{"id":"a6186956-459f-475c-8a76-97478e082412","arxiv_id":"1909.01003","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The topological 4-genus of every 3-strand torus knot equals half its maximal signature and its untwisting number.","lead":"Three-strand torus knots have topological 4-genus exactly equal to half their maximal Levine-Tristram signature, a computable integer, and also equal to their untwisting number. The paper also lowers the known upper bound for the asymptotic ratio between topological 4-genus and Seifert genus of torus knots from 2/3 to 14/27.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Figure 1's local move is the linchpin of Theorem 1 and is not algebraically or computationally verified.","rationale":"The reader's verdict of CONDITIONAL is appropriate. The central theorem depends on an upper bound on the untwisting number, which in turn depends on the local move in Figure 1. That move is only justified by a diagram, not by an algebraic sequence of braid relations, so it constitutes a real, localized gap. The other components of the proof, including the signature lower bounds and the braid calculus in Lemma 5, appear checkable and likely correct based on the textual derivations. The concern is not a demonstrated counterexample but an unverified load-bearing step, which justifies a conditional acceptance rather than a rejection or a stronger verdict. The proposed algebraic or computational check of Figure 1 would settle the matter directly.","tokens_in":7293,"tokens_out":49744,"duration_ms":419102,"concrete_test":"Formalize Figure 1: in B4, take the braid word for abbaabba on strands 1-3 with strand 4 straight, insert the full twist (sigma1 sigma2 sigma3)^4 on the four strands at the marked position, then perform the indicated crossing change. Verify by explicit braid relations and Markov moves, or with a braid calculator (e.g., SageMath braid functions), that the closure is isotopic to the closure of b^2 (with strands 1 and 4 straight). If this equivalence holds, Lemma 5's untwisting bound is sound; if it fails, the central theorem does not follow.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1 is proved by the chain sigma_hat/2 <= g_t <= t <= ceil(2n/3) = sigma_hat/2. The lower bound sigma_hat/2 = ceil(2n/3) follows from Litherland's jump formula and is carefully checked. The upper bound t <= ceil(2n/3) is the entire content of Lemma 5. Every non-base case in Lemma 5 hinges on the assertion in Figure 1 that the closure of the 3-braid abbaabba is related to the closure of bb by one null-homologous twist on four strands followed by one crossing change. This assertion is not derived algebraically; it is a visual claim. If it were false, the untwisting upper bound would fail for the infinite families T(3,6k+16) and T(3,6k+19), and Theorem 1 would collapse. The same local move is reused in Proposition 2 and in the asymptotic argument of Theorem 3, so the entire upper-bound machinery depends on this single unverified pictorial step. The paper itself sets a higher standard elsewhere, explicitly flagging signature equalities in Proposition 2 as unproved, which makes the absence of an algebraic verification of Figure 1 a genuine gap rather than a stylistic choice.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the topological 4-genus of torus knots. The main result (Theorem 1) asserts that for every n ≥ 4 not divisible by 3, the topological 4-genus of the 3-strand torus knot T(3,n) equals the maximal Levine-Tristram signature bound σ_hat/2 = ceil(2n/3), and that this value is also the untwisting number t(T(3,n)). The proof combines McCoy's inequality g_t ≤ t with explicit sequences of null-homologous twists (upper bound) and Litherland's signature jump formula (lower bound). The paper further claims (Proposition 2) that the same invariants for T(4,n) and T(6,n) are within 1 of the signature bound, and (Theorem 3) improves the asymptotic upper bound on g_t/g for torus knots from 2/3 to 14/27.","tokens_in":7504,"tokens_out":7960,"duration_ms":77878,"significance":"If Theorem 1 holds, it provides the first infinite family of torus knots for which the topological 4-genus is exactly equal to the signature bound, giving a clean topological analogue of the smooth local Thom conjecture; it also identifies the untwisting number with these invariants. The proof is elementary and mostly explicit, building on recent work of McCoy, and the paper gives concrete braid sequences for the upper bounds. The improvement