When an algorithmic group fits inside a simple group with finite structure in every dimension

← All digests

Family 250 · Group theory
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-22

1. The question

Suppose a group comes with an algorithm that decides whether any word in its generators equals the identity. Can that computational fact be recognized purely from the group’s position inside a larger algebraic object?

The Boone–Higman conjecture predicts a striking answer: a finitely generated group has such an algorithm exactly when it embeds in a finitely presented simple group. The larger group combines a finite description with the absence of proper nontrivial normal subgroups.

Family 250 contains three related claims. The first resolves that conjecture. The second asks for more geometric finiteness in the simple overgroup and claims it in every dimension. The third removes the word-problem hypothesis and constructs one universal group with the same higher finiteness property containing every finitely presented group, but does not require this universal target to be simple.

These are different embedding questions. In particular, the last universal target must accommodate groups with undecidable word problem, whereas a finitely presented simple group has decidable word problem.

2. Background and essential definitions

Algorithms for equality

Choose a finite generating set for a group $G$. A word is a finite string of generators and their inverses. The word problem asks whether that string represents the identity.

A decider must halt on every input and answer correctly both when the word is trivial and when it is not. This is stronger than an enumeration procedure that eventually lists all trivial words but may never settle a nontrivial one.

A recursively presented group has finitely many generators and a recursively enumerable list of defining relations. A finite presentation is a special case. Recursive presentation does not imply decidability of the word problem; finitely presented groups with undecidable word problem exist.

An embedding $G\hookrightarrow H$ is an injective homomorphism. It preserves every equality and inequality between elements, so it identifies $G$ with a subgroup of $H$.

Simplicity is an algebraic restriction

A subgroup is normal if conjugation by every ambient group element preserves it. A nontrivial group is simple when its only normal subgroups are the identity subgroup and the whole group.

Simplicity does not mean that multiplication is easy to compute, that the group is finite, or that it has few ordinary subgroups. A simple group can contain complicated subgroups that are not normal.

Finite presentation means a finite list of generators and relators determines the entire group. It does not generally pass to subgroups. Thus placing a group inside a finitely presented group is weaker than giving the subgroup itself a finite presentation.

Finiteness in every dimension

A classifying space $K(H,1)$ is a connected cell complex whose fundamental group is $H$ and whose universal cover is contractible. Its cells encode generators, relations, and successively higher relations among those relations.

A group has type $F_n$ if such a space can be chosen with finitely many cells through dimension $n$. Type $F_1$ is finite generation, and type $F_2$ is finite presentation.

Type $F_\infty$ means that one classifying space has finitely many cells in each dimension. It may still have infinitely many dimensions and infinitely many cells altogether. In particular, this condition does not imply torsion-freeness or finite cohomological dimension.

Homological conditions denoted $FP_n$ instead concern finite generation of modules in a projective resolution. They are related to geometric finiteness but should not be silently substituted for $F_n$. The manuscripts’ final claims are geometric $F_\infty$ statements.

3. Development history and previous partial results

Higman’s 1961 embedding theorem characterized the finitely generated subgroups of finitely presented groups: they are exactly the finitely generated recursively presented groups. It connected an effective enumeration of relations to a finite algebraic description in an overgroup.

Boone and Higman obtained a sharper characterization in 1974. A finitely generated group has decidable word problem exactly when it sits in a chain

$$G\leq S\leq P,$$

where $S$ is simple and $P$ is finitely presented. The two desirable properties belonged to different groups. Their conjecture asked whether the intermediate simple group could itself be finitely presented. Finite presentation of $P$ alone cannot justify that conclusion for its subgroup $S$.

Thompson later obtained a finitely generated simple overgroup while preserving decidability of the word problem. That advanced the finite-generation side, but finite generation remained weaker than finite presentation.

Recent positive cases include hyperbolic and contracting self-similar groups, through work of Belk, Bleak, Matucci and Zaremsky. Bux, Llosa Isenrich and Wu treated groups acting on locally finite trees, including Baumslag–Solitar groups and finite-rank free-by-cyclic groups. Belk, Fournier-Facio, Hyde and Zaremsky established results for automorphism groups of finite-rank free groups and mapping class groups of punctured surfaces.

