{"id":"42381b11-ce3f-4841-8d8e-1d3b99e11476","arxiv_id":"2507.15305","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":2.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Lecture notes presenting the cobordism maps on Khovanov and link Floer homology as invariants of knotted surfaces, with worked examples and exercises.","lead":"These lecture notes explain how link homology theories, such as Khovanov homology and link Floer homology, assign maps to the surfaces that knots bound in 4-dimensional space. A smart generalist might read them to see how these algebraic invariants distinguish knotted surfaces, including surfaces that are topologically but not smoothly the same.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 3.7's decisive claim H·δBN ≠ 0 rests on an unshipped KnotJob spectral-sequence computation; the notes do not identify δBN in Table 3 or justify the 'survive to second page' step.","rationale":"The reader's verdict (CONDITIONAL, low correctness risk) is fair for a lecture-note survey. The quoted functoriality theorems are settled results; a survey is entitled to use them as black boxes. I therefore do not treat the analytic foundations of Heegaard Floer theory as a load-bearing weakness of this manuscript: if those foundations failed, the cited literature would collapse, but the notes add no new dependence beyond citation. The one place where the notes make a self-contained claim with a proof that must stand on its own is Proposition 3.7, and there the decisive assertion H·δBN ≠ 0 is supported only by a spectral-sequence computation whose input (the diagram of K) is a hand-drawn figure and whose output (Table 3) lists E1/E2 ranks without identifying δBN or the third-page differentials. The logical step 'survive to the second page' → 'H·δBN ≠ 0' is also under-justified in the text. This does not undermine the survey body, but it does mean the notes' advertised demonstration of stabilization-distance exotica is not independently verifiable. That matches the reader's secondary concern and supports the CONDITIONAL verdict. If the KnotJob computation is shipped or the proposition is recast as a quotation of [Hay23, §5.2], the condition is satisfied. I credit the notes for being transparent that Proposition 3.7 is based on [Hay23, §5.2] and a proof sketch, and for giving the reader exercises to verify the Khovanov-level computations (Figures 24–26), which are checkable by hand.","tokens_in":25878,"tokens_out":8441,"duration_ms":86446,"concrete_test":"Encode the knot K from Figure 32 (e.g., as a braid or PD code) and the two slice disks D,D′ as movies; run KnotJob [Sch23] or an independent implementation to compute the Bar-Natan–Lee–Turner spectral sequence and the F2[H]-module structure of BN(K). Check specifically: (i) whether the spectral sequence collapses on page 3 as claimed; (ii) whether every class in BN^{0,1}(K), in particular δBN, lies in a free F2[H] summand so that H·δBN ≠ 0; and (iii) whether the cobordism maps BN(D) and BN(D′) differ after one internal stabilization. Alternately, extract the explicit chain-level representatives from Figure 32 and verify H·δBN ≠ 0 by hand in the finite-dimensional Bar-Natan complex, avoiding the spectral sequence. If the computation cannot be reproduced, Proposition 3.7 should be downgraded to a quotation of [Hay23, §5.2] with proof omitted.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The internally load-bearing claim is Proposition 3.7: the Bar-Natan cobordism maps of the slice disks D and τ(D) from Figure 32 differ even after one internal stabilization. The proof sketch reduces this to the assertion H·δBN ≠ 0 for δBN = BN(D)(1) − BN(D′)(1) ∈ BN^{0,1}(K). This is the decisive step, and it is not established by the notes. The only justification is §3.1's sentence: 'one can use the program KnotJob [Sch23] to compute the Bar-Natan–Lee–Turner spectral sequence ... and then determine the F2[H]-module structure of BN(K). This spectral sequence collapses on its third page, and the relevant portions of the first two pages are shown in Table 3. In particular, all elements in bigrading (0,1) survive to the second page, which implies that H·δBN ≠ 0.' The displayed Table 3 gives E1 and E2 bigraded ranks only; it does not identify δBN as one of the surviving classes, and 'survive to the second page' is not by itself equivalent to 'H·δBN ≠ 0' unless the third-page differentials are zero and the E2 class maps injectively to E∞. The KnotJob input (the diagram of K, which is only drawn as a picture) and the program output are not provided, so the computation cannot be reproduced or checked from the notes. Since Proposition 3.7 is the notes' own self-contained demonstration of the stabilization-distance phenomenon associated with Theorem 1.15, the central didactic claim rests at this point on unshipped computation. The other surveyed applications (Theorems 1.8–1.14) are quoted from refereed literature and are not affected by this gap; my concern is therefore specific and localized.