Computing Khovanov homology of tangles

Guo-Wei Wei; Jian Liu; Li Shen

REVIEW 3 major objections 3 minor 10 references

Computing Khovanov homology of tangles

T0 review · 3 major / 3 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper proves a closed form for the Poincaré polynomial of any simple tangle — P_T(x,y) = y^{−N+n++n−}(1+xy)^{n+}(x^{−1}y^{−3}+y^{−2})^{n−} — so the full bigraded Khovanov homology is fixed by the strand count and the two signed crossing

desk verdict The simple-tangle formula and tables are useful and appear correct, but the arc reduction theorem that is supposed to prove them is false as stated, not just unproven. read the letter →

arxiv 2508.14398 v1 pith:6UMYF24V submitted 2025-08-20 math.GT math.GN

Li Shen , Jian Liu , Guo-Wei Wei This is my paper · ORCID

classification math.GTmath.GN MSC 57K1857K10

keywords tangleKhovanovhomologysimplepurearcreductionPoincarépolynomiallow-crossingtangles

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Khovanov homology refines the Jones polynomial into a richer invariant, but for tangles — knotted strands with free ends — explicit computations have been scarce. This paper proves that for "simple" tangles (tangles in which no group of strands encloses a region), the entire Khovanov homology over a field is captured by a closed-form Poincaré polynomial: only the number of strands and the counts of right- and left-handed crossings appear. The proof is an arc-reduction: repeatedly peel off a strand that crosses the rest at most once and track how each removed crossing shifts the homological and quantum gradings — a right-handed crossing contributes the factor (1+xy), a left-handed the factor (x^{−1}y^{−3}+y^{−2}). The same machinery supplies explicit Poincaré polynomials for every tangle with at most three crossings, listed by crossing sign. If the reduction is sound, computing Khovanov homology of simple tangles becomes bookkeeping rather than linear algebra.

What carries the argument

The load-bearing object is the arc-reduction theorem (Theorem 2.1), which asserts that adding a single crossing on one arc splits the Khovanov complex into a direct sum of two copies of the original tangle's complex, with the homological and quantum gradings shifted by fixed pairs of integers. The theorem rests on the claim that the TQFT-induced differential between the two smoothings of the added crossing is the zero map, so the complex splits without interaction terms. The second ingredient is the class of "simple" tangles — every arc is pure, meaning both sides of each arc reach the boundary without crossing any other strand — for which Lemma 2.5 guarantees that some arc always intersects

What would settle it

Compute the Khovanov complex of the single-crossing tangle (type 1₁) directly from the defining TQFT: it has two smoothings and exactly one differential — the very map the proof asserts is zero. If that differential has a non-vanishing entry on w⊗w⊗x, Theorem 2.1 fails at the first nontrivial example; the tabulated polynomial x + y^{−1} (right-handed) or y^{−3} + x^{−1}y^{−4} (left-handed) would then be refuted. The same check on a two-crossing tangle such as type 2₃ tests the next induction step.

Watch

Extended reading notes

Core claim

The central claim is Theorem 2.7: if T is a simple tangle with N arcs, n+ right-handed crossings, and n− left-handed crossings, its Poincaré polynomial is P_T(x,y) = y^{−N+n++n−}(1+xy)^{n+}(x^{−1}y^{−3}+y^{−2})^{n−}. Each monomial in the expansion is one generator of the Khovanov homology in a specific bidegree, so the expansion encodes the whole bigraded homology. The route is Theorem 2.1, an arc-reduction rule: a tangle obtained by adding one arc with a single crossing splits into a direct sum of two copies of the smaller tangle's homology, with grading shifts (0,0)+(1,1) per right-handed crossing and (−1,−3)+(0,−2) per left-handed one. Iterating until one arc remains, each free arc remove

Load-bearing premise

The whole reduction rests on an unproved assertion in the proof of Theorem 2.1: the differential between the two smoothings of an added crossing is "necessarily the zero map" (w⊗w⊗x ↦ 0), stated without proof or citation; if that map is nonzero, the direct-sum splitting — and with it the closed-form Poincaré polynomial — does not follow.

