{"id":"eb9ffbe4-3b62-4b79-a6e3-4081ff8ed8ea","arxiv_id":"2607.19969","paper_version":2,"verdict":"REJECT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The claimed theorem—quantum Latin squares of order v exist for every cardinality in [v,v²]\\{v+1} for v≥8, v∉{9,11,23}—is not established because key lemmas are false or only give arithmetic progressions.","lead":"This paper claims that for almost every order v≥8, a quantum Latin square exists with every cardinality between v and v² except v+1. The proof is not valid as written: two load-bearing steps replace arithmetic progressions with full intervals and rely on a false distinctness lemma.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Affine cardinality formulas are treated as intervals; in Lemma 4.8 for v=17 the constructions produce only values ≡1 mod 4, missing c=20 required by Theorem 1.2.","rationale":"The reader's verdict is REJECT, and I agree with the outcome. The most decisive reason is not the (also real) false Lemma 2.2(1) but the affine-cardinality error. The main constructions in Section 3 have cardinality formulas that are linear functions of input set sizes with coefficient n (the order of the template Latin square). When an input set size c0 is allowed to run over an interval of integers, the output runs over an arithmetic progression with difference n. The proof repeatedly replaces that progression by the full interval. The sharpest instance is Lemma 4.8 for u=4, where the two cases produce only cardinalities congruent to 1 mod 4; c=20 is required by Theorem 1.2 but cannot be obtained. Because v=17 is not covered by Theorem 1.1(2) or any other part of the paper, this is a genuine hole in the central claim, not a minor presentational issue. I also confirmed the reader's Lemma 2.2(1) counterexample; that invalidates Lemma 2.3's coefficient comparison and thereby the disjoint-copy lemmas, so there are two independent serious flaws. My recommended disposition is unchanged (REJECT), and I set agreement_with_reader to 'partial' because the reader's stated weakest assumption is Lemma 2.2(1), while my primary concern is the interval/affine error they mention only secondarily.","tokens_in":22955,"tokens_out":19219,"duration_ms":169495,"concrete_test":"Check Lemma 4.8 for u=4 by enumerating the outputs of its two cases: S1={4c0+1 : c0∈[4,48]\\{5}} and S2={4c0+97 : c0∈[4,48]\\{5}}. Test whether 20 belongs to S1∪S2. (It does not.) Since 20 is in the interval [17,289]\\{18} claimed by Lemma 4.8, this shows the asserted coverage is false and Theorem 1.2 is unproved for v=17.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing defect is that the cardinality formulas are affine with coefficient n, so the intervals asserted in the proof do not follow. Lemma 3.7 gives |M| = n(c0+c1-c2-1)+c3 and Lemma 3.8 gives |M| = n(c0+c1-c2)+c3. In Lemma 4.8, taking n=u, h=1, Case 1 fixes c1=5, c2=4, c3=1 and lets c0 run over [4,16(u-1)]\\{5}; this produces {u·c0+1}, an arithmetic progression, not [4u+1,16u(u-1)+1]\\{4u+2} as claimed. For u=4 (v=17), Case 1 gives {4c0+1 : c0∈[4,48]\\{5}} and Case 2 gives {4c0+97 : c0∈[4,48]\\{5}}; both are subsets of the residue class 1 mod 4. The value c=20 lies in [17,289]\\{18}, which Theorem 1.2 requires for v=17, but it is not in either set. Since v=17 is handled only by Lemma 4.8, the proof of the main theorem is not internally consistent. The identical affine-image error also affects Lemma 4.3 and Tables 3–5. Separately, Lemma 2.2(1) is false because f1f4=f2f3=x^6, invalidating the coefficient comparisons in Lemma 2.3; but the cardinality gap is enough by itself.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to determine all possible cardinalities of quantum Latin squares of order v for every v ≥ 8 except v = 9, 11, 23, namely every c in [v, v^2] \\ {v+1}. The approach adapts classical Latin-square direct-product constructions to the quantum setting: a parameterized vector lemma (Lemma 2.3), a sequence of disjoint-copy corollaries, and block-union cardinality formulas (Lemmas 3.7 and 3.8) are used to assemble QLS(v)s with prescribed cardinalities. The main theorem is then obtained by treating separately the residue classes of v modulo 4 via Lemmas 4.3, 4.8, 4.9, and 4.10.","tokens_in":23402,"tokens_out":14708,"duration_ms":123528,"significance":"If the main theorem were correct, it would essentially settle the possible-cardinality problem for quantum Latin squares, leaving only five open orders. The paper contains useful raw ingredients: explicit