Malnormal Subgroups of Finitely Presented Groups

This paper refines classical embedding theorems by proving that recursively presented groups admit quasi-isometric malnormal embeddings into finitely presented groups with the congruence extension property and controlled decidability, while also establishing that any countable group can be malnormally embedded into a finitely presented group with a prescribed word length function.

The Gist

Imagine you have a very complex, messy instruction manual for a machine. This manual is so long and complicated that you can't even write it down completely; you can only list the rules one by one as you discover them. In math, we call this a recursively presented group. It's a set of rules that a computer could generate forever, but it's not a neat, finite list.

Now, imagine you want to put this messy machine inside a much bigger, perfectly organized factory (a finitely presented group) where the rules are finite, clear, and easy to check.

The famous Higman Embedding Theorem (from the 1960s) already proved that you can do this. You can always take your messy machine and hide it inside a clean factory. But the old method had a few problems:

1. The messy machine might get distorted (stretched or squished) so much that it's hard to recognize.
2. The factory might have "leaks" where the messy machine's rules accidentally mix with the factory's rules in weird ways.
3. If your messy machine had a problem you couldn't solve (an undecidable "Word Problem"), the factory might also become unsolvable, or vice versa, depending on how you built it.

Francis Wagner's paper is like a master architect who says, "I can build a better factory." He proves you can embed that messy machine into a clean factory with five superpowers:

1. The "Malnormal" Security System

The Metaphor: Imagine your messy machine is a secret club inside a huge city (the factory). In the old factories, if you walked around the city, you might accidentally bump into a member of your club and start a conversation that reveals your secrets.
Wagner's Fix: He builds the club so that it is Malnormal. This means the club is so isolated that if you walk around the city and try to "translate" the club's rules into the city's language (conjugate the subgroup), the only time you get a match is if you are standing right inside the club itself. If you step outside, the club's rules look completely alien and don't overlap with anything else. It's like a secret society that is invisible to the rest of the world unless you are already a member.

2. The "Congruence Extension" Passport

The Metaphor: Imagine the club has its own internal laws. In the old factories, if the club wanted to change a law or split into a smaller group, the city's government might not respect that change, or the city's laws might interfere.
Wagner's Fix: He ensures the Congruence Extension Property (CEP). This means the club is so well-integrated that any rule change or split the club decides on can be automatically extended to the whole city without breaking anything. The city respects the club's internal logic perfectly. It's like having a VIP passport that guarantees your internal rules are recognized as valid laws for the entire building.

3. The "No Distortion" Elevator

The Metaphor: In the old factories, if you tried to walk from point A to point B inside the messy machine, the factory might force you to take a detour that is 1,000 times longer than the direct path. The machine gets "distorted."
Wagner's Fix: He builds an elevator system where the distance you travel inside the messy machine is exactly the same as the distance you travel in the factory (up to a constant factor). This is called a quasi-isometric embedding. The machine isn't stretched or squished; it keeps its original shape and size.

4. The "Noise" Machine (The Secret Sauce)

How did he do it?
The paper introduces a new tool called "Noisy S-machines."

• Old S-machines: Think of these as robotic assembly lines that build groups. They are very precise but rigid. They use "commutator" rules (like saying "A then B is the same as B then A") which accidentally create overlaps, breaking the "Malnormal" security.
• Noisy S-machines: Wagner adds "noise" to the assembly line. Imagine the robot is supposed to move a block, but instead, it adds a little bit of static or a random sound effect to the block as it moves it.
  • This "noise" (extra letters in the code) ensures that if you try to mix the club's rules with the city's rules, the noise makes them incompatible. It breaks the accidental overlaps, securing the Malnormal property.
  • Crucially, Wagner shows that this noise is "tame." It doesn't ruin the shape (distortion) or the logic; it just acts as a perfect security guard.

5. The "Word Problem" Mirror

The Metaphor: The "Word Problem" is like asking, "Is this specific sequence of instructions actually a valid move, or is it nonsense?"

