Finite first-order recognition of and from subsemigroup lattices
Problem
For nonempty semigroups , let be the lattice of all subsemigroups, including . Can free groups and free abelian groups be recognized by first-order lattice sentences, uniformly in their rank?
The language is ; may be empty or infinite. The result below gives one finite sentence for each class.
Setup and result
For a semigroup , write
The empty subsemigroup is included. We consider first-order sentences in the two-function language
An elementary class of lattices is a class defined by first-order sentences in this language. The lattice identities can always be adjoined to an axiom set; they impose no additional restriction on the lattices .
Shevrin proved finite lattice recognition results for torsion-free groups and for torsion-free abelian groups among semigroups (4; 5). At the end of (5, p. 684), he asked whether free groups and free abelian groups admit analogous first-order axiomatizations. These are the two questions in Kourovka Problem 2.81.
Theorem 1. There are -sentences and such that, for every nonempty semigroup ,
Here is the free group on , and is the group of finitely supported integer-valued functions on . The set may be empty or have any infinite cardinality.
Elements and bases will be represented inside the lattice by
where is the set of irreducibles of a submonoid . Generation is expressed through containment in subsemigroups and definable subgroups. The free-group argument reconstructs an oriented multiplication graph by Beth’s theorem. These constructions use finitely many variables independently of the rank.
Sections “Recognizing torsion-free groups”–“A finite sentence for free abelian groups” prove (3b), including the initial recognition results. Sections “Recovering unordered multiplication”–“The finite sentence for free groups” prove (3a). We write , and for semigroup, monoid and group generation, respectively. The cyclic group is permitted throughout.
Recognizing torsion-free groups
Order, strict order and covering are definable by
The symbols denote the uniquely definable bounds. Put
Every -code is monogenic. Indeed, if is its greatest proper subsemigroup and , then . Here
A monogenic semigroup with a repeated power is finite and contains an idempotent. Conversely, if all positive powers of are distinct, deleting gives the greatest proper subsemigroup of . Thus codes are exactly the infinite monogenic subsemigroups. Atoms are exactly singleton idempotents.
For lattice elements , define
Define by and the existence of -codes such that
Our local group guard is
This is a finite first-order formula. It is a variant of the cyclic-interval recognition criteria of Shevrin (3; 5).
Lemma 2. A subsemigroup of a monoid satisfies if and only if its multiplication makes it a torsion-free group, with an identity that need not equal the identity of the ambient monoid.
Proof. Suppose holds and write . If has repeated powers, then is finite and contains an idempotent, necessarily . Unless , the interval
contradicts . Hence every has distinct powers.
Fix an interval satisfying and write , . The preceding observation and (9b) imply
We first show . If neither cross product equals , both belong to . Whenever , induction gives
and whenever it gives . The corresponding formulas for show that is a subsemigroup. This contradicts . By symmetry, take . If , then , and gives
contrary to injectivity of powers. The case is identical with the roles reversed. Thus .
Associativity now gives and . If , then idempotence implies , hence . If , multiplying by gives , which is impossible. Therefore is either or , and similarly is either or . The identities exclude mixed choices. If absorbs both rays, then
is a proper subsemigroup of containing and , contrary to (9a). Consequently is the identity on , and are mutual inverses. Every element of is a positive power of one of them or is .
Every lies in such an interval. All these cyclic groups have the same identity . It follows that is an identity on and every element of has an inverse in . The injectivity of powers already proved gives torsion-freeness.
Conversely, in a torsion-free group the only idempotent is . Every subsemigroup properly containing contains a nonidentity element and hence an infinite monogenic subsemigroup. Set
The two rays are disjoint, , and gives . Every nonidentity element of lies on one ray, so every infinite monogenic subsemigroup of lies on one ray. Thus all the required conditions hold. The trivial group satisfies the same formula, with the conditions involving infinite codes vacuous. ◻
The ambient-monoid convention in Lemma 2 can be removed within the finite lattice language. Adjoin a new identity to an arbitrary semigroup , obtaining . There is an order isomorphism
Terms in the product are evaluated coordinatewise. A quantifier over its second coordinate is a conjunction or disjunction of two cases. Induction on formulas therefore eliminates the finite factor. Applying this translation to at gives a finite sentence such that
The multiplication in the reconstructed group is the original multiplication of .
