{"id":"97791696-c1b1-4bd9-a6bf-f181cb13caaf","arxiv_id":"2507.05642","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For every m>=2, quantum Latin squares of order 4m exist with every cardinality in [4m,16m^2] except 4m+1.","lead":"This paper constructs quantum Latin squares of order 4m, for every integer m at least 2, that use every possible number of distinct vectors between 4m and (4m)^2, except the forbidden 4m+1. The construction assembles smaller 4x4 quantum Latin squares in a Latin-square pattern and carefully counts overlaps.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof hinges on unverified exact-overlap counts (P1)-(P4) and analogous 'easy to check' disjointness claims; if any count is off, the cardinality intervals for Lemma 3.2 and Theorem 1.1 collapse.","rationale":"The reader's weakest-assumption identification is the same one I would make: the proof of Lemma 3.2 and the second half of Theorem 1.1 depend crucially on the exact overlap counts (P1)-(P4) and on disjointness assertions for the W_{2i+3,2i+4} blocks, none of which are derived in the text. These are finite, in-principle-checkable claims, so the appropriate status is CONDITIONAL rather than REJECT: the construction is explicit and the missing pieces are verifiable, not conceptual impossibilities. I also concur with the reader that the t=25 exclusion in the final union is a typo; correcting it makes the stated union match the theorem, and the separate explicit QLS(12) covers the m=3 case. I do not see a deeper structural flaw: the tensor-product block construction is valid, Lemma 2.3 is sound (tensor products of two 2-dimensional vector sets are distinct when the parameter sets are disjoint), and the irrationality of W5,6 relative to W0,...,W4 and H'ℓ supports the disjointness claims. The paper would be strengthened by including exact verification code or a short derivation of the overlap counts, but the central claim is not contradicted by anything in the manuscript. Hence the reader's CONDITIONAL verdict should stand unchanged.","tokens_in":11971,"tokens_out":29036,"duration_ms":292356,"concrete_test":"Write a short exact-arithmetic script (Python Fraction or a CAS) to construct X_i, J_k, W_k (k=1,...,4), and H'ℓ, and compute |W0∩W1|, |W0∩W2|, |W0∩W3|, |W2∩W4|, checking they are 1, 4, 2, 6 respectively. Then for m=3 and m=4, verify symbolically that W5,6, W7,8, ... have 16(m−1) entries disjoint from all of W0,...,W4 and H'ℓ, and recompute the cardinality of the displayed QLS(12) from the end of Theorem 1.1 (it should be 105). If all counts match, the missing verification is supplied and the theorem's construction stands.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction's high-cardinality cases depend on exact intersection data that the paper asserts without proof. After Equation (9), the claims (P1)-(P4) give |W0∩W1|=1, |W0∩W2|=4, |W0∩W3|=2, |W2∩W4|=6, and on these rest the QLS(8) cardinalities 49-56 and 58-64 in Lemma 3.2, the special 57-construction, and the second range of Theorem 1.1. The proof of Theorem 1.1 additionally asserts, again without derivation, that the 16(m−1) entries of W_{2i+3,2i+4} (i=1,...,m−1) are all distinct from every W0,W1,W2,W3,W4 and from every H'ℓ. These are load-bearing finite computational facts, not mere convenience: if any overlap count were wrong by one, say |W0∩W2|=5 instead of 4, then W2 would contribute 11 new elements rather than 12, shifting the achievable cardinality set and potentially leaving a gap in [49,64] or in [16m^2−16m,16m^2]. The range-union typo including t=25 is real but is an arithmetic slip: removing 25 from the exclusion list repairs the union, and the explicit m=3 QLS(12) plausibly fills the remaining 105 hole. The structural vulnerability is therefore the missing verification of the overlap counts, not the union arithmetic. The matrices involved have rational entries (and the W_{2i+3,2i+4} entries are irrational relative to the rational W_j and H'ℓ), so these counts can in principle be settled exactly.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies the set of possible cardinalities of quantum Latin squares (QLS) of order 4m. The main result, Theorem 1.1, states that for every integer m≥2 and every integer c in [4m,16m²]\\{4m+1}, there exists a QLS(4m) with exactly c distinct vectors. The proof is constructive: it builds QLS(8) with all cardinalities in [8,64]\\{9} using carefully selected 4×4 blocks with controlled overlap (Lemmas 2.1, 3.2), and then obtains QLS(4m) by 'blowing up' a classical Latin square of order m with these 4×4 blocks (Theorem 1.1). The paper also proves the impossibility of cardinality n+1 for any order n (Lemma 3.1).","tokens_in":12355,"tokens_out":23794,"duration_ms":216112,"significance":"The result, if fully established, would determine the complete set of possible cardinalities for all QLS of order 4m (m≥2), since n+1 is provably impossible and the theorem covers all other values. The construction is explicit and the paper is generally well organized. The main theorem is attractive and would settle a natural question in the theory of quantum Latin squares. However, the