Skip to content

Semigroup with Involution

An associative algebraic system with a self-undoing unary operation that reverses product order.

Version
v1 · 2026-10-03 · History
Domain-specific #
13600
Domain group
Formal Sciences
Origin domain
Mathematics
Subdomain
Semigroup Theory → Mathematics
Aliases
Involution Semigroup, Involutory Semigroup, Star Semigroup

Core Idea

A semigroup with involution (also called an involution or star-semigroup) is a semigroup \(S\) with a selected unary operation \(x\mapsto x^*\) satisfying two laws:

\[ (x^*)^*=x,\qquad (xy)^*=y^*x^*\quad\text{for all }x,y\in S. \]

The underlying product is closed and associative. The first star law makes the operation self-undoing; the second makes it an anti-automorphism, reversing factor order. These conditions describe a formal reversal of multiplication. They do not say that \(x^*\) is a multiplicative inverse of \(x\): a general semigroup need not have an identity at all, and even a star-monoid need not satisfy \(xx^*=x^*x=1\).[1][2]

The distinction matters because different mathematical carriers exhibit the same laws. Transposing square matrices and taking the converse of binary relations both reverse composition, yet a singular matrix or a one-way relation need not be invertible. The object is therefore the carrier, product and chosen star together, not an isolated involutive function and not the whole theory of inverse semigroups.[1][3]

Structural Signature

Sig role-phrases: associative carrier → total unary star → double-star identity → order-reversing product identity → optional stronger subtype axioms.

  • Associative carrier. A nonempty set \(S\) is closed under a binary product whose parentheses can be reassociated without changing its value. Without this semigroup structure the named algebra is not defined.[2]
  • Total unary star. Every \(x\in S\) has a designated \(x^*\in S\). The star is part of the signature; it cannot be inferred from the presence of an associative operation alone.[1]
  • Double-star identity. Applying star twice returns \(x\). This makes star bijective and its own inverse as a function; it says nothing by itself about multiplicative inverses.[2]
  • Order-reversing product identity. Star sends \(xy\) to \(y^*x^*\). Thus \((xyz)^*=z^*y^*x^*\) follows by associativity and repeated application. A self-inverse map that fails this identity is not an involution of the semigroup in this sense.[1]
  • Optional stronger subtype axioms. An identity element, regularity condition, projection law or inverse-semigroup property may be imposed later. They locate narrower classes but are not fifth and sixth defining laws of the broad structure.[3][2]

What It Is Not

It is not a group disguised by notation. A group's inversion map is one example: \((xy)^{-1}=y^{-1}x^{-1}\) and \((x^{-1})^{-1}=x\). But for the full \(2\times2\) real matrix semigroup with transpose as star, a singular \(A\) has no multiplicative inverse. Even an invertible matrix can have \(A^{\mathsf T}\ne A^{-1}\). The star equations alone never license cancellation by \(x^*\).[1][4]

It is not the same as an inverse semigroup. That narrower class gives every element a unique semigroup inverse in the regular sense. Easdown and Munn's original result invokes further hypotheses to characterize it among involution semigroups. The mere existence of a star does not provide the inverse law \(xx^*x=x\).[3]

It is not an involutive monoid by default. A monoid has a two-sided identity; a semigroup need not. Conversely, a monoid need not come with any star. It is also much weaker than a C*-algebra, which requires complex linear, normed and completeness structure in addition to a compatible involution.

It is also not a catch-all for *-regular, Baer or Foulis semigroups, or for congruences constructed on involution semigroups. Those names impose additional laws or identify other objects; sharing this algebraic setting does not make them aliases of the broad structure.[3]

Scope of Application

The identity covers associative systems equipped with a reversal of multiplication: matrix semigroups under transpose, relation semigroups under converse, groups under inversion, and suitable free-word semigroups under reverse-and-star constructions. Which instances exist and which extra equations they satisfy depend on the selected carrier and star, not the name of the field in which they appear.[1][4]

For matrices, transposition is a natural star on all square matrices of a fixed size over a field: \((AB)^{\mathsf T}=B^{\mathsf T}A^{\mathsf T}\). For binary relations on a set, converse reverses every ordered pair and obeys \((R\circ T)^{-1}=T^{-1}\circ R^{-1}\) under the usual convention for composition. Zhang, Gao and Luo explicitly link the Boolean-matrix and finite-relation descriptions. Neither example needs to restrict to invertible elements.[1]

One must check that a proposed subcollection is closed under both operations. Upper-triangular matrices, for example, are not generally closed under ordinary transpose, although another chosen operation might make a particular submonoid involutive. The algebra is the triple \((S,\cdot,*)\), not simply the product set viewed without its domain restrictions.[1]

Clarity

The word involution has two meanings that can be conflated: an arbitrary self-inverse map, and an involutive anti-automorphism of a product. This entry requires both self-inverse behavior and reversal of product order. A calculation such as \(f(f(x))=x\) is therefore only half the test. Conversely, the product-reversal law alone does not establish double-star recovery.[2]