Commutativity in the subsemigroup lattice
For the remainder of this section let be a torsion-free group. Its -codes are exactly the subsemigroups
and is a bijection. Indeed, gives , with . If , injectivity of its powers gives . The identity is represented by the unique atomic code . In formulas below, quantification over elements means quantification over -codes; membership means .
Subgroups are definable by
To verify soundness, a positive word in equal to must contain , since has infinite order. Writing it as gives . Completeness follows by taking . Hence the least subgroup containing is definable and equals .
The inverse code is definable as well. For , is the greatest nonidentity element code in disjoint from . A nonidentity power has this disjointness precisely when . The identity code is excluded here; it is also disjoint from . At , inversion fixes .
Lemma 3. Let and let be the greatest proper subsemigroup of . Then if and only if is an element code and there are element codes such that
Proof. Here . If , write and let generate , where . Representing in the join forces
For , choose . The join has exponent set , so adding its only missing element gives .
Suppose and . The generator of is with , since would make the join equal to . This join lies in , and . The covering relation forces the join to equal . But representing both and with the exponents forces respectively and . This is impossible. ◻
The identity-code case completes a finite formula for . This is the square-code construction of (4, Lemma 4).
Write . Define the special-product relation by
The cardinality condition means the existence of three distinct element codes below . All parts are finite lattice formulas. A shortest positive word for proves
Indeed, the three-element condition supplies an element different from both generators; the defining implications then exclude and . A shortest word for therefore has length at least two. Removing its first letter gives a different element , and the defining implication identifies with the other generator.
We use the following local predicate, corresponding to (4, Lemma 6). Let assert:
-
, , and .
-
There is exactly one with .
-
Every satisfies for some , .
-
If , , , and
then and .
-
If , , , and , then .
Lemma 3 makes the fourth and fifth conditions finite lattice formulas.
Lemma 4. In a torsion-free group, implies . In a torsion-free abelian group, its first condition implies all five conditions.
Proof. For soundness, symmetry and (24) allow the unique special product to be . Suppose . Then , since either membership would imply commutativity. The third condition and (24) give
The case contradicts uniqueness of . The second equality gives
contrary to the first condition. In the first case,
If this square lies in neither original ray, condition 4 gives , , and hence , another contradiction. The square therefore lies in or . Conjugation gives
Using the expression appropriate to that ray gives . Condition 5 now puts in an original ray, the final contradiction.
For necessity, work in an abelian group. The first condition gives
It also gives
A zero exponent in (31) would put one generator in the other positive ray, or make it the identity. With both exponents positive, cancellation reduces the equality to the preceding positive identity relation.
To check , write and . Equation (31) gives
Thus . If , then , so . The other defining implication is symmetric. Uniqueness follows from (24).
For a mixed element , both exponents are positive. Neither nor is a positive power of the other, since that would put in an original ray. The same argument gives the required special product.
For condition 4, write , with . A representation
has because the square lies in neither ray. Cancellation gives
so and . Finally, squaring is injective in a torsion-free abelian group: gives , hence . Thus condition 5 holds. ◻
Define by requiring, for every such that
that or holds. All excluded pairs already commute. Lemma 4 therefore proves that implies commutativity. Conversely, in a torsion-free abelian group, both and cannot contain : equations
would give . Hence one orientation satisfies the first condition of , and the lemma applies. We have proved
A finite sentence for free abelian groups
Put
and define
In an abelian group, (24) gives
For the second disjunct, , , and cancellation gives the conclusion. Let be the finite sentence .
Lemma 5. Every free abelian group of rank at least two satisfies .
Proof. Use additive notation. If are independent over , independent integer coordinates prove by counting the two letters in a positive word. The identity cases are immediate.
Suppose are dependent. Choose independent of ; such a exists because the rank is at least two. A nonzero relation has , and it shows that
are independent whenever their first entries are nonzero. For example, from one obtains
so and then . Taking and proves the second disjunct of (39). If , its last partial product is an identity case. ◻
Assume now that is a torsion-free abelian group satisfying . By (40), defines its multiplication on element codes. For a submonoid , let
The following properties form a finite lattice formula :
Here means for some . Products and divisibility use . The first equality in (44b) is expressed by
The second uses the subgroup formula (20): every subgroup containing is . Thus no quantification over arbitrary subsets or over lists is introduced into the sentence.