proof as written contains several unproved computational overlap assertions (P1)-(P4) and 'easy to check' distinctness claims, plus an arithmetic error in the final range union; these must be repaired before the result can be considered rigorous. The manuscript does provide explicit matrices and formulas, so the missing checks are verifiable in principle.","major_comments":[{"comment":"The claims (P1)-(P4) are stated as \"It has been checked\" with no derivation. These exact intersection sizes are load-bearing: they lead to the existence of QLS(4)s with 12, 14, 15 new elements relative to W0, to the cardinality-57 construction, and to the second range in Theorem 1.1. Note also that (P1) must be read as exactly one common element (|11⟩) for the cardinality calculations to work, although the text writes only \"have a common element\". Please supply a rigorous case check, e.g., an appendix listing the 16 vectors of each W_i and the intersections, or a verifiable computer algebra script.","section":"Section 3, after Eq. (9)"},{"comment":"The assertions that W_{5,6}, W_{7,8}, and the family W_{2i+3,2i+4} have all their elements distinct from the reference blocks (H_ℓ, and later W_0,...,W_4, H'_ℓ) are essential for the counting formulas, but are backed only by \"it is easy to check\" and a one-line argument in Lemma 2.3. Since the parameters a=2k−1, b=2k produce entries with square-root factors, distinctness is not immediate from coefficient inspection; a complete proof or an explicit verification for all relevant k is needed. Without it, the cardinality formulas in the two ranges of Theorem 1.1 lack support.","section":"Section 3, Lemma 3.2 and Theorem 1.1"},{"comment":"The union computation contains an extra exclusion t=25: the second range was derived with missing values t∈{1,3,5,7,9,11,13}, yet the final union subtracts 16m²−16m+t also for t=25. Taken literally, this would leave c=121 (m=3) and c=217 (m=4) uncovered, even though the second-range construction achieves those values (for m=3, s=25 is 15+8+2 from the allowed set {0,2,4,6,8,12,14,15,16}). Removing 25 from the exclusion list repairs the union; for m=3 the explicit QLS(12) with cardinality 105 fills the remaining gap. This correction is necessary for the proof of Theorem 1.1.","section":"Section 3, proof of Theorem 1.1, final paragraph"}],"minor_comments":[{"comment":"Lemma 2.2 is cited from the preprint [12] without proof; since it is used to construct W0 and Wa,b, a short proof or attribution to a published source would improve self-containedness.","section":"Section 2, Lemma 2.2"},{"comment":"Equation (1) is difficult to read in the current typesetting; the matrix entries should be separated more clearly.","section":"Section 2, Eq. (1)"},{"comment":"The phrase \"there are two QSL(4)s\" should be \"QLS(4)s\"; similar \"QSL\" typos appear elsewhere.","section":"Section 3, Lemma 3.2"},{"comment":"The statement that W5,6 has 16 elements not in W0 \"since the coefficients ... are irrational\" is too terse; please spell out why no rational vector from W0 can coincide with an irrational-coefficient vector from W5,6.","section":"Section 3, Lemma 3.2"},{"comment":"The notation X_{0,j} is used both for the actual block chosen in row 0 and as an option for later rows; this is confusing and should be renamed.","section":"Section 3, Theorem 1.1"},{"comment":"The proof of Lemma 2.3 says \"since k−t≠0, 1+kt≠0 and 1−kt≠0, all 32 elements ... are distinct\"; this is not a complete argument and should be expanded.","section":"Section 2, Lemma 2.3"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within scope for a quantum information or combinatorics journal. The main theorem is interesting and likely correct, but the proof needs the noted repairs: the overlap counts (P1)-(P4) must be rigorously verified, the distinctness claims for the W_{2i+3,2i+4} families need justification, and the t=25 exclusion in the final union must be removed. These are all fixable within the manuscript's scope, so I recommend major revision rather than rejection. The self-citation to [12] is mild and can be resolved by adding a proof of Lemma 2.2."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know: this paper settles the cardinality spectrum for quantum Latin squares of order 4m for all m≥2, leaving out only 4m+1, which is provably impossible. That is a genuine result, not a small extension. The construction is explicit: they build sub-QLS(4)s with controlled overlaps and arrange them in a Latin square of order m, then count the union. I spot-checked several small cardinalities (8, 57, 105) and the formulas work out. The method is new, and the main theorem is a clean completion for an infinite family.\n\nWhat's good: the counting is transparent, the matrices are fully written out, and the hard part—making intermediate cardinalities like 14 and 57—is handled by explicit QLS(4)s with known intersection sizes. Lemma 2.2 from the co-author's preprint is only used as a black-box tensor product, so the self-citation is not a real problem.