Topological constraints on clean Lagrangian intersections from $\mathbb{Q}$-valued augmentations

This paper proves that for knots containing specific components like the $(2,q)$-torus knot or the figure-eight knot, no compactly supported Hamiltonian diffeomorphism can move their conormal bundles to intersect the zero section cleanly along an unknot, a result established by deriving a unique algebraic constraint on the augmentation variety over the rational numbers using symplectic field theory.

The Gist

Here is an explanation of the paper "Topological constraints on clean Lagrangian intersections from Q-valued augmentations" by Yukihiro Okamoto, translated into everyday language with creative analogies.

The Big Picture: The "Untangling" Problem

Imagine you have a piece of string tied in a complex knot (like a pretzel or a figure-eight). Now, imagine this string is floating inside a giant, invisible, elastic sheet (this is the mathematical space called $T^*\mathbb{R}^3$).

The paper asks a very specific question: If you have a knotted string, can you stretch, twist, and pull the elastic sheet around it (using "Hamiltonian diffeomorphisms," which are just fancy, smooth, energy-conserving moves) so that the string ends up looking like a simple, unknotted circle (the "unknot")?

In the world of this math, the "string" is actually a higher-dimensional shape called a Lagrangian submanifold (specifically, the conormal bundle of a knot). The "sheet" is the zero section. The question is: Can you deform the knotted shape so that it intersects the sheet cleanly in the shape of a simple circle?

The Answer: For certain types of knots (specifically those involving torus knots or the figure-eight knot), the answer is NO. You cannot untangle them into a simple circle just by stretching the space around them. They are "rigid."

---

The Detective Work: Using "Augmentations" as a Fingerprint

How does the author prove this? He doesn't try to physically untie the knots. Instead, he uses a mathematical "fingerprint" called an Augmentation Variety.

Think of a knot as a complex machine with many gears.

1. The Machine (Knot DGA): Mathematicians have built a complex algebraic machine (a Differential Graded Algebra, or DGA) for every knot. This machine has many parts (variables) that interact in specific ways.
2. The Fingerprint (Augmentation): An "augmentation" is like a way of simplifying this machine. You take all the complex gears and assign them simple numbers (from a field $k$) to see if the machine still works without breaking.
3. The Variety: The set of all possible number assignments that make the machine work forms a shape called the Augmentation Variety ($V_k(K)$).

The Logic:
If you could magically transform Knot A (the complex one) into Knot B (the simple unknot) by stretching the space, then the "fingerprint" of Knot B would have to fit inside the "fingerprint" of Knot A. In math terms, there must be a way to map the solutions of the simple knot into the solutions of the complex knot.

The Twist: Why "Rational Numbers" ($\mathbb{Q}$) are the Key

This is where the paper gets clever. Most mathematicians usually look at these fingerprints using Complex Numbers (which include everything: real numbers, imaginary numbers, etc.). In the world of complex numbers, almost everything has a solution. If you look for a root of a polynomial in complex numbers, you will almost always find one.

The Problem: If you use complex numbers, the fingerprints of the complex knot and the simple knot look too similar. The complex knot's fingerprint is so "full" that it can easily swallow the simple knot's fingerprint. It's like trying to prove a specific lock is unique by looking at a keyring that has every possible key in the universe.

The Solution: The author decides to use Rational Numbers ($\mathbb{Q}$) instead. These are just fractions (like 1/2, 3/4, -5).

• In the world of fractions, things are much stricter. A polynomial might have a solution in complex numbers, but no solution in rational numbers.
• The author constructs a specific algebraic "trap" (a polynomial equation) derived from the complex knot.
• He proves that for the complex knot, this trap requires a solution that does not exist in the world of rational numbers.
• However, for the simple unknot, the trap is easy to solve in rational numbers.

The Metaphor:
Imagine the complex knot is a safe with a very specific, rare combination lock.

• If you look at the safe using "Complex Numbers" (a master key that opens every door), you can't tell the difference between the complex safe and a simple box.
• But if you look at the safe using "Rational Numbers" (a specific, limited set of keys), the complex safe cannot be opened by the keys that open the simple box.
• Therefore, the complex safe cannot be transformed into the simple box, because the "keys" (mathematical solutions) don't match up.

The Specific Knots

The paper focuses on two types of knots:

1. Torus Knots: Knots that wrap around a donut shape (like a (2, 5) knot).
2. The Figure-Eight Knot: The simplest knot that isn't a simple loop.

The author shows that if your knot is made by tying a simple knot to one of these "special" knots, you can never stretch the space to turn it into a simple circle.

The "Arithmetic" Argument

The proof relies on a famous theorem called Hilbert's Irreducibility Theorem.

