{"id":"e91949ba-46fa-4f18-a479-b19ae232ee24","arxiv_id":"1908.09458","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"New formulas express the braid index and HOMFLY polynomial of a rational link in terms of its reduced alternating continued fraction expansion, via a new primitive-block conversion to all-even form.","lead":"This paper gives an algorithm that converts the reduced alternating diagram of a rational link into the special 'all-even' continued fraction form, and uses it to write the braid index and HOMFLY polynomial directly in terms of the minimal diagram. It removes the need to switch to a non-minimal representation when computing these two invariants.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.4's sign-replacement rules are the load-bearing, unverified pivot: a parity error there would corrupt both invariant formulas, and the proof delegates the key case check to the reader.","rationale":"I read the paper as a genuine derivation of braid-index and HOMFLY formulas from reduced alternating diagrams via an explicit continued-fraction conversion. The arithmetic manipulations in Section 3 are concrete and the example in Example 4.5 checks out once one uses the equivalent nonalternating form ending in [5,2,1] rather than [5,3]; the output evaluates to 34651/49654. This supports the authors' claim that the conversion is computationally meaningful. The reader's conditional verdict correctly targets Theorem 4.3/4.4, and I find that to be the weakest load-bearing step: it is an informal automaton argument rather than a case-checked proof, and Theorem 7.5 explicitly leaves the essential verification to the reader. The braid-index formula has some independent support through its equivalence with the earlier formula in Section 6, but the HOMFLY formula has no such external check. Since no counterexample or internal inconsistency is shown, I do not move the verdict from CONDITIONAL. The proposed exhaustive comparison is the most direct way to settle whether the sign-parity rules in Theorem 4.4 are correct for all rational links.","tokens_in":22792,"tokens_out":25643,"duration_ms":223721,"concrete_test":"Run an exact-arithmetic exhaustive comparison for every reduced rational p/q with 1 ≤ q ≤ 300 and pq even: (1) compute the all-even form from Theorem 4.4 and verify the output evaluates to p/q with all denominators even; (2) compute P(K) via Theorem 7.5 and compare it with the Lickorish–Millett all-even formula (7.2), and, where applicable, with the Duzhin–Shkolnikov formula; (3) check Theorem 5.5 against the Cromwell–Murasugi index computed directly from the all-even form. Include cases with a_n = ±1 and with crossing-sign changes at block boundaries. If every case matches, the concern is resolved; if not, the failing rule is isolated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claims (Theorems 5.5 and 7.5) both depend on converting the primitive-block nonalternating continued fraction into the unique all-even form. The conversion is implemented by the replacement rules of Theorem 4.4, whose sign exponent τ(i) mixes block-boundary detection with a parity count. The proof of Theorem 4.4 is not a formal case enumeration: it invokes the automaton of Figure 5 and states that the rules follow 'after verifying that (−1)^{τ(i)} sign(a1) is the correct sign'; Theorem 7.5 similarly ends with 'the details of the verification are left to the reader.' Theorem 5.5 uses only coarse crossing-sign parity and is therefore less sensitive, but Theorem 7.5 assigns a matrix factor to every original a_i, so a single wrong sign or parity in a rule—especially rule (4), which produces length-|a_i|−1 alternating strings and is used at block boundaries—would silently yield a wrong HOMFLY polynomial. No independent check against the Duzhin–Shkolnikov formula or a machine enumeration is provided. This is not a demonstrated error, but it is the point where the paper's central claim is least secure.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an algorithmic conversion of a rational link's continued fraction from the nonalternating (reduced alternating diagram) form to the all-even form used in Murasugi's braid-index formula and Lickorish-Millett's HOMFLY formula. The conversion is organized around primitive blocks (Definition 3.3) and a replacement rule (Theorem 4.4) that tracks crossing signs via the automaton of Figure 5. From this conversion the paper derives a braid-index formula (Theorem 5.5), compares it with the authors' earlier formula from [4] (Section 6), and derives a matrix-product HOMFLY formula stated directly in terms of the original partial denominators and crossing signs (Theorem 7.5). Worked examples include the knots and links 1402/1813, 3244/4195, and the two-component link 5075/17426.","tokens_in":23013,"tokens_out":11354,"duration_ms":100071,"significance":"Should the formulas hold in full generality, they give a practical way to compute both invariants from a minimal alternating diagram without first constructing the highly non-minimal all-even