of the asymptotic ratio is a modest but real advance. The main weakness is that the key local move (Figure 1) and several secondary moves (Figure 2, Figure 3) are verified only pictorially, and the lower bounds in Proposition 2 are explicitly left unproved.","major_comments":[{"comment":"The assertion that the braids abbaabba and bb are related by one null-homologous twist on four strands followed by one crossing change is the linchpin of Lemma 5 and hence of the entire upper bound in Theorem 1; it is verified only by a drawing, not by an algebraic braid computation. This move is used for both infinite families T(3,6k+16) and T(3,6k+19), and a failure of this pictorial claim would invalidate the untwisting upper bound and Theorem 1. Please provide an explicit algebraic sequence of braid relations (or a short computer script) establishing this move.","section":"Section 3, Lemma 5 and Figure 1"},{"comment":"The proof of Proposition 2 states that 'by a similar calculation ... one may prove' the inequalities σ_hat(T(4,n)) ≥ 2n and σ_hat(T(6,n)) ≥ 3n+1, and then 'We note (without proof) that the stated inequalities for σ_hat are in fact equalities.' Since Proposition 2 is stated as a theorem and advertised in the abstract and introduction, these bounds need a proof; if they are omitted intentionally, the proposition should be rephrased as conditional or its proof sketched. The absence of proof is especially notable because the paper itself flags it.","section":"Section 4, Proposition 2"},{"comment":"The 'crucial move' transforming (abc)^12 into (bc)^12 by four twist operations is presented via Figure 2 without algebraic verification, and this move underpins all upper bounds for t(T(4,n)) in Proposition 2. As with Figure 1, a pictorial verification is not sufficient for a load-bearing step; please provide a braid-word or computational check.","section":"Section 4, Figure 2"},{"comment":"The proof states (items (1)–(4)) that a single twist transforms T(2n,2n+1) into the 'disjoint union' of copies of T(n,2n+1), whereas the preceding sentence and the analogous argument in Section 4 refer to the connected sum. The untwisting number is defined for knots, and untwisting each component of a split union into an unknot would produce an unlink, not the unknot; the bound on t(T(3·2^k, 3·2^k+1)) therefore requires the connected-sum interpretation. Please reconcile the terminology and make explicit how the untwisting operations are applied to the summands.","section":"Section 5, proof of Theorem 3"}],"minor_comments":[{"comment":"There is a typo in the abstract: 'off' should be 'off'.","section":"Abstract / general"},{"comment":"The caption reads 'The second twist is a crossing changes'; this should be 'a crossing change'.","section":"Figure 1 caption"},{"comment":"In the paragraph on signature bounds for T(6,n), the expression 'T(6,k+5)' should presumably be 'T(6,6k+5)' (or another integer congruent to 5 modulo 6), since T(6,n) with n coprime to 6 is a torus knot; please clarify the indexing.","section":"Section 4"},{"comment":"The list in items (1)–(4) uses 'disjoint union' while the surrounding prose and Section 4 use 'connected sum'; please use one consistent term throughout, as discussed in the major comments.","section":"Section 5"},{"comment":"The notation '5k' in 'L[3,5k,4,3,5k,4]' is explained as a sequence of length k, but the same notation is also used for a scalar in Lemma 5; a brief reminder of this convention would help the reader.","section":"Section 3, after Lemma 4"}],"recommendation":"major_revision","confidential_remarks":"The paper is interesting and the main chain of inequalities for Theorem 1 is transparent and likely correct. However, the pivotal local move in Figure 1 is verified only visually, and the paper itself acknowledges missing proofs in Proposition 2; the referee report requests fixing these gaps. The disjoint-union/connected-sum inconsistency in Section 5 is minor if the connected-sum interpretation is intended, but needs to be resolved. Overall, the manuscript is within scope for a serious journal and the result is worth publishing after the requested revisions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main result is real and worth knowing: for every 3-strand torus knot, the topological 4-genus equals both the maximal Levine–Tristram signature bound and the untwisting number. That is a genuinely new infinite family, not just one more example. The proof strategy is sensible: lower bound from Litherland's jump formula, upper bound via explicit null-homologous twist sequences. The asymptotic improvement from 2/3 to 14/27 is a nice, short application of McCoy's induction, and the braid calculus in Section 2 is both useful and clearly explained.