small-order examples, a quantum version of the singular direct product, and a framework for controlling intersections of blocks. These are potentially valuable. However, several load-bearing steps are invalid: a key algebraic distinctness lemma is false, and the cardinality formulas used in the residue-class constructions are affine images of parameter intervals rather than the intervals themselves. These are not presentation issues; the main theorem is not established by the submitted proofs.","major_comments":[{"comment":"Lemma 2.2(1) is false. With f_k(x) = x^{(4k-1)/3}, one has f_1(x)f_4(x) = x^{(3+15)/3} = x^6 and f_2(x)f_3(x) = x^{(7+11)/3} = x^6, yet {1,4} ≠ {2,3}. Lemma 2.3 relies on Lemma 2.2 to compare coefficients of distinct powers after substitution, specifically in deriving Eq. (2.5). Since equal powers can merge, the coefficient comparison 4a_n^2 a_k^2 = 0 is not justified. Consequently Lemma 2.3 and the dependent disjoint-copy results (Lemma 2.7, Lemma 2.10, Corollaries 2.8, 2.12, 2.13) are unsupported, and these are the tools used throughout §4 to control intersections of blocks.","section":"§2.2, Lemma 2.2(1)"},{"comment":"The cardinality formulas in Lemma 4.8 do not produce the claimed intervals. Lemma 3.7 gives |M| = n(c0+c1-c2-1)+c3. In Case 1 with n=u, c1=5, c2=4, c3=1, this reduces to |M| = u c0 + 1. As c0 runs over [4,16(u-1)]\\{5}, only values congruent to 1 modulo u are obtained, not the full interval [4u+1,16u(u-1)+1]\\{4u+2}. Concretely, for u=4 (v=17), Case 1 gives the set {4c0+1} and Case 2 gives {4c0+97}; neither contains c=20, which lies in [17,289]\\{18} and is required by Theorem 1.2. Since v=17 is handled only through Lemma 4.8, the main theorem is not proved by this argument. The identical affine-image gap recurs in Lemmas 4.9 and 4.10 and in Tables 3–5.","section":"§4, Lemma 4.8"},{"comment":"The proof of Lemma 4.3 as written is a non sequitur. It says that after applying Construction 3.9 with m=u and n=4, the 'u-fold sums' of the sets from Lemma 4.2 yield [4u,16u^2]\\{4u+1}. But Construction 3.9 outputs cardinality mc, where c is the union cardinality of the m blocks, so with m=u and c0 in the Lemma 4.2 set the outputs are u c0, not arbitrary u-fold sums. For u=3 this produces only multiples of 3 from c0 ∈ [4,48]\\{5}, so c=14 ∈ [12,144]\\{13} is not obtained. The cited prior result [27] may imply Lemma 4.3, but the 'self-contained' derivation supplied here is invalid.","section":"§4, Lemma 4.3"}],"minor_comments":[{"comment":"The phase-normalization step implicitly assumes (a_n,b_n) ≠ (0,0). The case a_n = b_n = 0 is not separated; the conclusion may be easy there, but the proof as written skips it.","section":"§2.3, Lemma 2.3 proof"},{"comment":"Several table rows state intervals as 'attainable cardinalities' although the formulas displayed in the same row yield arithmetic progressions. For example, Table 3, u=3, first row gives values 3c0+1, which are only the numbers congruent to 1 modulo 3. The tables should distinguish intervals from the subsets actually established.","section":"§4, Tables 3–5"},{"comment":"The entry sets in Tables 1 and 2 are asserted without a complete derivation of the additions and intersections. A brief generating-function or induction verification would improve reproducibility.","section":"§4, Lemmas 4.5–4.6"}],"recommendation":"reject","confidential_remarks":"The paper depends heavily on a cluster of prior results by the same or overlapping authors ([23]–[28]), and the only new route to the main theorem is the block-intersection calculus that fails in the ways described above. The core construction idea is interesting, but the central claims are not supported in the present form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The claimed classification is not proven. I agree with the reader's verdict: reject. Two independent errors break the main theorem.\n\nWhat the paper does well: the target is genuine and natural — if Theorem 1.2 were true, it would settle the possible-cardinality problem for QLS for all but five orders. The adaptation of singular direct product and direct product constructions to the quantum setting is a reasonable extension, and the explicit small-order examples are useful. The paper is clearly written and engages the right prior work.\n\nThe soft spots are severe. First, Lemma 2.2(1) is false: for f_k(x)=x^{(4k-1)/3}, f_1(x)f_4(x)=x^6=f_2(x)f_3(x). The distinct-products claim is used to compare coefficients in Lemma 2.3, which underpins the entire disjoint-copy machinery (Lemmas 2.7–2.13). Without Lemma 2.2(1), that part of the proof does not go through.