Colored Heegaard Floer Homology

• Colored Heegaard Floer homology is a framework that encodes ‘colors’ via basepoint algebra and cable stabilization to enhance invariants for knots and links.
• It integrates multi-pointed Heegaard diagrams, cobordism maps, and hyperboxes to systematically propagate colored data across surgeries and link components.
• Stable-limit constructions using cable knots yield explicit module structures, recurrence relations, and holonomic properties that refine classical invariants.

Colored Heegaard Floer homology denotes a family of constructions in Heegaard Floer theory in which “color” is encoded either by multivariable basepoint algebra or by stable limits of cable knot Floer complexes. In the multi-pointed and link Floer formalisms, a coloring identifies variables attached to basepoints or components and is propagated by cobordism maps, hyperboxes, and surgery formulas. In recent stable-limit constructions, the color $S^r$ or $n$ is realized by taking cables such as $K_{r,rn+1}$ or $K_{n,mn}$ and passing to a limit under full-twist maps, producing Heegaard Floer analogues of colored Khovanov and Khovanov–Rozansky theories (Zemke, 2015, Manolescu et al., 2010, Cooper et al., 2 Jan 2025, Alishahi et al., 29 Aug 2025).

1. Foundational meanings of color in Heegaard Floer theory

The papers considered here use “color” in related but non-identical senses. In the graph cobordism formalism, the basic object is a multi-pointed complex over a polynomial ring in the basepoint variables. For a multi-pointed Heegaard diagram $$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$ and $s\in\mathrm{Spin}^c(Y)$, the minus complex is the free $\mathbb{F}_2[U_{\mathbf{w}}]$–module

$$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$

with differential

$$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$

A coloring is a map $\sigma:\mathbf{w}\to \mathcal{P}$ to a finite set of colors, and it induces

$n$0

where $n$1. In this sense, coloring means identifying variables according to prescribed color classes (Zemke, 2015).

In link Floer surgery theory, color is attached to link components and to the bookkeeping of sublinks, orientations, and handle blocks. The generalized subcomplexes $n$2, the free complexes $n$3 in the link-minimal case, and the complete systems of hyperboxes assemble colored data for every sublink and orientation choice. The reduction maps

$n$4

record the linking correction in the $n$5-th color. The surgery complex

$n$6

packages the colored cube-of-complexes for integral surgery, while four-colored framed links encode 1–, 2–, 2–, and 3–handle blocks in closed 4–manifold constructions (Manolescu et al., 2010).

Framework	Coloring mechanism	Primary output
Multi-pointed graph TQFT	$n$7 identifies $n$8-variables	Colored modules and cobordism maps
Link surgery hyperboxes	Components and sublinks carry the colors	Surgery complexes and mixed invariants
Stable cable limits	Color realized by cabling and stabilization	Colored knot Floer homology

This foundational usage is algebraic and functorial. It predates the stable-limit constructions and supplies part of the ambient language in which “colored” Heegaard Floer structures are organized.

2. Stable-limit constructions of colored knot Floer homology

A recent direction defines colored knot Floer homology by taking stable limits of cable complexes. In "Colored knot Floer homology: structures and examples" (Alishahi et al., 29 Aug 2025), for an oriented $n$9-component link $K_{r,rn+1}$0 and an unknot $K_{r,rn+1}$1 bounding a disk $K_{r,rn+1}$2 that intersects each component $K_{r,rn+1}$3 positively exactly once, one lets $K_{r,rn+1}$4 be obtained from $K_{r,rn+1}$5 by inserting $K_{r,rn+1}$6 full twists supported near $K_{r,rn+1}$7. The “full version” link Floer chain complex $K_{r,rn+1}$8 is defined over

$K_{r,rn+1}$9

with the relation that $K_{n,mn}$0 are pairwise chain-homotopic, hence act as a common variable $K_{n,mn}$1 on homology. The connecting maps

$K_{n,mn}$2

are link Floer cobordism maps induced by blowing down the $K_{n,mn}$3–framed unknot $K_{n,mn}$4 in the $K_{n,mn}$5 structure $K_{n,mn}$6 satisfying

