Khovanov Homology for Tangles in Connected Sums

This paper extends Khovanov homology to links in 3-manifolds that are connected sums of orientable interval bundles over surfaces by constructing type D and type A tangle invariants that, when glued along a separating sphere, recover the link's Khovanov homology.

The Gist

Imagine you have a giant, complex knot made of string. In mathematics, this is called a link. For a long time, mathematicians had a way to describe these knots using a special formula called the Jones Polynomial. It's like a barcode for a knot: if two knots have different barcodes, they are definitely different. But if they have the same barcode, they might be the same, or they might just be very similar.

Then, a mathematician named Mikhail Khovanov came along and said, "Let's upgrade that barcode." Instead of a single number (the polynomial), he created a massive, multi-layered library of information called Khovanov Homology. Think of it as taking that single barcode and turning it into a 3D hologram. If two knots have the same hologram, they are definitely the same knot. This "categorification" (turning a number into a structure) is a huge deal in math.

The Problem:
So far, this hologram library only worked for knots floating in a perfect, empty sphere (the 3-sphere, or $S^3$). But what if the universe isn't empty? What if the space around the knot is weird? Imagine the space is made of several "rooms" (3D shapes) glued together. In math, this is called a connected sum. The paper by Alan Du asks: Can we build this hologram library for knots living in these weird, glued-together universes?

The Solution: The "Tangle" Strategy
Instead of trying to build the whole library at once for a knot in a weird universe, Du uses a clever construction trick. He treats the universe like a house with a separating wall.

1. Cutting the House: Imagine you have a weird 3D shape made of several rooms glued together. You slice it right down the middle with a giant, invisible sphere. Now you have two halves: a Left Half and a Right Half.
2. The Tangles: The knot might pass through this cut. The part of the knot in the Left Half is a "Left Tangle" (it has loose ends sticking out into the cut). The part in the Right Half is a "Right Tangle."
3. The Two Languages (Type A and Type D):
   • For the Left Tangle, Du builds a "Type A Structure." Think of this as a Menu. It lists all the possible ways the knot can behave on the left side, but it's written in a language that expects to be fed information from the right.
   • For the Right Tangle, he builds a "Type D Structure." Think of this as a Order Form. It lists the possibilities on the right side and is ready to send information to the left.
4. The Glue (The Box Tensor Product):
   • Now, you take the Menu (Left) and the Order Form (Right) and smash them together. In math, this is called a Box Tensor Product.
   • When you combine them, the loose ends match up perfectly. The "Menu" and the "Order Form" talk to each other, canceling out the loose ends, and suddenly, you have the full hologram (Khovanov Homology) for the entire knot in the weird universe.

The Magic of "Handleslides" and "Mirror Moves"
To make sure this library is real and not just a trick, Du has to prove that it doesn't matter how you draw the knot.

• Reidemeister Moves: You can wiggle the string, twist it, or untwist it without cutting it. The math proves the library stays the same.
• Finger and Mirror Moves: Because the universe is weird (it has these "rooms" or surfaces), you can push a piece of the knot through a wall or reflect it. Du had to invent new rules (like "Finger Moves" and "Mirror Moves") to handle these weird topological tricks. He proved that even if you do these strange moves, the final hologram remains unchanged.

The "Cleaved Links" Algebra
To make all this math work, Du had to invent a new dictionary called the Cleaved Links Algebra.

• Imagine the knot is cut by the separating sphere. The pieces of the knot that touch the cut are "cleaved."
• This algebra is a set of rules for how these cut pieces can interact. It's like a rulebook for a board game where the pieces are the cut ends of the string.
• Du defined specific "moves" in this rulebook (like merging two loops or splitting one loop) and proved that the order in which you make these moves doesn't change the final result. This ensures the math is consistent.

Why Does This Matter?
This paper is a bridge. It takes a powerful tool (Khovanov Homology) that was stuck in a simple, perfect world and taught it how to survive in complex, "glued-together" worlds.