Lemma 6. If an abelian group contains a submonoid satisfying (44a)–(44c), then is free abelian on . Conversely, the nonnegative cone of a free abelian group has these properties.
Proof. A divisor in of another element of equals that element: the remaining factor must be a unit, hence by (44a). Primality therefore implies that an irreducible dividing a finite product of irreducibles occurs among its factors. Cancelling that factor and inducting on the number of factors proves that the product map
is injective. It is surjective by (44b). The induced group homomorphism
is injective: split a finitely supported integer vector into its positive and negative parts, and apply injectivity of the positive product map. Its image contains , so it is surjective by the second part of (44b).
For , take . Its only unit is , its irreducibles are the standard basis vectors, and each is prime by the corresponding nonnegative coordinate. These vectors generate and generate as a group. ◻
Let assert that the greatest lattice element is the least subgroup containing some element code. In a torsion-free group this means that the group is trivial or infinite cyclic. Define
Theorem 7. For every nonempty semigroup ,
In particular, the class in part (b) of Kourovka Problem 2.81 is recognized by a single finite sentence in the pure lattice language.
Proof. If holds, (18) reconstructs a torsion-free group with the given multiplication, and (37) makes it abelian. The cyclic alternative is free abelian of rank zero or one. Otherwise (40) and identify the multiplication used in with the group multiplication, and Lemma 6 applies.
Conversely, a free abelian group satisfies and . In rank zero or one it satisfies . In every other rank, Lemma 5 supplies and the nonnegative cone supplies . A semigroup isomorphism induces an isomorphism of subsemigroup lattices and preserves the sentence. ◻
Recovering unordered multiplication
The special-product relation of Section “Commutativity in the subsemigroup lattice” is also useful without commutativity. Let
This is a finite lattice formula: the least subgroup containing an element code is definable by (20). Its truth implies .
We impose two finite conditions on a torsion-free group :
Here means that is a cyclic subgroup, rather than the global sentence . Denote their conjunction by .
Lemma 8. If a torsion-free group satisfies , every nonidentity centralizer is cyclic, and
Every free group satisfies .
Proof. If commute, (24) gives at most one special product. Thus (51a) implies . The reverse implication always holds. Consequently the subgroup in (51b) is exactly .
In a free group, for , so commuting pairs satisfy . If , the subgroup theorem for free groups gives
Indeed, the rank is at most two by generation and is not at most one by noncommutativity. The images of generate and form an integer basis. Every positive word equal to or has exactly one occurrence of each generator. The defining implications for follow by substituting a positive word for the intermediate element and counting the two letters. Hence
which establish (51a). The centralizer itself supplies (51b). ◻
Lemma 9. Let be torsion-free, with cyclic centralizers of nonidentity elements. If , and , then . In particular, conjugate elements of the same nonidentity centralizer are equal.
Proof. Elements commuting with the same nonidentity element lie in a cyclic group and commute with each other. Thus
Conjugation by preserves . Its effect on is therefore or . In the first case commutes with and hence with .
In the second case , and by torsion-freeness. Both and commute with , so they belong to its cyclic centralizer and commute. This contradicts . Finally, if and , the first assertion puts in the cyclic group , giving . The case is immediate. ◻
Continue to write for the partial product defined before (39). Define as the disjunction
All variables range over element codes. Thus is a fixed finite lattice formula.
Proposition 10. For a torsion-free group satisfying ,
For every noncyclic free group, the converse also holds.
Proof. The first two alternatives follow from (24). In (56c), let . Then , , and . The partial-product witnesses give
Since , the first and last possibilities are and . Equality or gives by cancellation. Equality or makes conjugate to inside , so Lemma 9 gives the same conclusion, including .
Each remaining equality contradicts malnormality. More explicitly,
In all four cases a nonidentity element of is conjugated into by or , contrary to .
