{"id":"f14819bc-0757-4847-9c90-54ff12f9f256","arxiv_id":"1908.06701","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For every nonnegative integer m, there are pairs of 2-knots and pairs of slice discs whose stabilization distance is exactly (or at least) m, including pairs indistinguishable by abelian Alexander invariants.","lead":"The paper proves that the number of 1-handle stabilizations needed to turn one embedded surface into another can be arbitrarily large, even for slice discs with the same boundary knot. It also shows a refined homology invariant can detect differences that the classical Alexander invariant cannot.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem C rests on the Proposition 7.8 splitting of twisted homology of the satellite and slice-disc exteriors; this splitting is asserted through a long Mayer-Vietoris diagram and is not independently verified.","rationale":"I read the paper as genuinely trying to prove that generalized stabilization distance is unbounded even when rational Alexander-module kernels coincide. The proof of Theorem C is a substantial and well-structured argument, and the individual pieces (Proposition 6.2, the generating-rank technology, the extension of characters over the cobordism in Claim 7.17) are mostly plausible. The weakest point is Proposition 7.8, which is the linchpin connecting the metabelian twisted homology of the satellite to the pattern and companion pieces. Its proof is a complex Mayer-Vietoris diagram with several assertions that are not fully expanded: the computation of H_1(T^2), the vanishing of j_J, and the surjectivity of the companion character. These are exactly the kind of places where a sign, a gluing identification, or a spectral sequence boundary argument can fail. The reader flagged the same general region but emphasized the diagrammatic computations of Examples 5.4 and 7.3; I agree those are risky too, but they are more localized and easier to check by direct calculation. The Proposition 7.8 splitting is more load-bearing because it is the only step that injects the companion's Alexander module A_ξ(J) into the lower bound; if that splitting is not correct, the final generating-rank comparison loses its foundation. I do not claim to have found a definite error, but the argument is not yet independently verified at this point. A direct computational check of the splitting for a concrete pattern-companion pair would settle whether the construction works. Until such a check is done, I would recommend conditional acceptance rather than unqualified acceptance, with the condition being a successful verification of Proposition 7.8 for at least one nontrivial companion.","tokens_in":111,"tokens_out":33916,"duration_ms":1022561,"concrete_test":"Compute both sides of Proposition 7.8 for one concrete satellite and check the claimed isomorphism and kernel splitting. Take R = 6_1, η as in Figure 8, n = 3, and χ a nontrivial character Z_9 → Z_3; choose J = 3_1 with a standard ribbon disc D, and J = unknot as a control. Using a Wirtinger presentation of X_{Rη(J)} and a handle decomposition of X_{∆_D}, compute H_1^{φχ∘f_*}(·; Z[ξ_3]) and the inclusion-induced map, and verify that it is isomorphic to H_1^{φχ}(X_R; Z[ξ_3]) ⊕ A_{ξ_3}(J)^{1⊕\\bar{1}} and that the kernel equals PR ⊕ P_J. A failure for any nontrivial J would invalidate the proof of Theorem C.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The lower bound in Theorem C (Theorem 7.15) depends entirely on Proposition 7.8's splitting H^{φχ∘f_*}_1(X_{Rη(J)}) ≅ H^{φχ}_1(X_R) ⊕ A_ξ(J)^{1⊕\\bar{1}}, together with the corresponding kernel splitting PR ⊕ P_J^{1⊕\\bar{1}}. The proof of Proposition 7.8 is a large Mayer-Vietoris diagram in which several identifications are asserted rather than demonstrated: for example, that H_1(T^2) is generated by the two deck translates of α and that j_J = 0 because [λ_J] = 0 in H_1(X_J^∞), and the surjectivity of the restricted character π_1(X_J) → Z_n. If any of these identifications is wrong, the generating-rank lower bound in Claim 7.18 (that P_2/(P_1∩P_2) has rank at least m) collapses, and Theorem C no longer follows as written. The reader's verdict identifies diagrammatic computations in Examples 5.4 and 7.3 as the main risk; I agree those matter, but the more load-bearing spot is the untested splitting in Proposition 7.8, since it is the only place where the companion's Alexander module enters the lower bound.