• Real-world analogy: It's like taking a GPS that only works in a flat, empty parking lot and upgrading it so it can navigate a city with tunnels, bridges, and multi-level parking garages.
• Specific Achievement: It specifically solves the problem for spaces made of "Interval Bundles over Surfaces." This includes tricky shapes like the Real Projective Space ($RP^3$), which is a fundamental shape in topology that behaves very differently from our normal 3D space.

In a Nutshell:
Alan Du figured out how to take a knot in a weird, multi-room universe, cut it in half, describe each half with a specialized "language" (Type A and Type D), and then glue those descriptions back together to get a perfect, unshakeable mathematical fingerprint of the whole knot. He did this by inventing a new rulebook (the algebra) and proving that even if you wiggle the knot through the walls of the universe, the fingerprint stays the same.

Technical Summary

Here is a detailed technical summary of the paper "Khovanov Homology for Tangles in Connected Sums" by Alan Du.

1. Problem Statement

Khovanov homology is a powerful invariant for links in the 3-sphere ($S^3$) that categorifies the Jones polynomial. While it has been extended to various 3-manifolds (e.g., $S^2 \times S^1$, $\mathbb{R}P^3$, and orientable interval bundles over surfaces), a unified framework for links in connected sums of orientable interval bundles over surfaces ($M = M_1 \# \dots \# M_r$) was lacking.

Specifically, the paper addresses the challenge of extending Lawrence Roberts' tangle-based construction of Khovanov homology (which uses type $D$ and type $A$ structures over a differential graded algebra) to these more complex 3-manifolds. The goal is to define a Khovanov homology for links in $M$ by cutting the manifold along separating spheres, constructing invariants for the resulting tangles in the two halves, and recovering the global invariant via a box tensor product.

2. Methodology

The author employs a "cut-and-glue" strategy, generalizing Roberts' work on $S^3$ to the connected sum setting. The methodology involves the following key steps:

A. Geometric Setup and Diagrams

• Manifold Decomposition: The 3-manifold $M = M_1 \# \dots \# M_r$ is cut along $r-1$ separating spheres ($S^2$). This decomposes $M$ into pieces: $M_1 \setminus \mathring{D}^3$, $M_i \setminus (\mathring{D}^3 \sqcup \mathring{D}^3)$, and $M_r \setminus \mathring{D}^3$.
• Surface $\Sigma$: These pieces are glued along their boundaries to form a surface $\Sigma$ embedded in $M$. Links in $M$ are projected onto $\Sigma$.
• Tangle Decomposition: A link $L$ is split into two tangles, $T_1$ (left) and $T_2$ (right), separated by a specific sphere $\partial D_p$. The tangles live in $M_{left} \setminus \mathring{D}^3$ and $M_{right} \setminus \mathring{D}^3$.
• Isotopy Moves: The paper establishes that isotopies in $M$ correspond to standard Reidemeister moves plus specific moves across the separating spheres: Finger moves, Mirror moves, and Handleslides.

B. The Cleaved Links Algebra ($B\Gamma_n$)

The core algebraic structure is a bigraded differential algebra $B\Gamma_n(M, p)$, analogous to the algebra used in Roberts' $S^3$ construction but adapted for the connected sum.

• Vertices: Decorated $n$-cleaved links (links where every component intersects the separating sphere $\partial D_p$) with signs ($+$ or $-$) assigned to components.
• Edges: Generated by bridges (arcs connecting components) and decoration changes.
  • Bridge edges: Correspond to surgery on arcs (merging or splitting circles).
  • Decoration edges: Correspond to changing the sign of a component.
• Relations: The algebra is defined by quotienting a quiver algebra by relations derived from commuting squares of bridges. Crucially, the relations distinguish between "left" (Type A) and "right" (Type D) bridges to handle the non-trivial topology of the connected sum.
• Differential: A $(1,0)$-differential $d_{B\Gamma_n}$ is defined, primarily acting on the left side (Type A) via specific bridge surgeries that correspond to 1-1 bifurcations or sign changes.

