Is there a finite lattice such that every join-irreducible element is left-modular and that is not semimodular?

I've been able to prove there are no such bounded atomistic lattices (not necessarily finite), using the fact that such a lattice is semimodular iff it is geometric.


I argue the contrapositive, that if $L$ is a finite lattice that is not semimodular, then $L$ has a join irreducible that is not left modular.

Since $L$ is not semimodular, it contains elements $a, b$ such that $a\wedge b\prec a$ but $b\not\prec a\vee b$. Choose any $c$ such that $b < c < a\vee b$. Now choose any $j\leq a$ minimal for the property that $j\not\leq b$; $j$ is join irreducible. Moreover, $j$ is not left modular, since $(j,c)$ is not a modular pair. To verify this last claim, compute that

  • $b\vee (j\wedge c) = b\vee (j\wedge a\wedge c) = b\vee (j\wedge a\wedge b) = b$,
  • $(b\vee j)\wedge c = (b\vee (b\wedge a)\vee j)\wedge c = (b\vee a)\wedge c = c$,
    and these are different.

  • $\endgroup$
    • $\begingroup$ Is finiteness of $L$ used in any step or is it sufficient to provide that every down-directed set has a minimum? $\endgroup$ – Michał Zapała Dec 9 at 18:35
    • 1
      $\begingroup$ @MichałZapała: The only place I use finiteness is where I assume that the difference of intervals $(a] - (a\wedge b]\;\; (= \{x\leq a|x\not\leq b\})$ has a minimal element. (This is not a down-closed set.) $\endgroup$ – Keith Kearnes Dec 9 at 19:07

    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.