It also clarifies what “formal inverse behavior” buys. It allows one to reverse a product and then restore it by applying star again; it does not assert that multiplying an element by its star returns a neutral element. In the matrix example \(AA^{\mathsf T}\) can be a nonidentity positive-semidefinite matrix, and for a relation \(R\), \(R\circ R^{-1}\) need not be the identity relation.[1]

Manages Complexity

The two star equations compress many concrete reversal calculations into one equational schema. Once they hold on a closed associative carrier, the star of any finite product is the starred factors in reverse order. A proof can therefore distinguish statements true in every involution semigroup from statements requiring a monoid, a group, a regular star-semigroup, or a particular matrix or relation model.[1][3]

This compression is disciplined rather than expansive. The same abstract laws cover both matrix transpose and relation converse, but their element-level properties differ. Matrix rank, relation domain/range and positivity are not consequences of the bare two equations. Carrying such conclusions from one model to all models would lose the very assumption boundary that the algebraic abstraction is meant to expose.[1]

Abstract Reasoning

To classify a proposed system, type the carrier and verify closure and associativity of its product. Check that star is total on that carrier, then prove \((x^*)^*=x\) and \((xy)^*=y^*x^*\) for arbitrary elements. If those conditions hold, an induction yields \((x_1x_2\cdots x_n)^*=x_n^*\cdots x_2^*x_1^*\). This is a legitimate inference without an identity or inverse law.[2]

To decide whether an intended stronger result follows, look for the extra hypothesis. For example, \(xx^*=1\) is false for \(A=\begin{pmatrix}1&1\\0&0\end{pmatrix}\) under transpose. A theorem that needs this equation is about a narrower class; relabeling the example does not repair the proof. In a relation model, similarly test whether converse actually undoes a relation or merely reverses its arrows.[1]

Knowledge Transfer

The identity transfers literally across algebraic carriers because the same equations can be checked for matrices, relations, groups and words. A star on one carrier can be represented differently from a star on another while the two laws remain identical. But the transfer is mathematical: it does not make the named algebra a general metaphor for reversing a process or undoing a decision.[1][4]

The live Semigroup prime carries the broader closed-associative product that every instance here has. The child adds an order-reversing, self-inverse unary map. The proposed strict subsumption edge is therefore justified by forgetting star; a separate edge to Group or Monoid would be false because their identity and inverse commitments are not inherited by all involution semigroups.

Examples

Square matrices under transpose. Let \(S=M_2(\mathbb R)\), all \(2\times2\) real matrices, with ordinary multiplication and \(A^*=A^{\mathsf T}\). Matrix multiplication is closed and associative, and the two transpose identities are exactly the star laws. For \(A=\begin{pmatrix}1&1\\0&0\end{pmatrix}\), however, \(AA^{\mathsf T}=\operatorname{diag}(2,0)\ne I\). This exact calculation is derived from the original paper's matrix model and shows that star is not general group inversion.[1]

Mapped back: associative carrier = \(M_2(\mathbb R)\) under multiplication; total unary star = transpose; double-star identity = transpose twice returns \(A\); order-reversing product identity = \((AB)^{\mathsf T}=B^{\mathsf T}A^{\mathsf T}\); optional stronger subtype axioms = the carrier happens to have \(I\), but a singular \(A\) fails multiplicative inversion.

Binary relations under converse. Let \(X=\{1,2\}\) and \(S\) be all binary relations on \(X\), with relational composition and \(R^*=R^{-1}\) formed by reversing every ordered pair. Converse twice restores \(R\), and converse reverses the order of composition. For the one-way relation \(R=\{(1,2)\}\), the composites \(R^{-1}\circ R\) and \(R\circ R^{-1}\) are different one-point diagonal relations; neither is the identity relation \(\{(1,1),(2,2)\}\). The original Boolean-matrix source identifies transpose with this converse operation.[1]

Mapped back: associative carrier = endorelations on \(X\) under composition; total unary star = converse; double-star identity = pair reversal twice restores \(R\); order-reversing product identity = converse of a composite reverses its factors; optional stronger subtype axioms = the identity relation exists, but not every relation has an inverse.

Boundary negative. All square matrices under multiplication, considered without a specified star, form a semigroup but not yet a specified semigroup-with-involution instance. The star and its two equations have to be supplied.