• If your messy machine has a problem you can't solve (you can't tell if a sequence is nonsense), the factory will also be unsolvable.
• If your messy machine can be solved, Wagner's factory can also be solved.
• The Breakthrough: He proves that the factory's solvability is exactly tied to the messy machine's solvability. You don't accidentally make a solvable machine unsolvable, or vice versa.

Why Does This Matter?

This paper is a "refinement" of a 60-year-old theorem. It's like taking a basic car and upgrading it to a self-driving, bulletproof, fuel-efficient vehicle that fits perfectly into any garage.

• For Mathematicians: It solves a long-standing question posed by Denis Osin: "Can we always embed these groups in a way that keeps them isolated (malnormal)?" The answer is a definitive YES.
• For Computer Science: It connects the complexity of algorithms (how hard it is to solve problems) with the geometry of shapes (how groups are distorted). It shows we can control the "shape" of these abstract mathematical objects with incredible precision.

In short: Wagner built a perfect, secure, and undistorted "home" for any computable group, ensuring it stays true to its original form while being completely isolated from the outside world, all while keeping the complexity of its internal logic intact.

Technical Summary

This is a detailed technical summary of the paper "Malnormal Subgroups of Finitely Presented Groups" by Francis Wagner.

1. Problem Statement

The paper addresses a long-standing open question in geometric group theory regarding the Higman Embedding Theorem.

• Background: Higman's theorem (1961) establishes that a finitely generated group is recursively presented if and only if it can be embedded into a finitely presented group.
• The Gap: While various refinements of Higman's theorem exist (preserving properties like the Word Problem, bi-Lipschitz distortion, or conjugacy decidability), it was unknown whether a malnormal refinement was possible. A subgroup $G \leq H$ is malnormal if for all $h \in H \setminus G$, $h^{-1}Gh \cap G = \{1\}$.
• The Question: Can every finitely generated recursively presented group be embedded as a malnormal subgroup into a finitely presented group, while simultaneously preserving other geometric and algorithmic properties (such as word length equivalence and the decidability of the Word Problem)?

2. Methodology

The author employs a sophisticated construction based on generalized S-machines, specifically introducing a new variant called "noisy S-machines."

• S-Machines: Originally introduced by Sapir, Birget, and Rips, S-machines are computational models used to construct finitely presented groups with specific geometric properties. They simulate Turing machine computations via group relations.
• The Innovation (Noisy S-machines): Standard S-machines use commutator relations that often prevent malnormality. Wagner generalizes the model to allow the application of predetermined automorphisms of free groups to tape words during computation.
  • "Noise": This generalization introduces "noise" (specifically, auxiliary letters and relations resembling Baumslag-Solitar groups) into the tape words.
  • Function: This noise prevents the conjugation of non-trivial elements of the embedded subgroup by elements outside the subgroup, thereby enforcing malnormality.
  • Control: Despite the introduction of noise (which typically complicates the Dehn function), the author carefully controls the geometry to ensure the embedding remains quasi-isometric (bi-Lipschitz).
• Construction Steps:
  1. Initial Embedding: A finitely generated recursively presented group $R$ is first embedded into a recursively presented group $R_C$ with specific combinatorial properties (using a "standard trick" to ensure positive relators).
  2. Machine Construction: A complex hierarchy of machines is built:
     • $M_A^1$: A noisy machine ensuring malnormality.
     • $M_L^2$: Simulates a Turing machine enumerating the relators of $R$.
     • $M_L^3, M_L^4, M_L^5$: Compositions and "reflected" copies of the previous machines to create symmetry and circular structures.
     • $M_L^6, M_L$: The final "main machine" $M_L$ combines these components to simulate the computation of $R$ while maintaining the necessary algebraic constraints.
  3. Group Presentation: The group $G(\Omega(ML))$ is constructed from the machine $M_L$ by adding specific relations (disk relations and "a-relations") that force the machine's accepted configurations to represent the identity.
  4. Diagrammatic Analysis: The core of the proof involves analyzing van Kampen diagrams over the group presentation. The author introduces:
     • Weight Functions: To bound the area of diagrams.
     • Transposition Techniques: To move bands (representing computation steps) across cells and disks.
     • Minimal Counterexample Arguments: Proving that if the subgroup were not malnormal or if the distortion were too high, one could construct a "minimal counterexample" diagram that leads to a contradiction via combinatorial arguments (e.g., Euler characteristic, graph theory on bands).