$K_{n,mn}$7

For the colored theory one uses $K_{n,mn}$8, renormalizes Alexander grading by subtracting $K_{n,mn}$9 from each coordinate, and defines

$$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$0

For a knot $$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$1 and its $$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$2–cables,

$$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$3

A parallel but distinct construction appears in "Holonomicity from a Heegaard-Floer Perspective" (Cooper et al., 2 Jan 2025). There the color $$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$4 is realized via $$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$5–cables $$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$6 and a stable limit of their knot Floer complexes. Using the immersed-curve model for bordered Heegaard Floer theory, one computes $$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$7 as intersection Floer homology $$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$8 in the appropriate cover $$H=(\Sigma,\boldsymbol{\alpha},\boldsymbol{\beta},\mathbf{w})$$9, then colors by taking $s\in\mathrm{Spin}^c(Y)$0 parallel horizontally-scaled copies, shifting copies of $s\in\mathrm{Spin}^c(Y)$1 vertically, and using a single zig-zag meridian $s\in\mathrm{Spin}^c(Y)$2. After shifting so that the bottom generator $s\in\mathrm{Spin}^c(Y)$3 lies in bidegree $s\in\mathrm{Spin}^c(Y)$4, the colored filtered chain complex is

$s\in\mathrm{Spin}^c(Y)$5

The limit exists by Rozansky’s convergence criterion, and its homology is a knot invariant.

The two constructions differ in flavor and indexing: one is a colimit in the full link Floer setting over $s\in\mathrm{Spin}^c(Y)$6–cables, and the other is a stable limit in the hat setting over $s\in\mathrm{Spin}^c(Y)$7–cables. Both realize color by cabling and stabilization, and both are explicitly framed as Heegaard Floer analogues of colored Khovanov-type theories.

3. Gradings, stabilization, and the geometry of the limit

The stable-limit theories are controlled by precise grading data. In the full link Floer construction, $s\in\mathrm{Spin}^c(Y)$8 carries a Maslov grading $s\in\mathrm{Spin}^c(Y)$9 and an Alexander multigrading

$\mathbb{F}_2[U_{\mathbf{w}}]$0

preserved by the differential, with

$\mathbb{F}_2[U_{\mathbf{w}}]$1

For the full-twist cobordism maps, the degree shifts are

$\mathbb{F}_2[U_{\mathbf{w}}]$2

After normalizing by $\mathbb{F}_2[U_{\mathbf{w}}]$3, the map $\mathbb{F}_2[U_{\mathbf{w}}]$4 preserves both $\mathbb{F}_2[U_{\mathbf{w}}]$5 and the normalized Alexander grading $\mathbb{F}_2[U_{\mathbf{w}}]$6 (Alishahi et al., 29 Aug 2025).

The central stabilization theorem states that there exists a constant $\mathbb{F}_2[U_{\mathbf{w}}]$7 independent of $\mathbb{F}_2[U_{\mathbf{w}}]$8 such that if $\mathbb{F}_2[U_{\mathbf{w}}]$9 satisfies $$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$0 for all $$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$1, then there is a canonical isomorphism

$$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$2

Consequently, each graded piece of $$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$3 is finite-dimensional. The proof uses special Heegaard diagrams with winding blocks near the $$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$4–basepoints, the generator labels $$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$5, a normalization lemma giving

$$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$6

and upper bounds on the Alexander gradings of interior generators. The paper also formulates the conjecture that for fixed $$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$7, the maps $$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$8 are isomorphisms for $$CF^-(H,s):=\bigoplus_{\substack{x\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta\ s_{\mathbf{w}}(x)=s}} \mathbb{F}_2[U_{\mathbf{w}}]\cdot x,$$9 (Alishahi et al., 29 Aug 2025).