• Think of the polynomial equation as a machine that produces numbers.
• The author shows that if you plug in rational numbers, this machine will never spit out a valid answer for the complex knot.
• But for the simple knot, it does spit out answers.
• Because the "output" (the set of valid rational solutions) is different, the two knots cannot be the same.

Summary

1. The Goal: Prove you can't untangle certain complex knots into simple circles just by stretching the space around them.
2. The Tool: A mathematical "fingerprint" (Augmentation Variety) derived from the knot's algebraic structure.
3. The Trick: Most people look at these fingerprints using "Complex Numbers," where everything is possible. This author looks at them using "Rational Numbers" (fractions), where things are much harder to satisfy.
4. The Result: The complex knot's fingerprint requires a solution that doesn't exist in the world of fractions. The simple knot's fingerprint does. Since the fingerprints don't match, the transformation is impossible.

In short: The paper uses the strict rules of fractions to prove that some knots are permanently knotted, no matter how much you stretch the universe around them.

Technical Summary

Here is a detailed technical summary of the paper "Topological constraints on clean Lagrangian intersections from Q-valued augmentations" by Yukihiro Okamoto.

1. Problem Statement

The paper addresses a fundamental question in symplectic topology regarding the rigidity of Lagrangian intersections. Specifically, it investigates whether a Lagrangian submanifold associated with a non-trivial knot can be Hamiltonianly isotoped to intersect the zero section (identified with $\mathbb{R}^3$) cleanly along the unknot.

• Context: Let $T^*\mathbb{R}^3$ be the cotangent bundle of $\mathbb{R}^3$ with its canonical symplectic structure. For a knot $K \subset \mathbb{R}^3$, let $L_K$ be its conormal bundle, a Lagrangian submanifold intersecting the zero section $\mathbb{R}^3$ cleanly along $K$.
• The Question: If $\phi \in \text{Ham}_c(T^*\mathbb{R}^3)$ is a compactly supported Hamiltonian diffeomorphism, can $\phi(L_K)$ intersect $\mathbb{R}^3$ cleanly along a knot $K'$ that is isotopic to the unknot, even if $K$ is not the unknot?
• Previous Work: It was known that if $K$ is the unknot, the intersection cannot be changed to a different knot type (rigidity of unknottedness). However, the persistence of knottedness (i.e., can a knotted $L_K$ be unknotted via Hamiltonian isotopy?) remained open for general knots.

2. Methodology

The author employs Symplectic Field Theory (SFT), specifically the Chekanov-Eliashberg Differential Graded Algebra (DGA), combined with arithmetic geometry.

A. Symplectic Field Theory and DGAs

1. Knot DGA: The paper utilizes the knot DGA, a combinatorial invariant defined from a braid representation of a knot, with coefficients in the ring $R = \mathbb{Z}[\lambda^{\pm}, \mu^{\pm}, U^{\pm}]$.
2. Augmentation Variety: For a field $k$, the augmentation variety $V_k(K) \subset (k^*)^3$ is defined as the set of points $(x, y, Z)$ corresponding to ring homomorphisms (augmentations) from the knot DGA to $k$ that map the generators $\lambda, \mu, U$ to $x, y, Z$ respectively.
3. Cobordism Induced Maps: The existence of a clean intersection between $\phi(L_{K_0})$ and $\mathbb{R}^3$ along $K_1$ implies the existence of an exact Lagrangian cobordism between the unit conormal bundles $\Lambda_{K_1}$ and $\Lambda_{K_0}$. This induces a DGA map $\Phi: A_*(K_0) \to A_*(K_1)$ (after appropriate base changes).
4. Injectivity: This induces an injective map between the augmentation varieties:
   $$V_k(K_1) \hookrightarrow V_k(K_0)$$
   defined by a specific monomial transformation involving integers $a, b, c, d$.

B. The Arithmetic Argument (The Novelty)

The core innovation lies in the choice of the coefficient field $k$.

• Previous Limitation: In prior work (e.g., [15]), the field was $\mathbb{C}$. Since $\mathbb{C}$ is algebraically closed, polynomial equations always have roots, making it difficult to derive contradictions based on the existence of roots.
• Current Approach: The author sets $k = \mathbb{Q}$ (the rational numbers).
• Strategy:
  1. Derive a specific polynomial constraint $P_m(T)$ (or $P(T)$) with coefficients in $\mathbb{Z}[\mu^{\pm}, U^{\pm}]$ such that any point in $V_{\mathbb{Q}}(K)$ must satisfy the condition that $P_m|_{\mu=y, U=Z}$ has a root in $\mathbb{Q}$.
  2. Show that for the unknot $O$, the variety $V_{\mathbb{Q}}(O)$ contains points $(y, Z)$ where the corresponding polynomial does not have a root in $\mathbb{Q}$.
  3. Use Hilbert's Irreducibility Theorem and the Rational Root Theorem to prove that for specific knot types, the required polynomial cannot have rational roots, thereby proving the injective map cannot exist.