Editorial extensions

If this is right

For any simple tangle, the full bigraded Khovanov homology over a field is fixed by three integers (N, n+, n−); the ranks in each bidegree are binomial coefficients, so no matrix calculations are needed.
The classification tables give explicit Poincaré polynomials for every connected tangle with at most three crossings, split by the signs of the crossings, and by the paper's orientation theorem each entry is complete for its (n+, n−) type.
Setting x = −1 in the closed form specializes to the Jones polynomial of the tangle, so the formula subsumes the Jones polynomial for simple tangles.
The 4-arc, three-crossing tangle — the only three-crossing case with N = n+ + n− + 1 — falls directly under the closed form, and Example 2.5 lists all four sign-orientation cases.
The arc-reduction theorem applies one crossing at a time: any tangle one crossing away from a known tangle gets its homology from two copies of the known one with explicit shifts.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If Theorem 2.7 holds, simple tangles are the degenerate base case of Khovanov homology: the invariant carries no hidden pairing or cancellation — nothing beyond the counts of strands and signed crossings — and the genuinely hard content begins with non-simple tangles, where strands enclose regions.
The same arc-peeling bookkeeping would plausibly extend beyond simple tangles to any tangle admitting a sequence of single-crossing arc removals, since Theorem 2.1 itself has no simplicity hypothesis; the closed form would then follow for any tangle with such a 'peeling order.'
A direct test of the orientation claim is suggested by the tables: two orientations of the same tangle with equal (n+, n−) should give identical Poincaré polynomials over a field, a property checkable within the paper's own classification.
The closed form yields a hand-checkable isotopy obstruction: simple tangles with different triples (N, n+, n−) cannot be isotopic, since their Poincaré polynomials differ.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The simple-tangle Poincaré formula (Thm 2.7) and the low-crossing tables are genuinely useful and appear correct. But the arc reduction theorem that is supposed to derive them (Thm 2.1) is not merely unproved; it is false as stated. The stress-test note is right: the assertion that the differential between the two smoothings of an added crossing is zero is wrong. For a crossing of two arcs, the two smoothings have different boundary pairings; the saddle map between them is the standard nonzero differential (for a single crossing it's the nontrivial mapping cone). The proof says both smoothings are T_s ⊔ Ω, which is incorrect.

A concrete check: apply Thm 2.1 to T = single arc (which has homology y^{-1}). For a right-handed crossing, Thm 2.1 gives y^{-1} + x^{-1} y^{-2}, but Table 1 gives y^{-1} + x. The shifts have the wrong sign. In fact, Thm 2.6's reduction step uses the opposite shift ((0,0)+(1,1) for a positive crossing), so Thm 2.1 and Thm 2.6 contradict each other. The later theorem's proof depends on the earlier one, so the derivation collapses.

The rest of the paper is likewise informal: Prop 2.2(iii) applies Euler's formula without checking connectedness, and the classification tables are assembled by hand. The authors lean on their own unpublished [10] for key structural facts.

What is genuinely new here is the closed-form formula for simple tangles and the tables. Those may well be correct. But the proof given does not establish them, and the central theorem is wrong. The authors should either correct Thm 2.1 (the shifts are reversed, and the differential is not zero) or provide a direct proof of Thm 2.7 that doesn't go through the false reduction. As written, this should be rejected. I wouldn't send it to a referee until the main theorem is fixed.

Referee Report

3 major / 3 minor

Summary. The paper proposes an arc-reduction method for computing Khovanov homology of tangles. Theorem 2.1 claims that adding a single crossing to a tangle splits the Khovanov chain complex into a direct sum of two copies of the original complex, with explicit homological and quantum grading shifts. Using this reduction, the authors define 'simple tangles' and derive a closed-form Poincaré polynomial (Theorem 2.7) in terms of the number of arcs and the numbers of positive and negative crossings. The paper then gives classification tables for tangles with at most three crossings and computes their Poincaré polynomials.