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper defines the 1-handle stabilization distance between surfaces properly embedded in a fixed 4-manifold, together with a generalized distance that also allows connected sum with arbitrary 2-knots at zero cost. It proves three main results: Theorem A, that for every m there are embedded 2-spheres in S^4 at 1-handle stabilization distance exactly m (with an explicit added note that this was first proved by Miyazaki); Theorem B, that for every m there is a knot with two slice discs whose generalized stabilization distance is exactly m, detected by comparing kernels of inclusion-induced maps on rational Alexander modules; and Theorem C, that for every m there is a knot with two slice discs whose rational Alexander-module kernels coincide while the generalized stabilization distance is at least m, detected by metabelian twisted homology with coefficients in the Eisenstein integers. The proofs are built on a common cobordism construction between surface exteriors, generating-rank inequalities for modules over PIDs, explicit computations for the knots 9_46 and 6_1, and a satellite construction with compatible degree-one maps. The algebraic arguments are stated in detail, and the paper is careful to identify which parts depend on previously published results.","tokens_in":31563,"tokens_out":18352,"duration_ms":184632,"significance":"If the results hold, the paper establishes that the generalized stabilization distance between slice discs is unbounded even when the first layer of abelian invariants, namely the kernels of rational Alexander-module maps, coincides. This is a genuine step beyond the Alexander-module method: it shows that metabelian twisted homology can distinguish choices of slice discs that abelian invariants cannot. The examples are explicit and checkable: the knots 9_46 and 6_1 are analyzed via Seifert matrices and handle decompositions, and the claimed distances are derived rather than fitted. The paper also gives credit where credit is due by acknowledging that Theorem A was previously proved by Miyazaki and by framing Theorem A as a pedagogical contrast. The lower-bound strategy via generating rank over PIDs is clean, and the use of a Mayer-Vietoris splitting in Proposition 7.8 is coherent. I do not find the stress-test concern about Proposition 7.8 to be a demonstrated gap: the proof, though compressed, supplies the relevant Mayer-Vietoris diagram and the key identifications are standard consequences of the zero-winding satellite setup.","major_comments":[],"minor_comments":[{"comment":"The computation of H_1(T^2) as (Z[ξ]/(ξ−1))^{1⊕\\bar{1}} and the statement that j_J = 0 because [λ_J] = 0 in H_1(X_J^∞) are highly compressed; a short explicit description of the twisted chain complex of T^2 and of why each deck translate of λ_J bounds a lift of a Seifert surface would make the proof substantially easier to verify.","section":"Section 7.2, proof of Proposition 7.8"},{"comment":"The linear-algebra fact imported from [KL05, Theorem 6.1] is not stated in the paper; please state the precise lemma, or quote the theorem, and indicate explicitly how the condition that at least N−m of the χ_i are nonzero follows from the cited argument.","section":"Section 7.3, Claim 7.17"},{"comment":"The verification that the banded diagram on the right of Figure 6 represents the standard unknotted torus is carried out visually; a short sentence describing the final cancellation would help. Since the added note already attributes Theorem A to Miyazaki, this part is not essential for novelty, but the exposition would be clearer with one more sentence.","section":"Section 5, proof of Proposition 5.1"},{"comment":"The roles of η, γ, and the genus-one Seifert surface F should be stated explicitly in the caption or text; currently the reader must infer that η is the curve used in Proposition 7.8 and that η generates A(R) as required.","section":"Example 7.3 and Figure 8"},{"comment":"The notation M^{1⊕\\bar{1}} is defined just before Proposition 7.8 and is then used immediately; a one-line reminder of this notation in the statement of Proposition 7.8 would avoid possible confusion.","section":"Notation 7.6 and Proposition 7.8"},{"comment":"The paper should mention at the start of the introduction that Theorem A is due to Miyazaki, rather than only in an added note, so that readers are not misled about the novelty of the first theorem.","section":"Section 1, Added in proof"}],"recommendation":"minor_revision","confidential_remarks":"I recommend minor revision. The paper is suitable for the journal. I weighed the stress-test concern about Proposition 7.8 carefully; in my reading it does not amount to a definite error, but the proof would benefit from the clarifications listed in the minor comments."