\n\nSecond, and even more decisively, the cardinality formulas in Lemmas 3.7 and 3.8 are affine functions of the input c0. The image of an interval under an affine map is an arithmetic progression, not a contiguous interval. The paper repeatedly treats these images as full intervals — for instance Lemma 4.3, Lemma 4.8, Lemma 4.9, and Tables 3–5. To give a concrete failure: in Lemma 4.8 with u=4 (v=17), Case 1 gives cardinalities {4c0+1} and Case 2 gives {4c0+97}, both contained in the residue class 1 mod 4. The value c=20 lies in [17,289]\\{18}, which Theorem 1.2 requires for v=17, but it is not in either set. Since v=17 is handled only by Lemma 4.8, the main theorem is internally inconsistent.\n\nThe self-citation pattern ([27], [28] carrying the heavy lifting) is not by itself a flaw, but it means the paper's independent contribution is thinner than it first appears.\n\nWho this is for: a specialist in quantum Latin squares or quantum combinatorial designs who wants to see whether the construction framework can be patched. The strategy may be salvageable — the false equality could be fixed by a different parameterization, and the interval claims could be replaced by genuine sums of arithmetic progressions — but the current manuscript does not establish the theorem.\n\nRecommendation: send to peer review? Yes, it deserves a serious referee because the target is important and the flaw is identifiable. But the reviewer will almost certainly return it for major revision or rejection. I would not cite the main theorem yet.","headline":"The claimed classification is not proven: a false equality in Lemma 2.2 and affine-image interval errors break the main theorem, though the construction strategy may be salvageable.","tokens_in":23837,"tokens_out":5103,"would_cite":false,"duration_ms":46504,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05B15"],"pacs":[],"model":"deepseek-v4-flash","headline":"Quantum Latin squares of every order v ≥ 8, except 9, 11, and 23, realize every cardinality from v to v² except v+1.","keywords":["quantum Latin square","cardinality","singular direct product","direct product construction","parameterized unit vectors","orthonormal basis","combinatorial designs","quantum information"],"falsifier":"Compute $f_1(x)f_4(x) = x^{(3+15)/3} = x^6$ and $f_2(x)f_3(x) = x^{(7+11)/3} = x^6$; this contradicts Lemma 2.2(1), on which Lemma 2.3's coefficient comparisons rest. Since those comparisons drive Lemmas 2.7–2.13 and the main constructions, the proof as written collapses at that point; the theorem could still be true, so a complementary test is to look for any cardinality in $[8,64] \\setminus \\{9\\}$ for $v=8$ that is not realized.","tokens_in":1478,"feed_emoji":"⚛️","tokens_out":1551,"duration_ms":73673,"temperature":0.7,"texified_at":"2026-08-05T21:35:25.850143+00:00","pith_summary":"A quantum Latin square of order $v$ is a $v \\times v$ array of unit vectors whose rows and columns are orthonormal bases; its cardinality is the number of distinct vectors in the array. This paper tries to prove that for every $v \\geq 8$ with $v$ not 9, 11, or 23, every integer $c$ in the full range $[v, v^2]$, with the single exception $c = v+1$, is realized by some quantum Latin square. If true, this gives a near-complete answer to the possible-cardinality problem, leaving only orders 6, 7, 9, 11, and 23 unresolved. The proof builds on new product constructions for quantum Latin squares, using a family of parameterized vectors to rotate copies so that their intersections are controlled.","texify_model":"deepseek-v4-flash","texify_usage":{"total_tokens":5847,"prompt_tokens":866,"completion_tokens":4981,"prompt_tokens_details":{"cached_tokens":0},"prompt_cache_hit_tokens":0,"prompt_cache_miss_tokens":866,"completion_tokens_details":{"reasoning_tokens":4155}},"feed_headline":"All v≥8 except 9, 11, 23 attain every cardinality from v to v²","feed_subtitle":"The result settles which counts of distinct vectors a quantum Latin square can have, leaving just five small orders open.","key_machinery":"The engine is a family of parameterized unit vectors $u_n(x) = (f_1(x), \\ldots, f_{n-1}(x), g_n(x))$, where $f_k(x) = x^{(4k-1)/3}$. These vectors are used to construct unitary matrices that rotate a given quantum Latin square so that its entries avoid any prescribed finite set of vectors. The crucial supporting lemma asserts that $|(u_n(x), \\alpha)| = c$ has