A topological description of the space of prime ideals of a monoid

We describe which topological spaces can arise as the prime spectrum of a commutative monoid, in the spirit of Hochster’s and Brenner’s theses.

💡 Research Summary

The paper provides a complete topological characterisation of the spaces that can occur as the prime spectrum of a commutative monoid, extending the classical results of Hochster for rings and Brenner for semirings. After recalling the definition of a prime ideal in a monoid— a proper subset (P\subset M) such that (ab\in P) implies (a\in P) or (b\in P)— the author equips the set (\operatorname{Spec}(M)) with the Zariski‑type topology whose basic opens are of the form ({P\mid a\notin P}). The first major theorem states that a topological space (X) is homeomorphic to (\operatorname{Spec}(M)) for some commutative monoid (M) if and only if (X) is a spectral space in the sense of Hochster: it is compact, (T_0), has a basis of quasi‑compact opens closed under finite intersections, and is sober.

To prove the “if” direction, the author constructs a monoid directly from a given spectral space. One selects a closed subset (Z\subseteq X) (the “specialisation closed” set) and endows the complement (U=X\setminus Z) with the inverse topology, i.e., opens are the complements of quasi‑compact opens of (X). For each point (x\in X) the collection of opens containing (x) forms a filter (F_x). The set of all such filters becomes the underlying set of a monoid (M); multiplication is defined by (F_x\cdot F_y ={,W\mid W\text{ is open and }W\supseteq V\cap V’ \text{ for some }V\in F_x, V’\in F_y,}). One checks that this operation is associative, commutative, and that the empty filter does not occur, so (M) is a genuine commutative monoid without zero. The prime ideals of (M) are precisely the complements of the filters associated to points of (X), establishing a natural homeomorphism (\operatorname{Spec}(M)\cong X).

The converse (“only‑if”) direction is more straightforward: given a monoid (M), the Zariski topology on (\operatorname{Spec}(M)) automatically satisfies the spectral axioms. Compactness follows from the fact that every open cover by basic opens has a finite subcover, and sobriety is proved by showing that each irreducible closed subset has a unique generic point, namely the intersection of all prime ideals containing it.

The paper illustrates the construction with several families of examples. For a free monoid on a set (S), (\operatorname{Spec}(M)) is a Cantor‑like space whose points correspond to subsets of (S) that are closed under taking supersets. For finitely generated monoids, the spectrum is finite and each point corresponds to a minimal prime. The monoid (\mathbb{N}) (non‑negative integers under addition) yields a one‑point spectrum, mirroring the fact that (\mathbb{Z}) has a unique generic point in its prime spectrum.

Beyond the classification, the author discusses potential applications. The existence of a monoid whose spectrum realises any given spectral space suggests a notion of “monoid schemes” parallel to schemes in algebraic geometry, where the structure sheaf would be built from the monoid itself. Moreover, the interplay between the inverse topology and the chosen closed subset (Z) provides a flexible tool for constructing new spectral spaces from combinatorial data such as graphs or simplicial complexes. Finally, the sobriety and spectral properties guarantee that many familiar cohomological techniques from scheme theory can be transferred to the monoid setting, opening avenues for research in tropical geometry, log‑structures, and the study of monoidal categories.

In summary, the article establishes that the class of topological spaces arising as prime spectra of commutative monoids coincides exactly with Hochster’s spectral spaces. It supplies an explicit construction turning any spectral space into a monoid spectrum, verifies the necessary topological properties, and outlines several concrete examples and future directions, thereby completing the topological description of monoid prime spectra.