3. Key Contributions

A. The Main Theorem (Theorem A)

The paper proves that a finitely generated group $G$ is recursively presented if and only if there exists an embedding $G \hookrightarrow H$ such that:

1. $H$ is finitely presented.
2. $G$ is a malnormal subgroup of $H$.
3. $G$ satisfies the Congruence Extension Property (CEP) (a new refinement in this context).
4. The embedding is quasi-isometric (the word length in $G$ is equivalent to the restriction of the word length in $H$).
5. The Word Problem for $H$ is decidable if and only if it is decidable for $G$.

B. Extension to Countable Groups (Corollary B)

The result is extended to arbitrary countable groups. For any countable group $G$ and a computable length function $\ell: G \to \mathbb{N}$, there exists a malnormal embedding into a finitely presented group $H$ such that the restriction of the word metric in $H$ to $G$ is equivalent to $\ell$. This refines a theorem by Ol'shanskii.

C. Congruence Extension Property (CEP)

The paper introduces the CEP condition as a novel feature of Higman embeddings. A subgroup $G \leq H$ has the CEP if every epimorphism from $G$ extends to an epimorphism from $H$. This is shown to hold for the constructed embeddings.

D. Decidability of the Word Problem

The construction explicitly handles the decidability of the Word Problem. If the input group has a decidable Word Problem, the resulting finitely presented group $H$ also has a decidable Word Problem. This is achieved by bounding the Dehn function of $H$ by a computable function derived from the time complexity of the underlying Turing machine simulation.

4. Results and Technical Proofs

• Malnormality: Proven by showing that any annular diagram representing a conjugation $g^{-1}hg = h'$ (where $h, h' \in G$ and $g \notin G$) must be trivial. This relies on the "noise" in the S-machines preventing non-trivial conjugations within the subgroup.
• Distortion: The embedding is shown to be bi-Lipschitz. The "noise" introduced to ensure malnormality is carefully bounded so that it does not distort the word length of the embedded group.
• Dehn Function: The author derives an upper bound for the Dehn function of the constructed group $H$. The bound is roughly polynomial in terms of the time function of the Turing machine simulating the group's relators, ensuring that if the original Word Problem is decidable, the Dehn function is computable, and thus the Word Problem for $H$ is decidable.
• Diagrammatic Rigor: The proof utilizes a "weight" system on diagrams to handle the complexity introduced by the "noisy" relations. It establishes that minimal diagrams cannot contain certain structures (like $\theta$-annuli or specific disk configurations) that would violate the malnormality or distortion properties.

5. Significance

• Resolution of Osin's Question: The paper positively answers a question posed by Denis Osin (attributed to D. Osin) regarding the existence of a malnormal refinement of the Higman embedding theorem.
• Unification of Properties: It demonstrates that algebraic (malnormality, CEP), geometric (quasi-isometry), and algorithmic (Word Problem decidability) properties can be simultaneously preserved in a Higman embedding. Previously, achieving malnormality often came at the cost of geometric distortion or algorithmic control.
• New Tools: The introduction of noisy S-machines provides a powerful new tool for constructing groups with specific subgroup structures. This technique bridges the gap between the rigid commutator relations of standard S-machines and the flexible, non-malnormal relations of Baumslag-Solitar type groups.
• Applications: The result implies that any recursively presented group can be realized as a malnormal subgroup of a finitely presented group with controlled distortion. This has implications for the study of relatively hyperbolic groups, as malnormal subgroups are central to their definition.

In summary, Francis Wagner's work provides a definitive and highly technical solution to a major open problem in geometric group theory, refining the foundational Higman embedding theorem to include malnormality and the congruence extension property without sacrificing geometric or algorithmic control.