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"These lecture notes, from the 2024 Georgia Topology Summer School, survey the use of link homology theories — principally Khovanov homology and link Floer homology — as functors from the category of links and link cobordisms in S^3×I to modules, and their applications to knotted surfaces in the 4-ball. The first lecture states the functoriality theorems (Theorems 1.2 and 1.4) and surveys applications: distinguishing slice disks, detecting exotically knotted surfaces, resolving Livingston's Seifert-surface question, and lower bounds on stabilization distance. The second lecture develops Khovanov homology from the cube of resolutions and the Frobenius algebra A = R⟨1,x⟩, discusses the cobordism maps with explicit Reidemeister I and II tables, and includes many worked examples, culminating in a detailed calculation distinguishing two slice disks for the knot 9₄₆. The third lecture introduces Bar-Natan homology over F2[H] and Bar-Natan's formal-complex category, sketches the proof of Reidemeister invariance via local relations, and proves (Proposition 3.7) that a pair of slice disks for a strongly invertible knot remain distinct after one internal stabilization, using a spectral-sequence computation attributed to the program KnotJob. The notes close with the ribbon-concordance application.","tokens_in":26084,"tokens_out":25889,"duration_ms":246377,"significance":"As an expository contribution, these notes are valuable: the survey portions are accurate and carefully attributed, the computations in Figures 21, 23, 25 and Tables 1–2 are explicit enough to serve as worked examples, and the exercises (1.16–1.19, 2.12–2.19, 3.3–3.27) genuinely guide the reader through substantial results from the literature. The notes are unusually candid about provenance, flagging where results are quoted from refereed sources, where proofs are sketched, and where computations come from the computer program KnotJob. The load-bearing claims of the survey — functoriality of Khovanov and link Floer cobordism maps and the applications that follow — rest on cited literature and are reported faithfully. The main caveat is that the notes' own self-contained demonstration (Proposition 3.7) depends on a spectral-sequence step that is not justified in the text and on a computation that is not shipped; this is the one point that needs substantive work before the notes can serve as a fully reliable self-study reference.","major_comments":[{"comment":"The decisive claim H·δBN ≠ 0 is not established by the argument as written. The text states that 'all elements in bigrading (0,1) survive to the second page, which implies that H·δBN ≠ 0,' but survival to the E2 page does not by itself imply the nonvanishing of H·δBN: one must also know that the E∞-class corresponding to δBN is a generator of the free F2[H]-tower summand in BN^{0,1}(K) rather than a torsion class, that the spectral sequence collapses at E3 (the collapse is asserted but the d2 differential is not shown in Table 3), and that the F2[H]-module structure of BN(K) is exactly two free towers plus torsion. None of these identifications is supplied, and δBN itself is never identified among the classes counted in Table 3. Because this is the load-bearing step for Proposition 3.7 — the notes' own demonstration of the stabilization-distance phenomenon — the proof should either be completed (defining the Bar-Natan–Lee–Turner spectral sequence, stating how its E∞ term relates to the associated graded of the H-filtration, and proving the free-tower lemma) or the notes should explicitly defer the verification to [Hay23, §5.2] rather than presenting the implication as established.","section":"§3.1, Proposition 3.7 (proof of H·δBN ≠ 0)"},{"comment":"The computation behind Table 3 is not reproducible from the notes: the KnotJob input (a machine-readable diagram of the knot K of Figure 32) and the program output are not provided, Figure 32 shows K only as a diagram, and Table 3 gives only the bigraded ranks of part of the first two pages. Consequently the assertions that 'this spectral sequence collapses on its third page' and that 'all elements in bigrading (0,1) survive to the second page' cannot be checked by a reader. Since Proposition 3.7 is advertised as illustrating the power of Bar-Natan homology, the notes should either supply the data as ancillary files or state clearly that the example is an illustration whose computational verification is deferred to the published source [Hay23, §5.2].","section":"§3.1, Table 3 and the KnotJob computation"}],"minor_comments":[{"comment":"The two displayed expressions defining q(α) are mutually inconsistent when n− ≠ 0: 'v+(α) − v−(α) + h(α) + n+ − n−' is not equal to 'deg(α) + h(α) + n+ + n−'; the intended