\n\nThe soft spots are exactly where the reader's report puts them. The local move in Figure 1—the claim that the braids abbaabba and bb are related by one twist on four strands plus one crossing change—is the linchpin of Lemma 5, and hence of the entire upper-bound machinery for Theorem 1. It is asserted visually, not algebraically. For a paper that is otherwise careful about proof, this is a genuine gap. I think the move is very likely correct, but a referee should ask for either a short algebraic braid-sequence verification or a computer-checkable certificate. This is not a fatal flaw, but it is load-bearing.\n\nProposition 2 also contains signature inequalities for 4- and 6-strand torus knots that are stated without proof, with the footnote that they are in fact equalities. Those are not needed for Theorem 1, only for the secondary results, and the same comment applies: they deserve a proof or a clear reference.\n\nThe citation pattern looks clean—previous work by McCoy, Litherland, and others is used as a tool, and the paper's own prior calculus is used appropriately. The writing is clear and the limitations are honestly flagged. The main theorem is significant enough that this deserves a serious referee, not a desk reject.\n\nBottom line: send it out. Ask for the Figure 1 move to be verified algebraically and for the unproved signature equalities in Proposition 2 to be either proved or replaced by a reference. After that, it is a solid paper.","headline":"Proves the topological 4-genus equals the signature bound for all 3-strand torus knots, with a clean asymptotic bonus, but the key local move in Figure 1 is verified only pictorially and should be pinned down algebraically before it goes to press.","tokens_in":8073,"tokens_out":1438,"would_cite":true,"duration_ms":17703,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57K10","57N13"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every 3-strand torus knot $T(3,n)$ with $n\\ge 4$ and $3\\nmid n$ has topological 4-genus exactly $\\lceil 2n/3\\rceil$, equal to its signature bound and its untwisting number.","keywords":["topological 4-genus","3-strand torus knots","Levine-Tristram signature","untwisting number","null-homologous twist","torus knots","positive braids"],"falsifier":"Run a braid-group computation on the word in Figure 1: write the null-homologous four-strand twist as an explicit braid word and check whether one further crossing change turns the closure into the unknot, i.e. whether $abbaabba$ really lies one twist plus one crossing change away from $bb$. A direct calculation of the Levine-Tristram signature of the intermediate four-strand twist closure would detect a mismatch: if its signature is not compatible with the claimed trefoil-summand diagram, the local move collapses and Theorem 1 loses its upper bound.","tokens_in":7080,"feed_emoji":"🪢","tokens_out":15009,"duration_ms":134567,"temperature":0.7,"pith_summary":"This paper proves that for every 3-strand torus knot $T(3,n)$ with $n\\ge 4$ and $n$ not divisible by three, the topological 4-genus—the smallest genus of a locally flat surface in the 4-ball bounded by the knot—is exactly $\\hat{\\sigma}(T(3,n))/2$, and this number is $\\lceil 2n/3\\rceil$. Equivalently, the signature lower bound is sharp for the whole family, and the minimal genus is also the untwisting number: the number of null-homologous twists needed to turn the knot into the unknot. The proof is combinatorial: explicit sequences of such twists reduce every $T(3,n)$ to the unknot in $\\lceil 2n/3\\rceil$ moves, while a computation of the maximal Levine-Tristram signature shows no fewer moves can suffice. The same braid-calculus method shows the signature bound is off by at most 1 for 4- and 6-strand torus knots, and it improves the known asymptotic upper bound for the topological 4-genus of general torus knots from $2/3$ to $14/27$.","feed_headline":"Signature bound is exact for 3-strand torus knots","feed_subtitle":"For each T(3,n), the signature lower bound is sharp and equals the untwisting number.","key_machinery":"The engine is a calculus for positive 3-braids written as $[k_1,\\dots,k_n] = a^{k_1}b\\,a^{k_2}b\\cdots a^{k_n}b$, together with the local move drawn in Figure 1: the braid word $abbaabba$ is