Search

We investigate whether the set of subfit elements of a distributive semilattice is an ideal. This question was raised by the second author at the BLAST conference in 2022. We show that in general it has a negative solution, however if the semilattice is a lattice, then the solution is positive. This is somewhat unexpected since, as we show, a semilattice is subfit if and only if so is its distributive lattice envelope., Comment: 14 pages, 1 figure

We generalize the concept of the pointwise supremum of real-valued functions to the pointfree setting. The concept itself admits a direct and intuitive formulation which makes no mention of points. But our aim here is to investigate pointwise suprema of subsets of $\mathcal{R}L$, the family of continuous real valued functions on a locale, or pointfree space. Our setting is the category $\mathbf{W}$ of archimedean lattice-ordered groups ($\ell$-groups) with designated weak order unit, with morphisms which preserve the group and lattice operations and take units to units. A main result is the appropriate analog of the Nakano-Stone Theorem: a (completely regular) locale $L$ has the feature that $\mathcal{R}L$ is conditionally pointwise complete ($\sigma $-complete), i.e., every bounded (countable) family from $\mathcal{R}L$ has a pointwise supremum in $\mathcal{R}L$, iff $L $ is boolean (a $P$-locale). We adopt a maximally broad definition of unconditional pointwise completeness ($\sigma$-completeness): a divisible $\mathbf{W}$-object $G$ is pointwise complete ($\sigma$-complete) if it contains a pointwise supremum for every subset which has a supremum in any extension. We show that the pointwise complete ($\sigma$-complete) $\mathbf{W}$-objects are those of the form $\mathcal{R}L$ for $L$ a boolean locale ($P$-locale). Finally, we show that a $\mathbf{W}$-object $G$ is pointwise $\sigma$-complete iff it is epicomplete.