Significance. If the main theorem were valid, it would provide an unusually simple closed-form description of the Khovanov homology of a large class of tangles, which would be a useful computational tool. The explicit low-crossing tables are also potentially valuable. However, the central proof rests on an unproved and likely false assertion about the TQFT differential, and the supporting combinatorial lemmas contain serious gaps. No machine-checked proofs or code are provided, and several structural facts are cited to an unpublished preprint. The claimed results are therefore not established at the level required for publication.

major comments (3)

[§2.1, proof of Theorem 2.1] The proof rests entirely on the assertion that G(d(⋆,ζ2)) is 'necessarily the zero map' (w⊗w⊗x ↦ 0). This is neither proved nor cited, and it is incompatible with the standard tangle Khovanov TQFT used in the paper. The two smoothings of a crossing between two arcs pair the four boundary endpoints differently; they are not the same TQFT object in the way asserted when the text writes T'(0,s)=T_s⊔Ω and T'(1,s)=T_s⊔Ω. In the standard construction (Bar-Natan [1], Khovanov [5]), the saddle map between the two smoothings of a crossing is the nonzero merge/split map; for the one-crossing two-arc tangle this is the differential of a nontrivial mapping cone. If that map is nonzero, ∂T' does not preserve A, Eq. (1) collapses, and Theorems 2.6, 2.7, and the Section 3 tables lose their derivation. This is the load-bearing step of the paper.
[§2.2, Proposition 2.2(iii)] The proof applies Euler's formula m+f−l=2 to the graph formed by N arcs and m crossings. Euler's formula in this form requires a connected planar graph; the figure formed by N arcs can be disconnected, so the formula should include the number of connected components. The step 'f=1' is also not justified by 'no closed regions' unless connectedness and a fixed planar embedding are assumed. Moreover, the counting 'at least 2N−m are shared' is not established; for N=2,m=1 it would claim three shared incident arc ends, which is impossible. Since Lemma 2.5 and Theorem 2.6 depend on this bound, this is a second load-bearing gap.
[§3.2, four-arc row of Table 2] The classification of the four-arc, three-crossing tangle as simple uses Proposition 2.4, whose proof is also not sound: from a closed region formed by n arcs it infers that the remaining N−n arcs contribute at least N−n crossings, but crossings can be shared between the two groups, and no argument is given. This matters because the last block of Table 2 is computed from Theorem 2.7 and therefore inherits the problems of the main proof.

minor comments (3)

[§3.2] There is an incomplete sentence: 'Consider the case of a tangle with four arcs and three crossings. Note that .'
[§3] The classification of tangles with at most three crossings is presented as exhaustive, but the enumeration arguments are informal and are not accompanied by a verifiable algorithm or computer check. Given the failure of the main theorem, independent verification of the tables would be needed.
[§3.1] Several structural facts, including the tensor-product decomposition of Khovanov homology for disjoint unions and the statement that tangles with the same (n_+, n_-) have the same homology, are cited to the unpublished preprint [10]. These should be proved in the paper or cited to a publicly available source.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the simple-tangle formula is derived from the arc-reduction theorem, not assumed; the unproved zero-map assertion is a correctness gap, not a circular reduction.

full rationale

The central derivation chain is not circular. Theorem 2.7's Poincaré polynomial is obtained by translating the generator expansion in Theorem 2.6, which is obtained by iterating the arc-reduction Theorem 2.1 on a simple tangle. The input data (N, n+, n-) are not fitted to the output; the formula is a closed-form consequence of the reduction rule plus purity. The proof of Theorem 2.1 does contain a load-bearing but unproved assertion that the saddle map G(d(⋆,ζ2)) is 'necessarily the zero map' (Section 2.1). That is an omitted proof or potential mathematical error, not a circular step: the theorem is not defined in terms of the conclusion it is used to prove. Similarly, the paper cites the authors' own [6] for the underlying TQFT functor and [10] for component tensor-product and orientation-independence facts used in the tabulation section. These are self-citations and make the paper not fully self-contained, but the simple-tangle Poincaré polynomial does not reduce to those citations by construction; no parameter is fitted and no prediction is renamed as an input. Therefore the circularity score is 0, with the caveat that correctness is not assessed here.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The central formula rests on the TQFT functor from the authors' prior work [6], on an unproven zero-differential assertion, on an Euler formula application that may require connectedness, and on an orientation-independence theorem from the unpublished [10]. No free parameters are fitted.