In the $$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$0-colored hat theory, the Alexander and Maslov gradings scale with the color in a controlled manner. If $$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$1 is the bottom generator, then

$$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$2

while Maslov differences are affine linear in $$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$3:

$$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$4

For $$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$5, specifically $$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$6, the cable complex splits into a “head” supported in the bottom row and a long “tail” supported along the horizontal connecting arc across rows. This yields the canonical decomposition

$$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$7

where $$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$8 with $$\partial x=\sum_{y\in \mathbb{T}_\alpha\cap \mathbb{T}_\beta}\ \sum_{\substack{\phi\in\pi_2(x,y)\ \mu(\phi)=1}} \#\mathcal{M}_J(\phi)\cdot \Big(\prod_{i=1}^n U_{w_i}^{n_{w_i}(\phi)}\Big)\cdot y.$$9, and $\sigma:\mathbf{w}\to \mathcal{P}$0 is a graded shift of the colored unknot complex (Cooper et al., 2 Jan 2025).

These stabilization statements supply the analytic and diagrammatic core of the colored theories. They replace a single finite model by a directed or convergent family whose asymptotic behavior is rigid enough to define an invariant.

4. Algebraic structures: cable algebras, colored unknots, and holonomic recurrence

The colimit theory carries a nontrivial algebra action. The cable algebra $\sigma:\mathbf{w}\to \mathcal{P}$1 is the $\sigma:\mathbf{w}\to \mathcal{P}$2–graded $\sigma:\mathbf{w}\to \mathcal{P}$3–algebra, where

$\sigma:\mathbf{w}\to \mathcal{P}$4

generated by $\sigma:\mathbf{w}\to \mathcal{P}$5 of degrees

$\sigma:\mathbf{w}\to \mathcal{P}$6

subject to the linear relations

$\sigma:\mathbf{w}\to \mathcal{P}$7

and the quadratic relations

$\sigma:\mathbf{w}\to \mathcal{P}$8

If

$\sigma:\mathbf{w}\to \mathcal{P}$9

then $n$00 is an $n$01–module in which $n$02 acts by $n$03. For the unlink $n$04, one has

$n$05

and this is a free rank-1 $n$06–module generated by $n$07 (Alishahi et al., 29 Aug 2025).

Localizing by $n$08 gives the colored algebra

$n$09

which admits the explicit presentation

$n$10

with

$n$11

The colimit $n$12 is then a graded module over $n$13, and for the unknot $n$14,

$n$15

As a graded vector space, this is $n$16 with the colored relations $n$17.

A different but complementary algebraic structure appears in the $n$18-colored hat theory. For sequences $n$19, one defines Weyl algebra operators

$n$20

satisfying $n$21. On the dg level there are functors $n$22 with a natural isomorphism

$n$23

categorifying the Weyl relation. A sequence is holonomic if it can be assembled, by a finite sequence of distinguished triangles, from the thick subcategory generated by the Weyl action. The main theorem states that for every knot $n$24, the sequence $n$25 is homologically $n$26-holonomic (Cooper et al., 2 Jan 2025).

At the Euler characteristic level, if

$n$27

then

$n$28

annihilates the reduced sequence $n$29, and the unreduced colored Alexander sequence is annihilated by

$n$30

This is a Heegaard Floer analogue of an Alexander-side AJ-type operator.

5. Computations and explicit families

For $n$31–space knots, the full colored theory is computable in closed form. If $n$32 is an $n$33–space knot, then the $n$34–cables are $n$35–space links for $n$36, and their link Floer homology is determined by the $n$37–function $n$38, equivalently by $n$39. The resulting theorem is

$n$40

where $n$41 is viewed as a module over $n$42 through

$n$43

Consequently $n$44 is finitely generated over $n$45, generated by the diagonal tower generators of $n$46 (Alishahi et al., 29 Aug 2025).

Writing

$n$47

with $n$48 and $n$49 for $n$50, and denoting the standard generators of $n$51 by $n$52, colored generators $n$53 satisfy

$n$54

and

$n$55

In particular, for $n$56,

$n$57

For torus links $n$58, the homology $n$59 is generated over $n$60 by elements $n$61, $n$62, with

$n$63

$n$64

and relations

$n$65

For the unknot,

$n$66

and for the small example $n$67 one has colored generators $n$68 with relations

$n$69

(Alishahi et al., 29 Aug 2025).