quantum-grading convention should be stated correctly.","section":"§2.1, definition of the quantum grading"},{"comment":"Example 1.5 uses U for the Bar-Natan variable, while the rest of the notes (e.g., §3.1) fixes R = F2[H]; the notes should adopt a single notation for this variable or explicitly note the change of convention.","section":"Example 1.5"},{"comment":"The sentence 'efficient software exists to compute the hat-flavored knot Floer homology [HFK of knots with > 100 crossings' contains an unclosed bracket, presumably a typo for '[HFK]'; the citation formatting should be fixed.","section":"§1.5, machine computation"},{"comment":"The caption does not state what the entries of Table 3 are; it should say that the entries are bigraded ranks over F2 and should identify which rows and columns are omitted and why those ranges suffice for the argument.","section":"Table 3, caption"},{"comment":"In the proof sketch, the verification that φ is mapped to 1 by Kh(−D) and to 0 by Kh(−D′) is left to the reader as an exercise in the middle of a proof; a brief indication of the movie (for example, which band moves occur) would make the sketch practicable.","section":"§3.1, proof of Proposition 3.7"},{"comment":"The citation key [Ozs06] is used in the text and in the reference list for a paper whose authors are Ozsváth and Szabó; the key should be [OS06] (or similar) for consistency with the surrounding OS citations.","section":"References, [Ozs06]"}],"recommendation":"major_revision","confidential_remarks":"The heavy self-citation in Section 1 (Theorems 1.11, 1.13, 1.15) is justified by the author's role as a principal contributor to those results and is disclosed transparently in the text; I see no novelty-disclosure concern. The stress-test concern about Proposition 3.7 does land: the notes' own central demonstration of the stabilization-distance phenomenon rests on a spectral-sequence step that is not justified as written and on an unshipped computation. My recommendation of major revision is driven by that gap alone; the rest of the manuscript is in good shape and the survey material is reliable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Kyle Hayden's notes are exactly what lecture notes should be: a coherent, honest tour of link homology cobordism maps and their use on knotted surfaces, aimed at graduate students. There are no new theorems, and the paper does not claim any. What is new is the selection and arrangement: the survey of applications (roll-spun disks, rim surgery, cables, stabilization distance) plus worked Khovanov and Bar-Natan computations, exercises, and a readable account of Bar-Natan's TQFT. I checked several grading calculations and the Reidemeister tables; they are internally consistent. The notes are unusually candid about provenance, with Section 1.1 warning that the overview is not mathematically precise and surveyed theorems attributed to refereed sources. The heavy self-citation is a natural consequence of the author being a main contributor to these applications; I do not see circularity.\n\nThe one real soft spot is Proposition 3.7, and it is localized. This is the notes' own demonstration that Bar-Natan maps can distinguish slice disks even after one internal stabilization. The proof sketch reduces the claim to delta_Kh being nonzero, which is fine, and then to H times delta_BN being nonzero. The second step rests entirely on a KnotJob computation of the Bar-Natan-Lee-Turner spectral sequence. Table 3 shows only bigraded ranks of pages 1 and 2; it does not identify delta_BN among the surviving classes, and the text's 'survive to the second page' does not by itself imply H times delta_BN is nonzero unless third-page differentials vanish and the E2 class maps injectively to E-infinity. The KnotJob input is just a picture of the knot, and no program output is shipped. So the decisive claim is not reproducible from the manuscript. This is a minor flaw in a survey rather than a load-bearing failure of the whole notes, but it should be fixed: either include the KnotJob data, or state Proposition 3.7 as a quotation from Hay23 with proof omitted.\n\nThe Heegaard Floer foundations are treated as a black box, but that is appropriate for lecture notes and external to the claims made here. I agree with the reader's conditional verdict. The paper deserves a serious referee; the referee should ask for the KnotJob reproducibility, and the rest of the notes can be checked against standard sources. I would bring this to a reading group for the survey sections, and I would cite it as an entry point to the area.","headline":"A useful, honest set of lecture notes on link homology cobordism maps and knotted surfaces, with one localized reproducibility gap in the self-contained Bar-Natan example.","tokens_in":26851,"tokens_out":1959,"would_cite":true,"duration_ms":22454,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57K18","57K10"],"pacs":[],"model":"deepseek-v4-flash","headline":"The