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nI basically agree with the reader's verdict: accept, soundness around 8, with the genuinely new content being Theorem B and especially Theorem C. The real result to care about is Theorem C: pairs of slice discs for the same knot with generalized stabilization distance at least m, even when the kernels of the inclusion maps on rational Alexander modules coincide. That shows metabelian twisted homology sees things the abelian theory cannot, and it gives real traction on the slice-disc enumeration problem. The authors also disclose in an 'Added in proof' that Theorem A is due to Miyazaki; keeping it as exposition is fine.\n\nWhat I like: the lower bounds come from honest invariants, generating ranks of Alexander modules and twisted homology, and the upper bounds from explicit handle additions. There is no parameter fitting anywhere. The proofs of B and C are detailed and mostly checkable. The paper is well organized and honest about what it does not know, e.g., the exact value of d2 in Theorem C is left open.\n\nSoft spots, in proportion. Proposition 7.8 is the load-bearing wall for Theorem C. The stress-test note worries the splitting there is asserted rather than proved. Having read it, I would put it differently: it is a long Mayer-Vietoris diagram chase, and the key identifications are argued, not assumed. The [lambda_J] = 0 step is a standard fact about infinite cyclic covers, and the surjectivity of the restricted character follows from eta generating A(R). It is intricate enough that a referee should check the chase line by line, but it is not a hand-wave. Minor spots: the computations in Examples 5.4 and 7.3 lean on [CP19] for details, though the Seifert matrix data is given; Claim 7.17 uses a linear algebra existence result from [KL05, Thm 6.1] without proof; and Theorem A's upper bound is a fiddly banded-diagram manipulation, but it is illustrated and followable. The citation pattern is healthy.\n\nWho it is for: anyone working on slice discs, 2-knots, or stabilization questions in 4-manifolds. The paper deserves a serious referee. I would send it to review, asking the referee to check Proposition 7.8 carefully and to ask for a bit more detail on the [KL05] input. Expect minor-to-moderate revisions, not a rewrite.","headline":"Worth refereeing: the genuinely new content is Theorem C, and the delicate spot is the Proposition 7.8 splitting, which is argued in detail but should be checked line by line.","tokens_in":32113,"tokens_out":12429,"would_cite":true,"duration_ms":117889,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57N13","57N65"],"pacs":[],"model":"deepseek-v4-flash","headline":"Slice discs can be arbitrarily far apart in stabilization distance.","keywords":["1-handle stabilization distance","2-knots","slice discs","Alexander module","twisted homology","metabelian representations","cyclic covers","generating rank"],"falsifier":"Recompute $A_{\\mathbb{Q}}$ for the knot $9_{46}$ and its left/right band slice discs directly from the diagrams in Figure 3: if $A_{\\mathbb{Q}}(9_{46})$ is not $\\mathbb{Q}[t^{\\pm 1}]/(2t-1) \\oplus \\mathbb{Q}[t^{\\pm 1}]/(t-2)$, or if the inclusion maps do not project onto the two displayed summands with kernels $P_1 = \\mathbb{Q}[t^{\\pm 1}]/(t-2)$ and $P_2 = \\mathbb{Q}[t^{\\pm 1}]/(2t-1)$, then the distance claim of Theorem B fails; alternatively, exhibit an explicit sequence of fewer than $n$ stabilizations and 2-knot sums relating the two canonical slice discs of $\\#^n 9_{46}$, which the theorem predicts does not exist.","tokens_in":31104,"feed_emoji":"🪢","tokens_out":9323,"duration_ms":81998,"temperature":0.7,"pith_summary":"The paper defines the 1-handle stabilization distance between two surfaces in a 4-manifold and proves that it can take any prescribed value. It first shows that for every