diagram, and they cover the case pq odd via mirror images. The paper provides several internal consistency checks: Example 5.4 verifies the Cromwell-Murasugi index of 1402/1813 both with Theorem 5.3 and with Definition 5.1; Example 6.5 computes the braid index of a two-component link in several ways; Example 7.4 illustrates the HOMFLY block product. These checks lend credence to the main formulas. The main weakness is proof completeness: the key sign-tracking steps underpinning Theorem 4.4 and Theorem 7.5 are delegated to the reader rather than proved or machine-checked.","major_comments":[{"comment":"The replacement rules in Theorem 4.4 are the pivot on which both Theorem 5.5 and Theorem 7.5 rest, yet their proof is not a formal case enumeration. The proof states that the rules \"follow from Proposition 3.2, after verifying that (-1)^tau(i) sign(a1) is the correct sign,\" and the rest is an informal description of the automaton states. In particular, rule (4) generates alternating strings of length |a_i|-1, so an off-by-one parity error at a block boundary would silently change every subsequent sign and hence both invariant formulas. Please supply a complete proof, or a machine-checked enumeration, covering all primitive-block types and all four replacement rules, and specify exactly how the automaton determines the parity count entering tau(i).","section":"Section 4, Theorem 4.4"},{"comment":"Theorem 7.5 is the paper's central HOMFLY result, but its proof ends with \"The details of the verification are left to the reader.\" This is not a minor omission: the passage from the block products (7.10)-(7.11) to the per-entry matrices H(a_i) requires matching the all-even form produced by Theorem 4.4 with the order and conjugation conventions in Proposition 7.1, case by case. Please provide the detailed verification, and ideally add an independent check of a nontrivial example against Proposition 7.1 or the Duzhin-Shkolnikov formula.","section":"Section 7, Theorem 7.5"}],"minor_comments":[{"comment":"The displayed computation \"1 + 1 + 2/2 + 3/2 + 5/2 + 3/2 = 8\" is arithmetically 8.5; the intended value 8 is obtained as 1 + (1 + 2 + 3 + 5 + 3)/2. Please correct the displayed formula.","section":"Example 5.4"},{"comment":"The notation M(-2/2), M(-4/2), M(6/2), M(4/2) is ambiguous: read literally, M(-2/2) would be M(-1), which is not of the form M(2r) used in the paper. Please clarify whether these denote M(-2), M(-4), M(6), M(4) or the corresponding r-values, and correct the example.","section":"Example 7.4"},{"comment":"The automaton in Figure 5 is introduced under the assumption ai > 0 and epsilon(B1) = +1, but Theorem 4.4 is stated for a general nonalternating continued fraction with s = sign(a1). Please state explicitly how the automaton and the parity rules are adapted when all ai have the opposite sign.","section":"Theorem 4.4 and Figure 5"},{"comment":"The proof of Theorem 6.4 verifies cases (i) and (ii) in detail and leaves cases (iii) and (iv) to the reader. Since this is a consistency check rather than a load-bearing step, a brief completion of the remaining cases would remove the asymmetry.","section":"Section 6, Theorem 6.4"}],"recommendation":"major_revision","confidential_remarks":"My recommendation is driven by the incomplete proof of the sign-tracking conversion, not by a detected counterexample. A resubmission with a completed proof of Theorem 4.4 and Theorem 7.5, or a machine-checked case enumeration, would settle whether the formulas are fully correct."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a genuine contribution, not a repackaging. The authors give an arithmetic algorithm (primitive block decomposition) that converts the nonalternating continued fraction of a rational link into the all-even form, and then state braid index and HOMFLY formulas directly in terms of the alternating diagram. The braid index formula is closely related to their earlier preprint [4], but the proof via Murasugi and the primitive-block formulation is different. The HOMFLY formula as a product of 2x2 matrices built from Fibonacci polynomials appears to be new. The examples are consistent; I spot-checked the 1402/1813 computation and the Cromwell-Murasugi index comes out 8 both ways. That is real evidence the mechanism works.\n\nThe soft spots are exactly where the proofs get informal. Theorem 4.4, which encodes the sign replacement rules, is proven by saying the rules 'follow' from the automaton after verifying a sign formula; Theorem 7.5's verification is explicitly left to the reader. This is where I'd want more care, because a parity slip in rule (4) would silently corrupt the HOMFLY product. I don't see an actual error in the examples, and the automaton reading is plausible, but the paper is asking the reader to trust the most load-bearing step without a full case enumeration. The absence of a comparison with the Duzhin-Shkolnikov HOMFLY formula is also a missed opportunity and would be a cheap independent check.