notes argue that link homology theories, via the maps they assign to link cobordisms, are effective invariants that distinguish knotted surfaces in 4-space.","keywords":["link homology","Khovanov homology","link Floer homology","cobordism maps","knotted surfaces","slice disks","exotic surfaces","Bar-Natan homology"],"falsifier":"A direct chain-level calculation of the two Khovanov cobordism maps for the $9_{46}$ slice-disk pair depicted in the notes would settle the distinguishing claim: the notes state one map sends the class $\\phi$ to $1$ and the other to $0$, so equal images would refute it. Independently, an explicit isotopy rel boundary between two movies of the same link cobordism whose induced Khovanov maps differ by more than a sign would falsify the central invariance theorem.","tokens_in":25469,"feed_emoji":"🪢","tokens_out":13466,"duration_ms":135495,"temperature":0.7,"pith_summary":"These lecture notes aim to install one working picture: link homology theories are functors from the category of oriented links and link cobordisms to modules, so a surface $\\Sigma$ in $S^3 \\times I$ with boundary links $L_0$ and $L_1$ induces a map between the homology groups of those links. For Khovanov homology, the induced map is bigraded of degree $(0, \\chi(\\Sigma))$, well-defined up to sign, and invariant under isotopy rel boundary; for link Floer homology, decorated cobordisms induce well-defined invariant maps. The payoff is 4-dimensional: these maps distinguish slice disks that share a boundary knot, detect exotically knotted surfaces, and answer a long-standing question by showing that equal-genus Seifert surfaces for a knot need not be isotopic in the 4-ball. The notes then make the machinery hands-on, computing Khovanov cobordism maps through the cube of resolutions and the Bar-Natan category, so the reader can reproduce the applications by hand or with software.","feed_headline":"Cobordism maps turn link homology into a 4D surface detector","feed_subtitle":"Khovanov and link Floer maps distinguish slice disks, exotic surfaces, and Seifert surfaces up to isotopy.","key_machinery":"The load-bearing object is the cobordism map. For Khovanov homology it is assembled from the cube of resolutions: each link diagram gives $2^n$ smoothings, each circle is assigned the rank-two Frobenius algebra $A=R\\langle 1,x\\rangle$, and the elementary cobordisms---birth, merging saddle, splitting saddle, death, and Reidemeister moves---are converted into the maps $\\iota$, $m$, $\\Delta$, $\\epsilon$, and the local cobordisms encoded in the notes' tables. For link Floer homology, the analogous maps come from counting holomorphic Whitney disks in the symmetric product $\\mathrm{Sym}^g(\\Sigma)$ that avoid, or intersect with multiplicity, the basepoint divisors. The Bar-Natan category, the category of formal complexes of planar tangles modulo the sphere relation, the torus relation, and the 4-tube relation, is what makes the Khovanov complex and its cobordism maps invariant: invariance under Reidemeister moves is proved by chain homotopy equivalences whose chain homotopies use exactly those local relations.","core_discovery":"The central claim is functoriality. A link cobordism $\\Sigma \\subset S^3 \\times I$ between links $L_0$ and $L_1$ induces a map $\\mathrm{Kh}(\\Sigma): \\mathrm{Kh}(L_0) \\to \\mathrm{Kh}(L_1)$ that is bigraded of degree $(0,\\chi(\\Sigma))$, well-defined up to sign, and invariant under isotopy of $\\Sigma$ rel boundary; the corresponding statement for link Floer homology is that every decorated link cobordism induces a well-defined, isotopy-invariant map $\\mathrm{HFL}(\\Sigma)$. The notes use these maps as the organizing tool for a survey of knotted surfaces: they distinguish pairs of slice disks with the same boundary, detect exotically knotted surfaces in $B^4$, show that Seifert surfaces of equal genus need not be smoothly isotopic in the 4-ball, and bound stabilization distance. The second half of the notes builds the Khovanov side from scratch---cube of resolutions, the Frobenius algebra $A = R\\langle 1,x\\rangle$, merge/split/birth/death maps, and the Bar-Natan category of formal complexes modulo local relations---so that the cobordism maps can be computed explicitly.","pith_inferences":["Beyond the notes, the explicit chain-level recipes for Reidemeister and saddle maps suggest that distinguishing slice disks by Khovanov maps can be automated: one could compute the images of the two $9_{46}$ disks over $\\mathbb{Z}$ with a program and turn the existence proof into a routine check.","Beyond the notes, the invariance proof through the Bar-Natan category indicates that any Frobenius algebra satisfying the same local relations yields a functorial link homology theory, so new theories built this way would immediately come