nonnegative integer $m$ there are 2-knots in $S^4$ at distance exactly $m$ from the unknot. It then proves that a generalized distance, which ignores connected sums with arbitrary 2-knots, is also unbounded for slice discs of a fixed knot in $S^3$. The main mathematical claim is that even when abelian invariants computed from cyclic covers—namely the kernels of the rational Alexander module maps—coincide, the generalized stabilization distance can still be arbitrarily large, with metabelian twisted homology providing the extra distinguishing power.","feed_headline":"Slice discs can be arbitrarily far apart in stabilization distance","feed_subtitle":"For each m a knot has two slice discs exactly m stabilizations apart; metabelian homology catches what abelian invariants miss.","key_machinery":"The load-bearing object is the rational Alexander module $A_{\\mathbb{Q}}(L) = H_1(X_L; \\mathbb{Q}[t^{\\pm 1}])$ of a knot or slice-disc exterior, together with the kernel of the inclusion-induced map $A_{\\mathbb{Q}}(J) \\to A_{\\mathbb{Q}}(D_i)$ for a slice disc $D_i$. A single 1-handle stabilization fits into a short exact sequence $0 \\to \\mathbb{Q}[t^{\\pm 1}]/(p) \\to A_{\\mathbb{Q}}(F_1) \\to A_{\\mathbb{Q}}(F_2) \\to 0$, so the generating rank drops by at most one; comparing the generating ranks of the two kernels and their intersection gives the lower bound in Proposition 6.3. For the second-order distinction, the machinery is metabelian twisted homology with representations $\\varphi_\\chi \\colon \\pi_1(X_K) \\to \\mathbb{Z}/2 \\ltimes \\mathbb{Z}_n$ factoring through the 2-fold branched cover, with $\\mathbb{Z}[\\xi_n]$ coefficients, and a splitting theorem (Proposition 7.8) that decomposes the kernel for a satellite slice disc into a piece from the pattern knot plus a conjugate pair of pieces from the companion. The companion is chosen as $J_0 \\# -J_0$, whose two standard ribbon discs have equal Alexander-module kernels but different behaviour after tensoring with $\\mathbb{Z}[\\xi_3]$.","core_discovery":"The paper claims that the 1-handle stabilization distance $d_1(F,F')$—the minimum number of 1-handle stabilizations needed to make two homologous surfaces ambiently isotopic—is unbounded and can be prescribed exactly, in the simplest setting of 2-spheres in $S^4$. For the coarser generalized distance $d_2$, which also allows connected sum with arbitrary 2-knots at zero cost, it claims that for every $m$ there is a knot $J$ in $S^3$ with two slice discs in $D^4$ whose generalized stabilization distance is exactly $m$. It further claims that such pairs exist with the kernels of the inclusion-induced maps on rational Alexander modules equal, so that all abelian cyclic-cover invariants agree; the separation is detected by metabelian twisted homology, specifically representations to $\\mathbb{Z}/2 \\ltimes \\mathbb{Z}_3$ with coefficients in the Eisenstein integers.","pith_inferences":["The generating-rank inequality in Proposition 6.3 is a general machine: any pair of slice discs whose Alexander-module kernels are complementary free summands should yield distance equal to the common rank, so many knots besides $9_{46}$ with suitable Seifert pairings should give similar examples.","The paper only bounds the distance in Theorem C below by $g$ and above by $4g$; determining the exact distance for these satellite examples would likely require a twisted analogue of the precise upper-bound construction used in Theorem B.","A natural next test is whether the same metabelian technique distinguishes slice discs whose metabelian invariants coincide, or whether still higher-order nilpotent twisted homology is needed; the kernel-splitting pattern suggests an entire hierarchy of slice-disc invariants indexed by solvable quotients.","Because Theorem C's examples have equal rational Alexander kernels, any invariant computed from cyclic covers—orders, torsion, or kernels—cannot certify the lower bound; readers should expect other slice-disc pairs that are abelian-indistinguishable but metabelian-distinguishable, possibly including spun versus non-spun ribbon discs for $K \\# -K$."],"forward_implications":["For every $m$ there are 2-knots in $S^4$ that require exactly $m$ stabilizations to become unknotted, so the metric $d_1$ is nontrivial even for null-homologous 2-spheres.","For every $m$ there is a slice knot with two