For completeness in a noncyclic free group, identity pairs and noncommuting pairs were treated in Lemma 8. Suppose commute and . Choose ; otherwise would be cyclic. Cyclic centralizers imply that does not commute with , and that does not commute with . If , then also does not commute with . Therefore
satisfy all three partial products in (56c), using the special products for these noncommuting pairs. When , the final partial product is an identity case. This proves completeness. ◻
Subsemigroup cuts and free bases
The next criterion uses no cyclic-centralizer assumption in its sufficient direction. Let have infinite order and let be a family of elements. For each , suppose that a set satisfies
The indices matter in (61c); different labels are not identified even if their values in agree.
Lemma 11. If and (61a)–(61c) hold, the homomorphism
is an isomorphism.
Proof. Let be the free generators. We prove the stronger statement: whenever a finite signed word evaluates in to , its value in the free group is . Induct on the number of letters of . For the empty word, the infinite order of gives . If the first letter is , , the remaining word evaluates to , so induction applies.
Suppose the first letter is . The path described by the remaining letters starts at and ends at . Choose its first crossing into . By (61b), the crossing is not a -letter. By (61c), the only possible crossing from outside to inside is the step labelled . Thus
Both and are shorter than . Their free values are respectively and , so the asserted equality follows. If the first letter is , the remaining word runs from to . Its first crossing out of is , giving
with the same values for , and the same argument applies.
Taking proves that every word in the kernel is the identity in the free group. Generation gives surjectivity. ◻
These cuts can be required using actual subsemigroups. For each , require such that
Indeed, satisfies (61a)–(61c); the power invariance follows from
Lemma 12. A free basis admits subsemigroups satisfying (65a)–(65b).
Proof. In the reduced-word model, let consist of the elements whose reduced word does not begin with the positive letter . Right multiplication by a basis generator changes this property only on the edge . To see this, cancellation at the right end cannot change the first letter unless the whole word cancels; in that case the two endpoints are precisely . This proves (61c), and applying it repeatedly to proves (61b).
Let count the exponent sum of , and put
Every displayed generator has -value . Therefore a conjugate belonging to must be a single generator, rather than a product of two or more. If
then . Consequently for some , and (61b) gives if and only if . We conclude that
The required subsemigroup cuts now follow from the set cuts. ◻
For a submonoid , use the same irreducible set as in Section “A finite sentence for free abelian groups”. Let assert
and, for every , the existence of a subsemigroup such that
where and . As before, generation by irreducibles is expressed by quantifying over subsemigroups containing and every irreducible. Hence is a finite formula in the lattice language together with a relation for multiplication on codes.
Proposition 13. A nontrivial torsion-free group is free if and only if holds for some and .
Proof. Assume . Set and . The two generation clauses imply
Since , it has infinite order. The cut clauses and Lemma 11 give a free basis.
Conversely, choose a generator of a nonempty free basis and let be the positive-word monoid on that basis. Word length shows that has no unit other than , and that its irreducibles are exactly the basis letters. They generate as a monoid and as a group. Lemma 12 supplies the cut clauses. Isomorphisms transport all the conditions. ◻
Eliminating the multiplication relation
We use Scott’s theorem in the following form (2): an injective map between groups satisfying
is either a homomorphism or an antihomomorphism. There is no cardinality restriction. The noncommuting-pair proof in (6) gives the same global orientation and is the proof used here.
Let be any first-order structure, let be defined by a formula, and let be a definable ternary relation. Name four elements
of , and add a ternary relation symbol . Consider the following finite theory :
-
implies , and .
-
For there is a unique with .
-
The operation on described by is associative, has identity , and every element has a left inverse.
-
For ,
-
and .
Each group condition is first-order in . For example, associativity is
The identity conditions are and for , and the inverse condition is . Together with totality and uniqueness they define an ordinary group on the nonempty set .
Lemma 14. For a fixed and fixed , at most one relation satisfies .
Proof. Suppose describe operations . Equation (75) gives
The identity map is a bijective half-homomorphism between the two groups. By Scott’s theorem, for all pairs, or for all pairs. The latter alternative contradicts the common orientation:
Thus the operations agree on . Both relations are false outside , so . ◻
Lemma 15. Let be a first-order theory in , where is a relation symbol. If each -structure has at most one expansion satisfying , then there is an -formula such that
Proof. This is the implicit-to-explicit definability theorem of Beth (1), in its at-most-one-expansion form. One standard proof uses compactness and the joint-consistency theorem: two theories with no common expansion admit a sentence of their common language separating them. Take disjoint copies of the added symbol and the corresponding renamed theories . Name a tuple in the common base language. Implicit uniqueness makes and jointly inconsistent. Separation yields a formula of the common base language; removing the names gives the defining formula.
