nLab multiplicatively idempotent semiring

Redirected from "multiplicatively idempotent rig".

This is about idempotent semirings whose multiplication is idempotent. For idempotent semirings whose addition is idempotent, see additively idempotent semiring.

This article is about Boolean semirings as defined by Toby Bartels. For other notions of “Boolean semiring” or “Boolean rig”, see Boolean semiring.


Contents

Idea

Recall that a semiring is a set RR equipped with two binary operations, denoted ++, and \cdot and called addition and multiplication, satisfying the ring (or rng) axioms except that there may or may not be either be a zero nor a negative nor an inverse, for which reason do check.

Definition

A semiring or rig is (multiplicatively) idempotent or Boolean if and only if multiplication is idempotent; that is, x 2=xx^2 = x holds for all xSx \in S.

Terminology

From now on we will assume that semirings and rigs are synonyms of each other; i.e. the semiring, SS, has a neutral element ε\varepsilon for + and one ee for \cdot. Moreover we assume that for all sSs\in S, sε=εs=εs\cdot \varepsilon =\varepsilon \cdot s = \varepsilon, making SS into a rig.

Multiplicatively idempotent semirings are sometimes called idempotent rigs, see e.g. Baez 2022. However, the term idempotent rig and idempotent semiring usually refers to additively idempotent semirings in the literature, and now redirects to the disambiguation page idempotent semiring on the nLab.

Multiplicatively idempotent semirings are called Boolean rigs or Boolean semirings by Toby Bartels, see e.g. Bartels 2020. The name originated from the fact that a Boolean ring is defined in the literature as a ring for which multiplication is idempotent (Bartels 2020, Rogers 2024). However, the terms “Boolean rig” and “Boolean semiring” have multiple meanings in the literature representing some generalization of Boolean rings from rings to rigs and semirings, and now redirects to the disambiguation page Boolean semiring on the nLab.

Properties

Let CMon be the concrete monoidal category of abelian groups, and let U:CMonSetU: CMon \to Set be the lax monoidal underlying-set functor.

Theorem

Every multiplicatively idempotent rig RR satisfies the equation

1 U(R)=(U(R)δU(R)×U(R)λU(RR)U(mult)U(R))1_{U(R)} = \left(U(R) \stackrel{\delta}{\to} U(R) \times U(R) \stackrel{\lambda}{\to} U(R \otimes R) \stackrel{U(mult)}{\to} U(R) \right)

where λ\lambda is a lax monoidal constraint.

Examples

The main examples are probably distributive lattices.

Of course, a Boolean ring is a multiplicatively idempotent semiring, since any ring is a semiring. However, since it's also a distributive lattice, a Boolean ring is actually a multiplicatively idempotent semiring in two different ways.

Similarly, a Boolean semiring as defined by Fernando Guzmán is a multiplicatively idempotent semiring by definition.

Smallest multiplicatively idempotent semirings

Let’s see what are the smallest multiplicatively idempotent semirings.

  • The only multiplicatively idempotent semiring of cardinality 11 is the zero ring 00.

There are exactly two multiplicatively idempotent semirings of cardinality 22. In such a multiplicatively idempotent semiring, we necessarily have 010 \neq 1, because 0=10=1 would imply that x=1x=0x=0x=1 \cdot x=0 \cdot x=0 for every xx and then the multiplicatively idempotent semiring would be the zero ring 00. The two elements of the multiplicatively idempotent semirings are thus 00 and 11. From the axioms of a semiring, we have 0+0=00+0=0, 1+0=0+1=11+0=0+1=1, 00=00 \cdot 0=0, 10=01=01 \cdot 0=0 \cdot 1=0 and 11=11 \cdot 1=1. We then have two possibilities for 1+11+1, either 00 or 11. The two possibilities give a multiplicatively idempotent semiring.

  • One of the two multiplicatively idempotent semirings of cardinality 22 is /2\mathbb{Z}/2\mathbb{Z} (00 can be interpreted as “False”, 11 as “True”, ++ as the exclusive disjunction and \cdot as the conjunction).
  • One of the two multiplicatively idempotent semirings of cardinality 22 is 𝔹={0,1}\mathbb{B}=\{0,1\} where 1+1=11+1=1 (00 can be interpreted as “False”, 11 as “True”, ++ as the disjunction and \cdot as the conjunction).

In either case, the multiplication is commutative and distributive over addition.

  • The three-element distributive lattice {0,u,1}\{0, u, 1\} with 0u10 \leq u \leq 1 is a multiplicatively idempotent semiring with cardinality 33.

  • The four-element distributive lattice {0,u,v,1}\{0, u, v, 1\} with 0uv10 \leq u \leq v \leq 1 is a multiplicatively idempotent semiring with cardinality 44.

  • The four-element distributive lattice {0,u,v,1}\{0, u, v, 1\} with 0u10 \leq u \leq 1 and 0v10 \leq v \leq 1 but uu and vv incomparable (i.e. ¬(uv)¬(vu)\neg (u \leq v) \wedge \neg (v \leq u)) is a multiplicatively idempotent semiring with cardinality 44.

  • The previous four-element distributive lattice is also a Boolean algebra, and so one can define a Boolean ring structure, resulting in yet another multiplicatively idempotent semiring with cardinality 44.

  • The initial multiplicatively idempotent semiring, i.e. the free multiplicatively idempotent semiring on the empty set, has cardinality 44 (Rogers 2024), and is defined as the quotient of the natural numbers by the relation n+4n+2n + 4 \sim n + 2. This semiring is commutative.

One can construct a non-commutative multiplicatively idempotent semiring of cardinal 99 (Rogers 2024). It is currently unknown if there are any smaller non-commutative multiplicatively idempotent semirings.

References

The term (multiplcatively) idempotent appears in:

The term idempotent rig appears in:

  • John Baez, Free Idempotent Rigs and Monoids, Azimuth Blog, 21 December 2022. [web]

The term Boolean rig was used by Toby Bartels in this current article on the nLab, before it was renamed to multiplcatively idempotent rig:

The term Boolean rig was also used in this Category Theory Zulip discussion:

  • J-B Vienney, Are Boolean rigs commutative?, Category theory Zulip, (web)

Last revised on June 14, 2025 at 11:45:20. See the history of this page for a list of all contributions to it.