.org/journals/proc/1950-001-02/S0002-9939-1950-0035279-X . $\endgroup$ Apr 28, 2016 at 0:59
  • 1
    $\begingroup$ Darij, those last two references answer my question perfectly: false in general, true when $P$ is connected. You could post it as an answer. $\endgroup$ Apr 28, 2016 at 1:10

1 Answer 1

8
$\begingroup$

Basically repeating what I said in the comments:

(Note: I have verified the counterexample, but I haven't looked at the proof.)

$\endgroup$
1
  • 2
    $\begingroup$ For some recent work on arithmetic of finite ordered sets, Ralph McKenzie has a couple of papers from 2000 showing that one can cancel exponents for certain bases and exponents. Gerhard "Exponents Aren't Just For Numbers" Paseman, 2016.04.27. $\endgroup$ Apr 28, 2016 at 2:44

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.