at most finitely many solutions in $x$, which relies on a claimed distinctness of the pairwise products $f_i(x)f_j(x)$. Around this, the paper develops intersection-controlled copies of quantum Latin squares (overlaps of size 0, 1, or 2), an extended tensor product operation $\\otimes_+$ and parameterized tensor products $\\otimes_r$ for the singular direct product constru","core_discovery":"The paper's central claim is Theorem 1.2: for every integer $v \\geq 8$ with $v \\notin \\{9, 11, 23\\}$, and for every integer $c$ in $[v, v^2]$ except $c = v+1$, there exists a quantum Latin square of order $v$ with cardinality $c$. The exclusion of $c = v+1$ is not accidental—earlier work shows that no quantum Latin square of order $u$ can have cardinality $u+1$. Thus the theorem asserts that, apart from this one structural obstruction, the entire cardinality range is filled. The proof is split by residue classes, producing QLS(4u), QLS(4u+1), QLS(4u+2), and QLS(4u+7) through extended direct and singular direct products, with finitely many exceptional orders handled in tables.","pith_inferences":["The false distinctness claim in Lemma 2.2(1) is a genuine gap in the proof as written, not a typo: the collision f_1 f_4 = f_2 f_3 = x^6 feeds directly into the coefficient comparisons of Lemma 2.3, which underpin the later disjoint-copy machinery.","A repair would likely require either a different parameterization of u_n(x) whose pairwise products are genuinely distinct, or a separate argument forcing equations (2.5)–(2.7) without relying on that distinctness.","Because the construction only needs finitely many x-values, the large-scale structure of the proof might survive a repair; a natural computational check would be to search for all cardinalities in [8, 64]\\{9} for order 8, the smallest order covered by the theorem."],"forward_implications":["The possible-cardinality problem for quantum Latin squares would be closed for all orders except 6, 7, 9, 11, and 23.","Every order of the form 4u would have a quantum Latin square for each cardinality in [4u, 16u²] except 4u+1.","The intersection-control lemmas provide a reusable way to force prescribed overlaps among many copies of a quantum Latin square, which is what powers the cardinality arithmetic.","The explicit tables for orders 6, 7, 9, 11, and 23 give concrete attainable sets and reduce the remaining uncertainties to short finite lists."],"fun_headline_variants":["Quantum Latin squares hit every cardinality from v to v², except v+1","All v≥8 but 9, 11, 23: quantum Latin squares fill cardinality range minus one","For v≥8 except 9,11,23, every cardinality v..v² except v+1 is achieved","Quantum Latin squares: full cardinality spectrum from v to v², missing only v+1","Almost every v≥8 yields quantum Latin squares of every size from v to v²"],"cache_read_input_tokens":25088,"weakest_assumption_plain":"The proof's engine is Lemma 2.2(1), the claim that products $f_i(x)f_j(x)$ of the parameter functions are pairwise distinct; this is false, since $f_1(x)f_4(x) = f_2(x)f_3(x) = x^6$, and the coefficient comparisons that power the rest of the construction depend on it.","fun_headline_variants_meta":{"raw":{"variants":["Quantum Latin squares hit every cardinality from v to v², except v+1","All v≥8 but 9, 11, 23: quantum Latin squares fill cardinality range minus one","For v≥8 except 9,11,23, every cardinality v..v² except v+1 is achieved","Quantum Latin squares: full cardinality spectrum from v to v², missing only v+1","Almost every v≥8 yields quantum Latin squares of every size from v to v²"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000154,"raw_usage":{"total_tokens":977,"prompt_tokens":603,"completion_tokens":374,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":347,"completion_tokens_details":{"reasoning_tokens":261}},"tokens_in":347,"tokens_out":374,"duration_ms":3888,"temperature":1.0,"reasoning_tokens":261,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T11:11:10.670527+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $f_1(x)f_4(x) = x^{(3+15)/3} = x^6$ and $f_2(x)f_3(x) = x^{(7+11)/3} = x^6$; this contradicts Lemma 2.2(1), on which Lemma 2.3's coefficient comparisons rest. Since those comparisons drive Lemmas 2.7–2.13 and the main constructions, the proof as written collapses at that point; the theorem could still be true, so a complementary test is to look for any cardinality in $[8,64] \\setminus \\{9\\}$ for $v=8$ that is not realized.","supporting_citations":[],"review_version":1}