Thompson-type constructions also supplied simple groups with strong higher finiteness properties. Brown proved type $F_\infty$ for the Thompson–Higman families. Belk and Zaremsky constructed one simple $F_\infty$ group containing all finitely generated right-angled Artin groups, while Belk, Hyde and Matucci obtained a simple $F_\infty$ group containing every countable abelian group.

The general obstacle was to build such an overgroup from an arbitrary word-problem decider. Belk and Zaremsky’s twisted Brin–Thompson construction provides a final transfer theorem, but its input requires a suitable permutation action and finiteness properties of stabilizers. Zaremsky later weakened the conditions needed for finite presentation alone.

A separate higher-dimensional Higman question asks for $F_\infty$ overgroups without decidability of the word problem. Leary obtained broad embeddings under the weaker homological condition $FP_2$. Fournier-Facio and Zaremsky analyzed how homological embeddings could yield geometric ones and identified limitations of particular rope-trick constructions. M3 addresses this separate direction with a single universal target.

4. What the new manuscripts claim

M1: finite presentation and simplicity together

M1 asserts that a finitely generated group $G$ has decidable word problem exactly when there is an embedding

$$G\hookrightarrow H$$

into some finitely presented simple group $H$.

For the constructive direction, the input includes a finite generating set and a word-problem decider. The input group need not be finitely presented, and the algorithm’s running time is unrestricted.

Its principal intermediate theorem embeds $G$ in a finitely presented group with a faithful two-transitive action on a countably infinite set and finitely generated point stabilizers. Faithful means that no nonidentity element acts trivially. Two-transitive means that any ordered pair of distinct points can be sent to any other such pair.

M2: a simple target of type F∞

M2 strengthens the target to a nontrivial simple group $H$ of type $F_\infty$. The input conditions remain finite generation and decidable word problem. The source also notes that its chosen target can be generated by two finite-order elements, using a further conclusion of the earlier twisted Brin–Thompson theorem.

The converse is classical. In a nontrivial finitely presented simple group, run two searches in parallel: one for a proof that a given word is trivial, and one for a proof that adding it as a relator collapses the whole group. A nontrivial element normally generates a simple group, so exactly the appropriate search terminates. Restricting this decision procedure to a finitely generated subgroup decides its word problem.

M2 imports specific tools from M1, including its bounded compiler, corner construction and Steinberg-kernel calculation. Its higher-finiteness assertion is not obtained merely by applying M1’s final theorem.

M3: one universal higher-finiteness group

M3 asserts the existence of a single group $U_\infty$ of type $F_\infty$ containing every finitely presented group. No word-problem assumption is imposed on those subgroups.

Equivalently, its finitely generated subgroups, up to isomorphism, are exactly the finitely generated recursively presented groups. The forward construction uses Higman’s embedding theorem; the reverse direction follows by enumerating relations inside the finitely presented ambient group.

M3 shares the finite-diagram and ascending-torus method with M2 and gives local proofs of the relevant identities and finiteness criterion. It does not assert simplicity of the universal target. All three manuscripts are dated 23 September 2026.

5. Why the advance matters

Importance: 5/5, provisional. The family claims a general algebraic characterization of decidable word problems, resolving a central embedding conjecture and extending it to geometric finiteness in every dimension.

M1 joins two properties that earlier general embeddings placed in different overgroups. M2 controls not only generators and relations but also the higher-dimensional relations needed to construct finite skeleta throughout a classifying space.

M3 has a different breadth: one $F_\infty$ group accommodates all finitely presented groups, including those with undecidable word problem. It would show that strong geometric finiteness of an ambient group can coexist with very complicated subgroup algorithms.

The significance is the exact match between computational hypotheses and algebraic embedding possibilities. The claims do not say that every group becomes computationally easy after embedding.

6. The main idea

Build a permutation action that a known theorem can use

The final simplicity step is supplied by the twisted Brin–Thompson construction. It turns a suitable faithful action on a countable set into an embedding in a simple group. The new task is to manufacture the required action while controlling its finiteness properties.

For M1, the action is affine: translations move one chosen point to zero, while a linear part moves nonzero vectors. A simple algebra helps provide the needed linear operations, because every nonzero element generates the entire algebra as a two-sided ideal.

The source keeps this simple algebra inside a different, finitely presented algebra. Passing to a later stage moves a finite set of coordinates into the simple part. This avoids assuming that one algebra automatically has both simplicity and finite presentation.