\n\nThe paper is honest about limitations: it notes when the all-even form doesn't exist directly (both p,q odd) and says what to do via mirror image. The earlier [4] connection is acknowledged, not hidden.\n\nWho is this for? Knot theorists working on rational links and people who want explicit formulas from minimal diagrams. It deserves a serious referee, but the referee should ask for a proof of Theorem 4.4 that doesn't delegate the key case verification, and ideally a comparison with Duzhin-Shkolnikov on a few examples.\n\nRecommendation: send to peer review, with revisions.","headline":"Useful new formulas for braid index and HOMFLY from minimal alternating diagrams, but the key sign-conversion proof leans too hard on 'left to the reader.'","tokens_in":23527,"tokens_out":2267,"would_cite":true,"duration_ms":21640,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57M25","57M27"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes that for every rational link, the braid index and the HOMFLY polynomial can be computed directly from the reduced alternating diagram (the nonalternating continued fraction form), by an algorithm that splits the…","keywords":["rational links","continued fractions","braid index","HOMFLY polynomial","alternating diagrams","primitive block decomposition","all-even form","Fibonacci polynomials"],"falsifier":"Compute the braid index of every rational link with denominator up to, say, $q=200$ using Theorem 5.5 and compare with the value from the all-even expansion of Definition 5.1; any mismatch would identify a counterexample to the sign rules, and the same comparison can be made for the HOMFLY matrix product against a direct skein computation for all rational links with at most 12 crossings.","tokens_in":22580,"feed_emoji":"🔗","tokens_out":6207,"duration_ms":62194,"temperature":0.7,"pith_summary":"Rational links can be represented by many continued fractions for the same rational number. This paper establishes that the two standard invariants—braid index and HOMFLY polynomial—can be computed directly from the reduced alternating diagram, i.e. from the nonalternating continued fraction in which all partial denominators share one sign. The key is an algorithmic conversion, block by block, of that fraction into the all-even form on which the classical formulas were based. This matters because the all-even expansion is usually non-minimal and does not even exist when both integers defining the link are odd. The resulting formulas (Theorems 5.5 and 7.5) read the invariants off the minimal diagram, with mirror-image substitution covering the odd–odd case.","feed_headline":"New direct formulas give braid index and HOMFLY for rational links","feed_subtitle":"A continued-fraction conversion computes both invariants from the reduced alternating diagram, skipping the old all-even step.","key_machinery":"The central object is the primitive block decomposition of a nonalternating continued fraction. A primitive block is either a single even partial denominator, or an odd-length stretch $a_m,a_{m+1},\\dots,a_{m+2k}$ whose end entries are odd, whose interior even-position entries are even, and whose entries share one sign; except for a possible final exceptional block, the decomposition is unique when it exists. Alongside this sits a finite sign-tracking automaton (Figure 5) that records for each twistbox whether its crossing sign is positive or negative depending on the parity of the partial denominator and the position inside its block. The automaton is what proves that crossing signs are constant within each primitive block and opposite in adjacent blocks, and it is the machine from which both invariant formulas are read.","core_discovery":"The paper's central claim is that the braid index and the HOMFLY polynomial of a rational link no longer need the special all-even continued fraction expansion. Starting from the reduced alternating diagram—equivalently, the nonalternating continued fraction $[0,a_1,\\dots,a_n]$—the authors give an explicit primitive block decomposition and a sign-tracking rule (Theorems 4.3 and 4.4) that converts the fraction into all-even form block by block. This conversion yields a closed formula for the braid index (Theorem 5.5): $b(K)=1+\\frac12$(sum of selected odd/even partial denominators chosen by the crossing signs of their twistboxes) plus a $0$ or $1/2$ correction term. It also yields a formula for the HOMFLY polynomial (Theorem 7.5) as the matrix product $H(a_n)\\cdots H(a_1)$ applied to the standard column vector $(1,(a^2-1)/(az))^T$, where each $H(a_i)$ is a $2\\times2$ matrix built from Fibonacci polynomials. When the numerator and denominator are both odd, so that no all-even form exists for the original diagram, the formulas apply to the mirror image and the substitution $a\\mapsto a^{-1}$ recovers the invariant.","pith_inferences":["Because the sign automaton has only finitely many states, the parity rule in Theorem 4.4 could be checked exhaustively over all primitive blocks up to a fixed size; such a computer enumeration would give independent confirmation of the informal automaton