with cobordism maps.","Beyond the notes, the link Floer stabilization-distance bounds leave open whether Khovanov or Bar-Natan maps can also force arbitrarily large stabilization distance, a question the notes do not answer.","Beyond the notes, the mostly zero or formulaic behavior on closed surfaces suggests that the predictive power of these functors is concentrated in surfaces with boundary; designing a closed-surface invariant would require adding structure beyond the maps surveyed here."],"forward_implications":["Slice disks with a common boundary knot that are non-isotopic rel boundary are distinguished by their Khovanov and link Floer cobordism maps (Theorems 1.8 and 1.9).","Exotically knotted surfaces in $B^4$ are detected: infinitely many knots bound infinitely many genus-one surfaces that are pairwise topologically isotopic but not diffeomorphic, and Khovanov homology detects such pairs in every genus (Theorems 1.10 and 1.11).","Equal-genus Seifert surfaces for a fixed knot need not be smoothly isotopic through surfaces in $B^4$, settling a long-standing question in the negative (Theorem 1.13).","The stabilization distance between exotically knotted slice disks can be made arbitrarily large, and Bar-Natan homology can distinguish surfaces even after one internal stabilization (Theorems 1.14 and 1.15).","Because the maps compose functorially, composing a distinguishing pair of cobordisms with a ribbon concordance preserves the distinction, so the detection propagates to new surfaces (Exercise 3.27 and the remarks around it)."],"supporting_citations":[{"why":"Introduces Khovanov homology and the Frobenius algebra $A=R\\langle 1,x\\rangle$ that the survey's chain-level methods use.","marker":"[Kho00]"},{"why":"Proves the Khovanov cobordism maps are invariant up to sign for surfaces in $\\mathbb{R}^3\\times I$.","marker":"[Jac04]"},{"why":"Extends that invariance to $S^3\\times I$ through the sweep-around move, the setting of the notes' main functoriality theorem.","marker":"[MWW22]"},{"why":"Builds the Bar-Natan category of formal complexes modulo sphere, torus, and 4-tube relations and gives an alternative invariance proof.","marker":"[BN05]"},{"why":"Introduces link Floer homology from doubly-pointed Heegaard diagrams, the theory to which Theorem 1.4 applies.","marker":"[OS04a]"},{"why":"Establishes the decorated link Floer cobordism maps and the graded flavors of the theory used in the survey.","marker":"[Zem19b]"},{"why":"Supplies the explicit Khovanov computation distinguishing the $9_{46}$ slice-disk pair that the notes reproduce.","marker":"[SS23]"},{"why":"Resolves the Seifert surface isotopy question using Khovanov cobordism maps, the content of Theorem 1.13.","marker":"[HKM+22]"},{"why":"Gives arbitrarily large stabilization distance lower bounds via link Floer cobordism maps, the content of Theorem 1.14.","marker":"[Gut24]"},{"why":"Produces exotically knotted surface pairs detected by Khovanov maps in all genera and informs the Reidemeister-move tables.","marker":"[HS24]"}],"fun_headline_variants":["Cobordism maps make link homology probe 4D surfaces","Link cobordism maps reveal 4D surface structure","Cobordism maps turn homology into 4D surface invariants","Khovanov cobordism maps track knotted surfaces in 4D","Functorial link homology detects exotic 4D surfaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole link Floer half of the survey assumes, as a black box, that the holomorphic-curve counts defining the homology are well-defined and independent of the auxiliary choices; the notes cite rather than prove this analytic foundation.","fun_headline_variants_meta":{"raw":{"variants":["Cobordism maps make link homology probe 4D surfaces","Link cobordism maps reveal 4D surface structure","Cobordism maps turn homology into 4D surface invariants","Khovanov cobordism maps track knotted surfaces in 4D","Functorial link homology detects exotic 4D surfaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000399,"raw_usage":{"total_tokens":2084,"prompt_tokens":942,"completion_tokens":1142,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":558,"completion_tokens_details":{"reasoning_tokens":1066}},"tokens_in":558,"tokens_out":1142,"duration_ms":8657,"temperature":1.0,"reasoning_tokens":1066,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:36:26.202119+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct chain-level calculation of the two Khovanov cobordism maps for the $9_{46}$ slice-disk pair depicted in the notes would settle the distinguishing claim: the notes state one map sends the class $\\phi$ to $1$ and the other to $0$, so equal images would refute it. Independently, an explicit isotopy rel boundary between two movies of the same link cobordism whose induced Khovanov maps differ by more than a sign would falsify the central invariance theorem.","supporting_citations":[],"review_version":1}