carried to $bb$ by one null-homologous twist on four strands followed by one crossing change. Iterating this move, with Lemma 4's presentations of powers of the full twist, produces the untwisting sequences behind Lemma 5's bounds $t(T(3,3k+4))\\le 2k+3$ and $t(T(3,3k+5))\\le 2k+4$, which are exactly $\\lceil 2n/3\\rceil$. The passage from untwisting to genus is the general theorem that a null-homologous twist changes the knot by a surface of genus one inside the 4-ball, so the untwisting number is an upper bound for the topological 4-genus. The lower side is supplied by the jump formula for the Levine-Tristram signature, which lets the paper compute $\\hat{\\sigma}$ from the positions of discontinuities and their signs. For the asymptotic result, the key operation is that a full twist on $2n$ strands can be transformed by one twist into two parallel double full twists on $n$ strands.","core_discovery":"On the paper's own terms, the discovery is that for $T(3,n)$ the topological 4-genus problem collapses to a braid-word problem. Theorem 1 states that for every natural number $n\\ge 4$ with $3\\nmid n$, $g_t(T(3,n)) = \\hat{\\sigma}(T(3,n))/2 = t(T(3,n)) = \\lceil 2n/3\\rceil$, where $\\hat{\\sigma}$ is the maximum of the Levine-Tristram signature over the unit circle away from roots of the Alexander polynomial and $t$ is the untwisting number. Because the inequality $\\hat{\\sigma}/2 \\le g_t \\le t$ holds for all knots, the two outer quantities are forced into equality by constructing untwisting sequences of exactly the signature bound's length; the minimal genus surfaces are never drawn but are guaranteed to exist by the theorem that null-homologous twists give locally flat surfaces. The paper also establishes upper bounds $n \\le g_t(T(4,n)) \\le n+1$ and $(3n+1)/2 \\le g_t(T(6,n)) \\le (3n+3)/2$ within the signature-to-untwisting window, and a limsup bound of $14/27$ for the ratio $g_t/g$ over all torus knots.","pith_inferences":["The same braid-calculus framework could be pushed to torus knots with more strands: Proposition 2's 'off by at most 1' for four and six strands suggests that exact equality might hold for fixed $p$ if an analogue of Figure 1's local move exists for $p$-strand full twists.","A computer search over braid words could test whether the Figure 1 move is the first of an infinite family of moves, and whether all 3-strand torus knots satisfy $t = \\lceil 2n/3\\rceil$ even when $n$ is divisible by three, where $T(3,n)$ is a 3-component link.","If the asymptotic ratio really converges to $1/2$, the $14/27$ bound here is not the end; the paper's induction from 3-strand bases, combined with sharper untwisting bounds at higher strand counts, is a natural route to close the gap.","Untwisting number is a combinatorial, unknotting-type invariant; its equality with a signature bound suggests that topological 4-genus for families of braid closures could be computed without constructing surfaces, which would be useful for computational knot tables."],"forward_implications":["The whole infinite family of 3-strand torus knots satisfies $g_t = \\hat{\\sigma}/2 = t = \\lceil 2n/3\\rceil$, so the signature lower bound is both sharp and attained by untwisting sequences rather than by explicit minimal surfaces.","Since the untwisting number equals the topological 4-genus for these knots, any algorithm that computes untwisting numbers for them also computes their topological 4-genus.","For 4- and 6-strand torus knots, the topological 4-genus lies within 1 of the signature bound, so the equality pattern nearly persists for two more strand counts.","The asymptotic ratio $\\limsup g_t/g$ over all torus knots is at most $14/27 \\approx 0.519$, improving the previous $2/3$ upper bound and supporting the conjecture that the ratio tends to $1/2$.","The untwisting construction is inductive and explicit: every $T(3,n)$ admits a prescribed sequence of null-homologous twists, so the genus bound is certified by a finite braid word."],"supporting_citations":[{"why":"supplies the central upper-bound tool: a null-homologous twist can be realized by a locally flat surface, so the untwisting number bounds the topological 4-genus; also provides the induction scheme for the asymptotic ratio.","marker":"[9]"},{"why":"establishes the lower bound on topological 4-genus by the signature and by the maximal Levine-Tristram signature.","marker":"[11]"},{"why":"gives the jump formula for discontinuities of the Levine-Tristram signature of torus knots, used to