If is finite, its reducts are defined by the finite conjunction . Indeed, a model of satisfies this conjunction because there. Conversely, expand a reduct satisfying by declaring to mean . Induction on formulas shows that this expansion satisfies every sentence of . ◻
Apply the lemma to , with the four named elements included in the base language. Removing their constant names gives a fixed formula
in the original language, with seven free variables. Put
This is a finite formula in the base language. For any interpretation of , its truth means precisely that interpreting by produces the required relational group.
Lemma 14 applies to every base structure . Thus Beth’s hypothesis holds before the specialization .
The finite sentence for free groups
Take , let be the element-code predicate , and let from Section “Recovering unordered multiplication”. Let and be the resulting formulas. In , replace every multiplication relation by , and denote the resulting pure lattice formula by
Inversion and conjugation in the cut clauses are expressed through existentially quantified intermediate products and the identity parameter . All intermediate elements are required to be -codes. The quantifiers over and the cut subsemigroups remain ordinary lattice quantifiers.
Define
The variables are lattice variables representing elements by their codes. Each displayed abbreviation expands to a finite first-order formula. In particular, (83) contains no quantifier over relations.
Lemma 16. Suppose a torsion-free group satisfies and its subsemigroup lattice has an expansion satisfying . Under the identification , the multiplication defined by is either the multiplication of or its opposite.
Proof. Let be the relational group operation on the element codes. By (75) and Proposition 10,
The decoding bijection is therefore a half-isomorphism from the relational group to . Scott’s theorem makes it an isomorphism or an anti-isomorphism. Only soundness of was used here; its completeness is not assumed for this . ◻
Proof of (3a). Assume . By (18), the multiplication of makes it a torsion-free group , up to semigroup isomorphism. If the cyclic alternative holds, is free of rank zero or one.
In the other alternative, interpret using the formula at the chosen parameters. Lemma 15 gives the relational group axioms, and Lemma 16 identifies its multiplication with that of or . Reversing multiplication preserves every subsemigroup as a set, and hence induces
Consequently the chosen satisfy in or . Proposition 13 proves that group free. Inversion is an isomorphism , so is free in either case.
Conversely, let be free. In ranks zero and one it satisfies . Otherwise is noncyclic and nonabelian. Lemma 8 gives and Proposition 10 gives
Choose with and put , , all represented by their element codes. The actual multiplication graph satisfies . By Beth definability, agrees with this graph. A nonempty free basis gives satisfying by Proposition 13. Hence (83) holds. Transport along a semigroup isomorphism finishes the proof for arbitrary . ◻
Theorem 7 proves (3b), so both assertions of Theorem 1 follow. Adjoining the finitely many lattice identities to each of the two sentences gives the required finitely axiomatized classes of lattices.
References
- E. W. Beth, On Padoa’s method in the theory of definition, Indagationes Mathematicae (Proceedings) 56 (1953), 330–339. doi:10.1016/S1385-7258(53)50042-3.
- W. R. Scott, Half-homomorphisms of groups, Proc. Amer. Math. Soc. 8, no. 6 (1957), 1141–1144. doi:10.1090/S0002-9939-1957-0095890-3.
- L. N. Shevrin, On lattice properties of semigroups, Sibirsk. Mat. Zh. 3, no. 3 (1962), 446–470 (Russian).
- L. N. Shevrin, The lattice-subsemigroup characterization of commutative nonperiodic groups, Sibirsk. Mat. Zh. 5, no. 3 (1964), 671–678 (Russian).
- L. N. Shevrin, Elementary lattice properties of semigroups, Sibirsk. Mat. Zh. 7, no. 3 (1966), 664–684 (Russian).
- W. Vipismakul, Another proof of half homomorphisms, Burapha Science Journal 25, no. 1 (2020), 187–193.