Structural Tensions

Formal reversal versus actual undoing. The weak star laws preserve a useful reversal syntax across noninvertible matrices and relations. Requiring actual undoing would sharply narrow the class; assuming it without proof yields false cancellation. On the other hand, retaining only the weak laws limits inverse-style conclusions. Diagnostic: Does the proposed argument use merely \((xy)^*=y^*x^*\), or does it secretly need \(xx^*=1\) or \(xx^*x=x\)?[1][3]

Minimal common variety versus stronger subtype theorems. The broad algebra joins disparate examples under two equations, making general proofs portable. Adding regularity, identity or inverse assumptions can yield stronger results, but then some original models or elements leave the class. Keeping the broad label while importing those stronger conclusions makes a theorem appear more general than it is. Diagnostic: Which extra equation is indispensable, and has it been checked for every claimed model?[2][3]

Structural–Framed Character

The entry is predominantly structural within a mathematical frame. Its vocabulary travels among matrix, relation, group and word structures through literal equations, not a superficial analogy. It carries almost no evaluative weight: an involution semigroup need not be useful, symmetric in an everyday sense, or invertible. Its institutional origin is mathematical naming and proof practice, not a rule created by a social institution. Its identity is not dependent on human participation once the carrier and operations are defined. Applying the term recognizes a checkable algebraic structure, although one must import the formal choice of product and star before testing it.[1]

The portable skeleton is already represented by the live Semigroup prime; the chosen anti-automorphism adds a specialist algebraic tier. Mathematical variety across examples is not automatically the encyclopedia's prime criterion across independent substrates. Its character: a structurally crisp, domain-specific formal object under the Semigroup prime, with no evaluative or institutional criterion in its defining laws.

Structural Core vs. Domain Accent

The inherited skeletal relation from Semigroup is closure plus associativity of a binary product. The domain accent is a second, unary operation governed by two exact equations: self-recovery and anti-automorphism. Matrix transpose, relation converse and group inversion are representations of that accent, not its definition. The failure boundary is visible when either star equation fails, or when an alleged result requires actual inverses not granted by them.[2]

Removing the star laws leaves only the parent semigroup. Removing the algebraic carrier and asking whether a narrative “reverses” a prior action leaves only metaphor. Thus the named child does not itself clear the prime bar even though the parent semigroup travels broadly. No separate future-prime identity is claimed here.

This entry is a kind of Semigroup.

The broader abstraction is Semigroup by strict subsumption. Every admitted instance remains a semigroup after forgetting star, while some semigroups have no selected star. Monoid is a related neighboring tier: matrix and relation examples are monoids, but the definition here does not require an identity. Group supplies one special implementation of star as actual inversion, not a parent of the whole class. The structured frontmatter asserts only the Semigroup edge.[4][1]

The live C*-Algebra is a much stronger specialist neighbor with additional operations and norm/completeness laws. Its star-bearing multiplicative reduct fits the two equations, but this broad entry should not inherit operator-algebra conclusions. Narrower star-semigroup species and congruences require separate identities rather than promotion as aliases of this broad class.

Relationships to Other Abstractions

Local relationship map for Semigroup with InvolutionParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Semigroup withInvolutionDOMAINPrime abstraction: Semigroup — is a kind ofSemigroupPRIME

Current abstraction Semigroup with Involution Domain-specific

Parents (1) — more general patterns this builds on

  • Semigroup with Involution is a kind of Semigroup Prime

    Forgetting the involution leaves the child's closed associative multiplication, a semigroup.

Hierarchy paths (4) — routes to 4 parentless roots

Neighborhood in Abstraction Space

Semigroup with Involution sits in a moderately populated region (57th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Algebraic Structures & Order Relations (18 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With

  • Plain semigroup: associative multiplication alone; no required star.
  • Involutive monoid: the same star laws plus a two-sided identity, so not a synonym for the broader class.
  • Inverse semigroup: unique regular inverses; not obtained solely from the two star laws.[3]
  • Baer or Foulis semigroup, *-regular semigroup: named subtypes with further conditions, not aliases of the general structure.
  • Dyck or Shamir congruence and free half group: named quotient/construction questions, not alternative names for the algebraic species.
  • An arbitrary order-two map: \(f(f(x))=x\) alone misses the anti-automorphism condition.

References

[1] Wen-Ting Zhang, Meng Gao and Yan-Feng Luo, “Equational Theories of the Boolean Matrix Monoid BRn with Involutions,” Journal of Algebra 685 (2026), pp. 225–270, original publisher article preview, Introduction definition and Boolean-matrix/relation paragraphs, Section 3 isomorphism snippet. The article preview supplies the structural claims; numerical two-element illustrations are directly calculated from its equations. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r

[2] Edmond W. H. Lee, “Embedding Finite Involution Semigroups in Matrices with Transposition,” Discrete Applied Mathematics 340 (2023), pp. 327–330, original publisher abstract and ‘Inverse semigroups’ snippet. Full article text was not inspected. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h

[3] D. Easdown and W. D. Munn, “On Semigroups with Involution,” Bulletin of the Australian Mathematical Society 48(1) (1993), pp. 93–100, original publisher extract. Only extract-level inverse-subtype claims are used. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h

[4] W. D. Munn, “Special Involutions,” in Semigroup Theory and its Applications (1996), pp. 157–165, original publisher chapter preview, introductory definition and Examples 1–2. This supports group and free-semigroup examples, not claims about every star-semigroup. registry ↩a ↩b ↩c ↩d