compute the sharp lower bound.","marker":"[6]"},{"why":"supplies periodicity of the ordinary signature for torus knots, used in the lower-bound estimates for $\\hat{\\sigma}$.","marker":"[4]"},{"why":"provides the disc theorem that turns null-homologous twists into embedded locally flat surfaces with controlled genus.","marker":"[3]"},{"why":"defines the untwisting number as the minimal number of null-homologous twists needed to reach the unknot, the invariant being sharpened.","marker":"[5]"},{"why":"introduces the positive 3-braid notation $[k_1,\\dots,k_n]$ and the rewriting calculus used throughout the untwisting constructions.","marker":"[1]"},{"why":"gives earlier bounds on the topological 4-genus of torus knots and the subadditivity principle that turns the asymptotic example into a general limsup bound.","marker":"[2]"}],"fun_headline_variants":["Signature bound sharp for all 3-strand torus knots","Exact 4-genus for 3-strand torus knots","Untwisting matches signature bound for T(3,n)","Torus knot ratio improved to 14/27"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the pictorial claim in Figure 1 that one null-homologous twist on four strands plus one crossing change transforms the braid $abbaabba$ into $bb$; the paper verifies that move only by the figure, not by an algebraic sequence of braid relations, and every untwisting upper bound in Lemma 5 is built from it.","fun_headline_variants_meta":{"raw":{"variants":["Signature bound sharp for all 3-strand torus knots","Exact 4-genus for 3-strand torus knots","Untwisting matches signature bound for T(3,n)","Torus knot ratio improved to 14/27"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000296,"raw_usage":{"total_tokens":1691,"prompt_tokens":889,"completion_tokens":802,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":505,"completion_tokens_details":{"reasoning_tokens":732}},"tokens_in":505,"tokens_out":802,"duration_ms":7960,"temperature":1.0,"reasoning_tokens":732,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T05:28:48.977542+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a braid-group computation on the word in Figure 1: write the null-homologous four-strand twist as an explicit braid word and check whether one further crossing change turns the closure into the unknot, i.e. whether $abbaabba$ really lies one twist plus one crossing change away from $bb$. A direct calculation of the Levine-Tristram signature of the intermediate four-strand twist closure would detect a mismatch: if its signature is not compatible with the claimed trefoil-summand diagram, the local move collapses and Theorem 1 loses its upper bound.","supporting_citations":[{"cited_title":"Null-homologous twisting and the algebraic genus","cited_arxiv_id":"1908.04043","evidence_quote":"supplies the central upper-bound tool: a null-homologous twist can be realized by a locally flat surface, so the untwisting number bounds the topological 4-genus; also provides the induction scheme for the asymptotic ratio."},{"cited_title":"Powell: The four-genus of a link, Levine-Tristram signatures and satellites , J","cited_arxiv_id":null,"evidence_quote":"establishes the lower bound on topological 4-genus by the signature and by the maximal Levine-Tristram signature."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"gives the jump formula for discontinuities of the Levine-Tristram signature of torus knots, used to compute the sharp lower bound."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies periodicity of the ordinary signature for torus knots, used in the lower-bound estimates for $\\hat{\\sigma}$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the disc theorem that turns null-homologous twists into embedded locally flat surfaces with controlled genus."},{"cited_title":"Ince: The untwisting number of a knot , Pacific J","cited_arxiv_id":null,"evidence_quote":"defines the untwisting number as the minimal number of null-homologous twists needed to reach the unknot, the invariant being sharpened."},{"cited_title":"Baader: Positive braids of maximal signature , Enseign","cited_arxiv_id":null,"evidence_quote":"introduces the positive 3-braid notation $[k_1,\\dots,k_n]$ and the rewriting calculus used throughout the untwisting constructions."},{"cited_title":"Baader, P","cited_arxiv_id":null,"evidence_quote":"gives earlier bounds on the topological 4-genus of torus knots and the subadditivity principle that turns the asymptotic example into a general limsup bound."}],"review_version":1}