3. Key Contributions

1. Refinement of Lagrangian Intersection Constraints: The paper establishes a stronger version of the relationship between augmentation varieties of knots connected by a Hamiltonian isotopy. It explicitly tracks the dependence on the coefficient ring $\mathbb{Z}[\lambda^{\pm}, \mu^{\pm}, U^{\pm}]$ rather than reducing to $\mathbb{Z}[\lambda^{\pm}, \mu^{\pm}]$ (setting $U=1$), which is crucial for the arithmetic argument.
2. Algebraic Constraints on Augmentation Varieties: The author derives explicit polynomial relations $P_m(T)$ and $P(T)$ that must be satisfied by the coordinates of the augmentation variety for connected sums involving torus knots and the figure-eight knot.
   • For $K' \# T(2, 2m+1)$, a polynomial $P_m$ is constructed using recursive sequences $h_m, k_m$.
   • For $K' \# 4_1$ (figure-eight knot), a cubic polynomial $P$ is explicitly computed.
3. Proof of Non-Existence via Arithmetic: The paper proves that for specific knots, the augmentation variety over $\mathbb{Q}$ is "too small" to contain the image of the unknot's augmentation variety under the required map. This is achieved by showing that the image would require the existence of rational roots for polynomials that are irreducible or lack rational roots.

4. Main Results

Theorem 1.1 (Main Theorem):
Let $K$ be a knot in $\mathbb{R}^3$ that is a connected sum of an arbitrary knot $K'$ with either:

• A $(2, 2m+1)$-torus knot $T(2, 2m+1)$ where $m \in \mathbb{Z} \setminus \{0, -1\}$, OR
• The figure-eight knot ($4_1$).

Then, there is no compactly supported Hamiltonian diffeomorphism $\phi \in \text{Ham}_c(T^*\mathbb{R}^3)$ such that $\phi(L_K)$ intersects the zero section $\mathbb{R}^3$ cleanly along the unknot.

Proof Sketch for Torus Knots ($K' \# T(2, 2m+1)$):

• The augmentation variety of the unknot $V_{\mathbb{Q}}(O)$ contains points $(y, Z)$ where $y \neq Z^n$ for any integer $n$.
• For the knot $K = K' \# T(2, 2m+1)$, the augmentation variety is constrained by the polynomial $P_m|_{\mu=y, U=Z}$.
• Using Hilbert's Irreducibility Theorem, the author shows there exist infinitely many rational pairs $(y, Z)$ in $V_{\mathbb{Q}}(O)$ such that $P_m|_{\mu=y, U=Z}$ is irreducible over $\mathbb{Q}$ (or has degree $\ge 2$ and no rational roots).
• Since the map $V_{\mathbb{Q}}(O) \to V_{\mathbb{Q}}(K)$ requires the image to be a root of this polynomial, the existence of such points in the domain with no corresponding roots in the codomain creates a contradiction.

Proof Sketch for Figure-Eight Knot ($K' \# 4_1$):

• A specific cubic polynomial $P(T)$ is derived.
• By setting specific rational values $\mu = -1$ and $U = -2$ (which are valid for the unknot), the polynomial becomes $4 - 3T - 4T^3$.
• Applying the Rational Root Theorem, it is shown that this polynomial has no rational roots.
• This contradicts the requirement that the image of the unknot's augmentation must map to a root of $P(T)$ in $\mathbb{Q}$.

5. Significance

• Rigidity of Knot Types: The paper provides a definitive negative answer to the question of whether "knottedness" of Lagrangian conormal bundles can be destroyed by Hamiltonian isotopy for a broad class of knots. It establishes that the topological type of the knot is an invariant of the clean Lagrangian intersection under Hamiltonian isotopy for these specific cases.
• Methodological Shift: It demonstrates the power of combining symplectic topology (SFT/DGAs) with number theory (arithmetic of fields). By moving from algebraically closed fields ($\mathbb{C}$) to non-closed fields ($\mathbb{Q}$), the author extracts topological constraints that were invisible in the complex setting.
• Generalization: The results generalize previous rigidity theorems (which were limited to the unknot) to connected sums involving torus knots and the figure-eight knot, suggesting a deeper connection between the arithmetic properties of knot invariants and symplectic rigidity.
• Technical Advancement: The paper refines the understanding of the coefficient rings in knot DGAs, showing that retaining the $U$ variable (related to the fiber class) is essential for detecting these arithmetic obstructions.

In summary, Okamoto proves that for a wide class of knots, the conormal bundle cannot be Hamiltonianly isotoped to intersect the zero section along the unknot, using a novel argument that leverages the lack of rational roots in specific polynomials derived from the knot's DGA.