analysis without requiring a formal case proof.","The same primitive-block scaffolding may extend to other alternating link families whose twistboxes share the same block-sign structure, potentially yielding HOMFLY formulas beyond rational links.","The Fibonacci-polynomial matrix entries suggest a path-counting interpretation: powers of the matrices $M(2)$ and $M(-2)$ count weighted lattice paths, so the HOMFLY polynomial of a rational link may be readable as a weighted path sum over the minimal diagram."],"forward_implications":["The braid index of any rational link can be computed from its minimal alternating diagram alone, without first constructing a larger all-even diagram.","The HOMFLY polynomial becomes a finite product of $2\\times2$ matrices with Fibonacci-polynomial entries, with the number of factors equal to the number of partial denominators in the reduced fraction rather than the inflated all-even expansion.","Rational links whose defining integers are both odd, which previously required a mirror-image workaround before applying the all-even formulas, are now handled by the same formulas with the substitution $a\\mapsto a^{-1}$.","The two braid-index formulations for rational links—one based on the preferred standard form and one based on the alternative standard form—are shown to agree through elementary continued-fraction manipulations."],"supporting_citations":[{"why":"Supplies the all-even braid-index formula that the new theorem converts into alternating-form terms.","marker":"[13]"},{"why":"Supplies the all-even HOMFLY matrix product that the new theorem rewrites using primitive-block data.","marker":"[11]"},{"why":"Gives the earlier braid-index formula for alternating links whose equivalence to the new formula is proven here.","marker":"[4]"},{"why":"Provides the statement of the all-even index used as the starting point for the braid-index computation.","marker":"[3]"},{"why":"Observes the existence and uniqueness properties of the all-even form that the paper generalizes to primitive blocks.","marker":"[5]"},{"why":"Introduces the HOMFLY polynomial, the invariant for which the paper gives a new formula.","marker":"[8]"}],"fun_headline_variants":["Braid index and HOMFLY from alternating diagrams directly","New formulas skip the all-even step for rational links","Direct invariants for rational links via alternating form","Reduced diagrams yield braid index and HOMFLY in one go"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the claim that, for every nonalternating fraction with a primitive block decomposition, crossing signs are constant within each block, opposite between adjacent blocks, and correctly encoded by the replacement rules of Theorem 4.4 with no hidden parity exception.","fun_headline_variants_meta":{"raw":{"variants":["Braid index and HOMFLY from alternating diagrams directly","New formulas skip the all-even step for rational links","Direct invariants for rational links via alternating form","Reduced diagrams yield braid index and HOMFLY in one go"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000152,"raw_usage":{"total_tokens":1233,"prompt_tokens":1005,"completion_tokens":228,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":621,"completion_tokens_details":{"reasoning_tokens":159}},"tokens_in":621,"tokens_out":228,"duration_ms":2668,"temperature":1.0,"reasoning_tokens":159,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:10:40.513226+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the braid index of every rational link with denominator up to, say, $q=200$ using Theorem 5.5 and compare with the value from the all-even expansion of Definition 5.1; any mismatch would identify a counterexample to the sign rules, and the same comparison can be made for the HOMFLY matrix product against a direct skein computation for all rational links with at most 12 crossings.","supporting_citations":[{"cited_title":"Murasugi On The Braid Index of Alternating Links, Trans","cited_arxiv_id":null,"evidence_quote":"Supplies the all-even braid-index formula that the new theorem converts into alternating-form terms."},{"cited_title":"Lickorish and Kenneth C","cited_arxiv_id":null,"evidence_quote":"Supplies the all-even HOMFLY matrix product that the new theorem rewrites using primitive-block data."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the earlier braid-index formula for alternating links whose equivalence to the new formula is proven here."},{"cited_title":"Cromwell, Knots and links, Cambridge University Press, 2004","cited_arxiv_id":null,"evidence_quote":"Provides the statement of the all-even index used as the starting point for the braid-index computation."},{"cited_title":"Duzhin and M","cited_arxiv_id":null,"evidence_quote":"Observes the existence and uniqueness properties of the all-even form that the paper generalizes to primitive blocks."},{"cited_title":"Freyd, D","cited_arxiv_id":null,"evidence_quote":"Introduces the HOMFLY polynomial, the invariant for which the paper gives a new formula."}],"review_version":1}