slice discs whose generalized stabilization distance is exactly $m$, so the number of slice-disc classes up to 2-knot connected sum is unbounded; for example $\\#^k 9_{46}$ has at least $2^k$ such classes.","Abelian invariants—the order and kernel of the rational Alexander module map—do not classify slice-disc pairs up to stabilization, because Theorem C exhibits pairs with equal kernels and arbitrarily large generalized distance.","Metabelian twisted homology distinguishes slice discs obtained from the same fixed metabolising link on a Seifert surface by different choices of bounding discs, detecting a genuinely second-order slicing phenomenon.","The distance $d_1$ is a metric on ambient isotopy classes of fixed-genus surfaces representing a fixed homology class, and the proof of the triangle inequality rearranges stabilizations before destabilizations."],"supporting_citations":[{"why":"Establishes that any two homologous surfaces become isotopic after finitely many 1-handle stabilizations, making the distances well-defined.","marker":"[BS15]"},{"why":"Contains the detailed Alexander-module computations for the $9_{46}$ left and right band slice discs used in Theorem B.","marker":"[CP19]"},{"why":"Proves the 2-knot stabilization-distance statement of Theorem A by a similar Alexander-module method.","marker":"[Miy86]"},{"why":"Introduces the metabelian dihedral representations whose twisted homology is the main tool for Theorem C.","marker":"[CG78]"},{"why":"Supplies the linear-algebra existence argument, cited in Claim 7.17, that produces the required character with many nonzero coordinates.","marker":"[KL05]"},{"why":"Provides the banded-knot-diagram moves used to show the upper bound in Theorem A, namely that the stabilized surface is unknotted.","marker":"[Swe01]"},{"why":"Defines an alternative stabilization distance that the paper contrasts with $d_1$ and $d_2$.","marker":"[JZ18b]"}],"fun_headline_variants":["Stabilization distance hits every integer exactly","Slice discs at any prescribed stabilization distance","Metabelian invariants reveal what abelian ones cannot","Exact stabilization distances for surfaces in 4-manifolds","Unbounded stabilization distance with exact values"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The lower-bound proofs depend on explicit diagrammatic computations: for the knot $9_{46}$ in Figure 3, that its rational Alexander module splits as $\\mathbb{Q}[t^{\\pm 1}]/(2t-1) \\oplus \\mathbb{Q}[t^{\\pm 1}]/(t-2)$ with the two band discs projecting onto different summands, and for the knot $6_1$ with infection curve $\\eta$ in Figure 8, that its Alexander module, kernel, and the splitting in Proposition 7.8 are as computed; if any of these module computations or kernel splittings is wrong, the claimed distances collapse, and Theorem C additionally relies on a linear-algebra existence result quoted without proof in Claim 7.17.","fun_headline_variants_meta":{"raw":{"variants":["Stabilization distance hits every integer exactly","Slice discs at any prescribed stabilization distance","Metabelian invariants reveal what abelian ones cannot","Exact stabilization distances for surfaces in 4-manifolds","Unbounded stabilization distance with exact values"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000231,"raw_usage":{"total_tokens":1475,"prompt_tokens":922,"completion_tokens":553,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":538,"completion_tokens_details":{"reasoning_tokens":481}},"tokens_in":538,"tokens_out":553,"duration_ms":6095,"temperature":1.0,"reasoning_tokens":481,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:36:55.691218+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute $A_{\\mathbb{Q}}$ for the knot $9_{46}$ and its left/right band slice discs directly from the diagrams in Figure 3: if $A_{\\mathbb{Q}}(9_{46})$ is not $\\mathbb{Q}[t^{\\pm 1}]/(2t-1) \\oplus \\mathbb{Q}[t^{\\pm 1}]/(t-2)$, or if the inclusion maps do not project onto the two displayed summands with kernels $P_1 = \\mathbb{Q}[t^{\\pm 1}]/(t-2)$ and $P_2 = \\mathbb{Q}[t^{\\pm 1}]/(2t-1)$, then the distance claim of Theorem B fails; alternatively, exhibit an explicit sequence of fewer than $n$ stabilizations and 2-knot sums relating the two canonical slice discs of $\\#^n 9_{46}$, which the theorem predicts does not exist.","supporting_citations":[],"review_version":1}