C. Type $D$ and Type $A$ Structures

• Type $D$ Structure ($J\!\!T_2\!\!\rangle$): Defined for the right tangle $T_2$. It is a left module over $B\Gamma_n$ equipped with a map $\delta: J\!\!T_2\!\!\rangle \to B\Gamma_n \otimes J\!\!T_2\!\!\rangle$. The differential involves the APS differential (from Asaeda-Przytycki-Sikora) and the algebra action.
• Type $A$ Structure ($\langle\!\langle T_1\!\!|$): Defined for the left tangle $T_1$. It is an $A_\infty$-module over $B\Gamma_n$ with operations $m_1$ (differential) and $m_2$ (action). Higher operations $m_i$ for $i \ge 3$ vanish.
• Invariance: The author proves that these structures are invariant under Reidemeister moves, finger moves, mirror moves, and handleslides. A key technical hurdle is proving invariance under handleslides across the separating spheres, which requires careful handling of "1-1 bifurcations" (surgeries that split a circle into two and then merge them back, or vice versa) and the use of "tangle diagrams with nodes" to track homotopy classes.

D. Pairing and Recovery

• Box Tensor Product: The global invariant is recovered by taking the box tensor product of the Type $A$ and Type $D$ structures: $\langle\!\langle T_1\!\!| \boxtimes J\!\!T_2\!\!\rangle$.
• Direct Construction: The author also defines a direct Khovanov chain complex $CKh(L)$ for the link $L$ in $M$ using resolutions and decorations, similar to the standard Khovanov construction but adapted for the surface $\Sigma$.

3. Key Contributions

1. Generalization of Roberts' Construction: The paper successfully extends the type $D$/type $A$ tangle decomposition of Khovanov homology from $S^3$ to connected sums of orientable interval bundles over surfaces.
2. Handling Handleslides: A significant technical contribution is the proof of invariance under handleslides across the separating spheres. The author demonstrates that the specific differential chosen (which includes maps merging/splitting non-trivial circles) is necessary for invariance, contrasting with previous constructions (like APS) where such invariance failed.
3. Unified Framework for $\#_r \mathbb{R}P^3$: As a special case, the construction provides a definition of Khovanov homology for links in $\#_r \mathbb{R}P^3$ (viewed as connected sums of twisted interval bundles over $\mathbb{R}P^2$) with $\mathbb{Z}$-coefficients, resolving previous limitations regarding coefficients in $\mathbb{R}P^3$.
4. Isomorphism Theorem: The paper proves that the box tensor product of the tangle invariants is isomorphic to the directly constructed Khovanov chain complex of the link.

4. Main Results

• Theorem 1.1: The chain homotopy type of the box tensor product $\langle\!\langle T_1\!\!| \boxtimes J\!\!T_2\!\!\rangle$ is an invariant of the link $L = T_1 \natural T_2$ under isotopy in $M$.
• Theorem 1.2: The directly constructed Khovanov chain complex $(CKh(L), d)$ is isomorphic to the box tensor product $(\langle\!\langle T_1\!\!| \boxtimes J\!\!T_2\!\!\rangle, \partial_\boxtimes)$. Consequently, the homology $Kh(L)$ is an invariant of the link.
• Invariance under Decomposition: Corollary 9.2 shows that the resulting homology is independent of the choice of the separating sphere used to decompose the link into tangles.

5. Significance

This work represents a major step forward in the categorification of link invariants in 3-manifolds. By generalizing the tangle decomposition method, it allows for the computation of Khovanov homology in complex 3-manifolds by breaking them down into simpler pieces (tangles) and using algebraic gluing.

• Computational Utility: The decomposition into Type $A$ and Type $D$ structures is computationally advantageous, as it allows for the calculation of invariants for large or complex links by computing smaller tangle invariants and combining them.
• Theoretical Depth: The rigorous handling of handleslides and the specific construction of the differential on the cleaved links algebra provide a robust framework for future extensions to other 3-manifolds or other homology theories (e.g., Heegaard Floer homology).
• Resolution of Coefficient Issues: The construction works with $\mathbb{Z}$-coefficients for links in $\#_r \mathbb{R}P^3$, overcoming previous restrictions to $\mathbb{Z}/2\mathbb{Z}$ coefficients in similar settings.

In summary, Alan Du provides a comprehensive and rigorous extension of Khovanov homology to a broad class of 3-manifolds, unifying previous scattered results and establishing a powerful algebraic toolkit for studying links in connected sums.