assumptions (3)

domain assumption The Khovanov TQFT functor from the tangle cobordism category to K[Z×Z]-modules, as constructed in [6], computes tangle homology.
The arc reduction proof in Section 2.1 works entirely inside this functor; the behavior of the differential (zero map) is a property of this functor that the paper asserts but does not prove.
domain assumption Euler's formula m+f-l=2 applies to the arc graphs considered in the proof of Proposition 2.2(iii).
The proof uses Euler's formula for planar graphs but does not establish that the graph formed by the N arcs is connected, which is required for the form m+f-l=2.
domain assumption For tangles, if two orientations have the same (n+, n-), their Khovanov homologies coincide ([10, Theorem 4.2]).
Used in Section 3.1 to compute polynomials for all orientations from the sign of crossings. The cited paper [10] is in preparation and not accessible.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Computing Khovanov homology of tangles." pith.science (2026). https://pith.science/paper/6UMYF24V

@misc{pith2026250814398,
  author       = {Pith},
  title        = {Pith review of: Computing Khovanov homology of tangles},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6UMYF24V}},
  note         = {Machine review of arXiv:2508.14398}
}

read the original abstract

The computation of Khovanov homology for tangles has significant potential applications, yet explicit computational studies remain limited. In this work, we present a method for computing the Khovanov homology of tangles via an arc reduction approach, and we derive the Poincar\'e polynomial for simple tangles. Furthermore, we compute the Poincar\'e polynomials of tangles with at most three crossings.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

10 extracted references · 9 canonical work pages

[10]

Khovanov homology of tangles: algorithm and compu- tation.In preparation, 2025

Li Shen, Jian Liu, and Guo-Wei Wei. Khovanov homology of tangles: algorithm and compu- tation.In preparation, 2025. 15

work page 2025
[1]

On Khovanov’s categorification of the Jones polynomial.Algebraic & Geo- metric Topology, 2(1):337–370, 2002

Dror Bar-Natan. On Khovanov’s categorification of the Jones polynomial.Algebraic & Geo- metric Topology, 2(1):337–370, 2002

work page 2002
[5]

A functor-valued invariant of tangles.Algebraic & Geometric Topology, 2(2):665–741, 2002

Mikhail Khovanov. A functor-valued invariant of tangles.Algebraic & Geometric Topology, 2(2):665–741, 2002. 13 Tangle type Sign of crossings Poincar´ e polynomial 31 {−,−,−} x−3y−9 +x −2y−5 +y −3 +y −1 3′ 1 {+,+,+} x3y9 +x 2y5 +y 3 +y 32 {−,−,−} x−3y−9 +x −2y−7 +y −3 3′ 2 {+,+,+} x3y7 +x 2y5 +y 33 {−,−,−} {+,+,−} x−3y−9 +x −2y−8 +x −2y−7 +x −1y−6 +y −4 x2...

work page 2002
[2]

Khovanov’s homology for tangles and cobordisms.Geometry & Topology, 9(3):1443–1499, 2005

Dror Bar-Natan. Khovanov’s homology for tangles and cobordisms.Geometry & Topology, 9(3):1443–1499, 2005

work page 2005
[3]

Fixing the functoriality of Khovanov homol- ogy.Geometry & Topology, 13(3):1499–1582, 2009

David Clark, Scott Morrison, and Kevin Walker. Fixing the functoriality of Khovanov homol- ogy.Geometry & Topology, 13(3):1499–1582, 2009

work page 2009
[4]