Elementary matrix operations are first represented in an abstract Steinberg group. Elements in the kernel of its map to actual matrices would spoil faithfulness. M1 constructs a transition that removes this whole kernel, then adjoins a stage-shifting element to package the sequence into a finitely presented group.

The algebra itself is built by encoding computation into finite word rules. A bounded evaluator ensures that recursive calls concern shorter inputs. A fixed-point construction supplies the self-referential program, and induction on input length supplies termination. The role of this mechanism is to preserve the original group’s equality information while creating the extra algebraic operations.

Higher finiteness requires stabilizers too

M2 constructs a highly transitive action: any two ordered tuples of distinct points of the same finite length lie in one orbit. It also arranges type $F_\infty$ for the acting group and every finite-point stabilizer, the subgroup fixing a prescribed finite set point by point.

This extra work is essential. A good action of a finitely presented group need not have stabilizers with the higher finiteness required by the transfer theorem. The source builds the action and stabilizers together through compatible matrix and ring constructions.

The shared device is an ascending torus, also called an ascending HNN extension. For an injective endomorphism $f:V\to V$, it has the form

$$T(V,f)=\langle V,t\mid t^{-1}vt=f(v)\text{ for }v\in V\rangle.$$

The new element $t$ records movement through successive copies of $V$. A factorization through a finitely presented group supplies finite presentation. Additional finite diagrams of commuting homomorphisms control higher-dimensional cycles.

Two primes force the same obstruction to vanish

The diagrams arise from systems of idempotents, elements satisfying $e^2=e$, over $\mathbb F_2$ and $\mathbb F_3$. Once lower-dimensional error terms have been removed, the two systems force the same relevant homology class $z$ to satisfy

$$2z=0,\qquad3z=0.$$

Then $z=3z-2z=0$. The difficult input is making both relations apply to the same class with enough uniform control.

For each finite set of allowed simplex types, the sources obtain one iteration and one finite set of output types that fill all relevant cycles. The lengths and coefficients of the filling chains need not be uniformly bounded. This specific uniformity lets the argument handle products of group-ring modules and apply the Bieri–Eckmann finiteness method. Finite presentation then helps convert the homological control into actual finite skeleta.

M3 begins instead with a classical universal finitely presented group $V$. It embeds a finite diagram enlargement back into $V$, obtaining an injective endomorphism $f$. The same ascending-torus method makes $T(V,f)$ type $F_\infty$, while retaining $V$ and therefore every finitely presented subgroup already inside it.

7. Scope, limitations, and open questions

The simple-overgroup theorems require decidability of the word problem. M3 removes that condition by making a different assertion about a universal target without simplicity. Combining the strongest adjectives from all three statements would produce a claim the sources do not make.

Type $F_\infty$ means finitely many cells in each dimension, not a finite-dimensional or finite classifying space. It gives no automatic bound on cohomological dimension and does not exclude torsion.

The embedding constructions preserve the group as a subgroup, but the stated theorems do not promise polynomial-time word-problem algorithms, small presentations, or controlled geometric distortion of the embedding. A decider supplied as input may take arbitrarily long to halt.

The higher-finiteness proof also requires uniformity over finite simplex types, not a universal bound on filling lengths. Confusing these different quantitative statements would overstate the result.

Questions about efficient embeddings, explicit small targets and additional geometric properties of the overgroups remain separate from the existence theorems summarized here.

8. Sources and further reading

Higman, Subgroups of finitely presented groups, and Boone–Higman, An algebraic characterization of groups with soluble word problem, supply the foundational embedding theorems. Thompson’s word-problem-preserving simple embeddings gives the finite-generation predecessor.

Belk and Zaremsky, Twisted Brin–Thompson groups, supplies the final simplicity and finiteness transfer. Zaremsky’s finite-presentability theorem weakens the action hypotheses needed for M1.

For the higher-dimensional background, see Bieri and Eckmann’s finiteness criteria, Brown’s geometric finiteness theory, and Fournier-Facio and Zaremsky’s Finiteness properties and Higman’s rope trick.

Source manuscripts

• M1. OpenAI, Finite algebraic envelopes and the Boone–Higman conjecture (September 23, 2026).
• M2. OpenAI, Simple F∞ overgroups of groups with decidable word problem (September 23, 2026).
• M3. OpenAI, A universal group of type F∞ (September 23, 2026).

Sources consulted on 22 October 2026 at repository revision 1581d4c. Download the manuscript citations.