The hat-theoretic stable-limit construction also admits explicit formulas. For the unknot,

$n$70

with $n$71, $n$72 and degrees

$n$73

Its Poincaré series is

$n$74

and

$n$75

For the right-handed trefoil,

$n$76

with $n$77 and

$n$78

Its reduced recurrence is governed by

$n$79

and the unreduced sequence is annihilated by $n$80 (Cooper et al., 2 Jan 2025).

Under the stabilization conjecture in the full-link theory, the normalized Euler characteristic stabilizes and satisfies

$n$81

In the $n$82-colored hat theory,

$n$83

These formulas make the Alexander-theoretic shadow of the colored constructions completely explicit.

6. Functoriality, comparison with other theories, and open questions

Colored Heegaard Floer structures are functorial in several senses. In the graph TQFT, a ribbon graph cobordism $n$84 induces chain maps

$n$85

equivariant over the appropriate colored ring. The construction factors through 1–, 2–, and 3–handle maps, free-stabilization maps

$n$86

and relative homology maps $n$87, $n$88. The trivial strand relation

$n$89

shows that adding a parallel strand of the same color does not change the map (Zemke, 2015).

In the full stable-limit theory, if $n$90 and $n$91 differ by one crossing, then there are colored crossing-change maps

$n$92

with degree shifts

$n$93

$n$94

and both commute with the $n$95–module action. The same paper records isotopy invariance, compatibility with composition, a braid-group action descending to an $n$96–action on $n$97 and $n$98, an orientation convention using $n$99 when some components intersect the disk negatively, and framing shifts that change normalized Alexander grading but not the vector space or Maslov gradings (Alishahi et al., 29 Aug 2025).

The relationship with colored Khovanov, colored Khovanov–Rozansky, and $K_{r,rn+1}$00–ified homology is algebraically concrete. The colored algebra relation

$K_{r,rn+1}$01

specializes the polynomial action in colored triply-graded homology

$K_{r,rn+1}$02

under

$K_{r,rn+1}$03

This supports a conjectural compatibility between colored KR and colored HFK via spectral sequences and large-color limits. In the $K_{r,rn+1}$04-colored hat theory, a conjectural spectral sequence from colored HOMFLY homology to knot Floer homology is one of the explicit motivations for the construction, and homological $K_{r,rn+1}$05-holonomicity is presented as consistent with that framework (Alishahi et al., 29 Aug 2025, Cooper et al., 2 Jan 2025).

Several basic questions remain open. In the full-link stable-limit theory, the isomorphism conjecture for $K_{r,rn+1}$06 is proved for $K_{r,rn+1}$07–space knots but open in general; stabilization for $K_{r,rn+1}$08–cables with fixed remainder $K_{r,rn+1}$09 is expected but not established at the chain level; and finite generation of $K_{r,rn+1}$10 as an $K_{r,rn+1}$11–module is known for $K_{r,rn+1}$12–space knots and open for arbitrary $K_{r,rn+1}$13. Extensions to links with mixed orientations, satellites and patterns, hat and minus flavors, 3–manifolds via trace cobordisms, and comparisons with plumbed-link formality are explicitly listed as further directions (Alishahi et al., 29 Aug 2025). In the holonomicity framework, open problems include extending the construction to links and to colors beyond the symmetric power $K_{r,rn+1}$14, making the spectral sequence from colored HOMFLY homology precise, studying cobordism functoriality of the recurrence, and relating the noncommutative annihilator $K_{r,rn+1}$15 to a Floer-theoretic Alexander $K_{r,rn+1}$16-polynomial (Cooper et al., 2 Jan 2025).

Taken together, these developments show that colored Heegaard Floer homology is not a single invariant but a structured domain of Heegaard Floer theory. Its algebraic basepoint-coloring formalisms govern cobordisms, surgeries, and 4–manifold constructions, while its stable-limit cable constructions produce genuine colored knot Floer homologies with module structures, explicit examples, Euler characteristic formulas, and categorified recurrence relations.