A categorification of the Jones polynomial.Duke Math

Mikhail Khovanov. A categorification of the Jones polynomial.Duke Math. J., 104(1):359–426, 2000

work page 2000
[6]

Persistent Khovanov homology of tangles.arXiv preprint arXiv:2409.18312, 2024

Jian Liu, Li Shen, and Guo-Wei Wei. Persistent Khovanov homology of tangles.arXiv preprint arXiv:2409.18312, 2024

arXiv 2024
[7]

Skein lasagna modules and handle decompositions.Advances in Mathematics, 425:109071, 2023

Ciprian Manolescu, Kevin Walker, and Paul Wedrich. Skein lasagna modules and handle decompositions.Advances in Mathematics, 425:109071, 2023

work page 2023

Show all 10 references

[8]

Invariants of 4-manifolds from Khovanov- Rozansky link homology.Geom

Scott Morrison, Kevin Walker, and Paul Wedrich. Invariants of 4-manifolds from Khovanov- Rozansky link homology.Geom. Topol, 26(8):3367–3420, 2022

2022
[9]

Springer, 2024

J´ ozef Przytycki, Rhea Palak Bakshi, Dionne Ibarra, Gabriel Montoya-Vega, and Deborah Weeks.Lectures in Knot Theory: An Exploration of Contemporary Topics. Springer, 2024

2024

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [10]

Khovanov homology of tangles: algorithm and compu- tation.In preparation, 2025

Li Shen, Jian Liu, and Guo-Wei Wei. Khovanov homology of tangles: algorithm and compu- tation.In preparation, 2025. 15

work page 2025

[2] [1]

On Khovanov’s categorification of the Jones polynomial.Algebraic & Geo- metric Topology, 2(1):337–370, 2002

Dror Bar-Natan. On Khovanov’s categorification of the Jones polynomial.Algebraic & Geo- metric Topology, 2(1):337–370, 2002

work page 2002

[3] [5]

A functor-valued invariant of tangles.Algebraic & Geometric Topology, 2(2):665–741, 2002

Mikhail Khovanov. A functor-valued invariant of tangles.Algebraic & Geometric Topology, 2(2):665–741, 2002. 13 Tangle type Sign of crossings Poincar´ e polynomial 31 {−,−,−} x−3y−9 +x −2y−5 +y −3 +y −1 3′ 1 {+,+,+} x3y9 +x 2y5 +y 3 +y 32 {−,−,−} x−3y−9 +x −2y−7 +y −3 3′ 2 {+,+,+} x3y7 +x 2y5 +y 33 {−,−,−} {+,+,−} x−3y−9 +x −2y−8 +x −2y−7 +x −1y−6 +y −4 x2...

work page 2002

[4] [2]

Khovanov’s homology for tangles and cobordisms.Geometry & Topology, 9(3):1443–1499, 2005

Dror Bar-Natan. Khovanov’s homology for tangles and cobordisms.Geometry & Topology, 9(3):1443–1499, 2005

work page 2005

[5] [3]

Fixing the functoriality of Khovanov homol- ogy.Geometry & Topology, 13(3):1499–1582, 2009

David Clark, Scott Morrison, and Kevin Walker. Fixing the functoriality of Khovanov homol- ogy.Geometry & Topology, 13(3):1499–1582, 2009

work page 2009

[6] [4]

A categorification of the Jones polynomial.Duke Math

Mikhail Khovanov. A categorification of the Jones polynomial.Duke Math. J., 104(1):359–426, 2000

work page 2000

[7] [6]

Persistent Khovanov homology of tangles.arXiv preprint arXiv:2409.18312, 2024

Jian Liu, Li Shen, and Guo-Wei Wei. Persistent Khovanov homology of tangles.arXiv preprint arXiv:2409.18312, 2024

arXiv 2024

[8] [7]

Skein lasagna modules and handle decompositions.Advances in Mathematics, 425:109071, 2023

Ciprian Manolescu, Kevin Walker, and Paul Wedrich. Skein lasagna modules and handle decompositions.Advances in Mathematics, 425:109071, 2023

work page 2023