\n\nThe soft spots are real but fixable. The proof leans on several 'it has been checked' claims: the intersection sizes (P1)–(P4) between W0 and W1..W4, and the disjointness of the W_{2i+3,2i+4} blocks from the Hℓ and W_i. These are load-bearing. If any count were off, the intervals in Lemma 3.2 and Theorem 1.1 would shift. The matrices are all explicit rational numbers, so the checks are finite and verifiable, but the paper doesn't show them or provide code. A referee should ask for an appendix or a short derivation. The disjointness of the W_{a,b} blocks can be argued by rationality (their entries are irrational in the computational basis, while Hℓ and W_i are rational), but the authors didn't say that. There is also a clear typo in the final range-union: the exclusion set includes t=25, which would leave a hole at 16m^2−16m+25; removing 25 fixes it. For m=3, the written union leaves out 105, but the explicit QLS(12) on the next line fills it—they just need to fold that into the statement.\n\nBottom line: the result is likely correct and worth publishing. The construction is the first complete cardinality spectrum for an infinite family beyond order 4. It deserves a serious referee, but the referee should insist on a verified computation for (P1)–(P4) and a corrected range argument. I'd send it to review rather than desk-reject, and I'd be comfortable citing it once the computational gaps are closed.","headline":"A real result—full cardinality spectrum for QLS(4m)—with a fixable gap: the proof leans on unverified overlap counts and a range-union typo.","tokens_in":12895,"tokens_out":17390,"would_cite":true,"duration_ms":183233,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05B15","81P68"],"pacs":[],"model":"deepseek-v4-flash","headline":"For every m ≥ 2 and every integer c from 4m to (4m)^2, except 4m+1, some quantum Latin square of order 4m has exactly c distinct vectors.","keywords":["quantum Latin square","cardinality","sub-QLS(4)","orthonormal basis","tensor product construction","Latin square","order 4m","unitary error bases"],"falsifier":"Perform an exact arithmetic check of the 16-vector sets defining $W_0,W_1,W_2,W_3,W_4$ and $H_\\ell$ for $\\ell = 0,\\dots,8$, counting pairs of vectors equal up to global phase; the asserted counts are (P1)–(P4) and the stated numbers of new elements for each $H_\\ell$. If any of these finite intersection counts differs, the cardinality formulas for order 8 and the upper range of Theorem 1.1 collapse.","tokens_in":11728,"feed_emoji":"🧩","tokens_out":10497,"duration_ms":111063,"temperature":0.7,"pith_summary":"The paper proves that for every m ≥ 2 the cardinality spectrum of quantum Latin squares of order 4m is exactly the interval $[4m,16m^2]$ with the single value $4m+1$ removed. A quantum Latin square is an $n \\times n$ array of unit vectors whose rows and columns are orthonormal bases, and its cardinality is the number of distinct vectors (up to global phase), so the result says every admissible size of “quantum content” occurs for these orders. The proof is a block construction: entries of a classical Latin square of order $m$ are replaced by carefully chosen $4\\times 4$ quantum Latin squares, and the total number of distinct vectors is controlled by summing the numbers of new vectors that each block contributes. Since the impossibility of cardinality $n+1$ is already known, this closes the question for all orders divisible by 4.","feed_headline":"For every order 4m, every cardinality except 4m+1 is realized","feed_subtitle":"The result pins down exactly which sizes a quantum Latin square can have, for every order divisible by four.","key_machinery":"The carrying object is the sub-QLS$(4)$: a $4\\times 4$ quantum Latin square in $\\mathcal{H}_2^{\\otimes 2}$ used as a building block. The argument depends on two families of such blocks: the squares $H_0,\\dots,H_8$, whose intersections with fixed squares are exactly $\\ell \\in \\{2,\\dots,8\\}$ elements, and the maximal-cardinality squares $W_0,\\dots,W_4$ and $W_{2k-1,2k}$, whose pairwise intersections are listed in assertions (P1)–(P4). The key counting mechanism is the tensor-product substitution $|a_{i,j}\\rangle \\otimes X_{i,j}$: since different $|k\\rangle$ factors live in orthogonal subspaces, the total cardinality is a sum over blocks, and choosing blocks with controlled overlaps turns the range of attainable sums into the interval $[4m,16m^2] \\setminus \\{4m+1\\}$.","core_discovery":"The central claim, Theorem 1.1, is that for any integer $m \\geq 2$ and any $c \\in [4m,16m^2] \\setminus \\{4m+1\\}$, there exists a QLS$(4m)$ in $\\mathcal{H}_m \\otimes \\mathcal{H}_2^{\\otimes 2}$ with cardinality $c$. The construction starts from the classical Latin square $a_{i,j} = j-i \\bmod m$ and replaces each entry $|a_{i,j}\\rangle$ by $|a_{i,j}\\rangle \\otimes X_{i,j}$, where $X_{i,j}$ is one of a finite list of explicit QLS$(4)$s. Because the subspaces $|k\\rangle \\otimes \\mathcal{H}_2^{\\otimes 2}$ are mutually orthogonal, the cardinality of the resulting $4m \\times 4m$ array is the sum of the cardinalities of its $m^2$ blocks, provided the blocks have no unintended shared vectors. The blocks drawn from $H_0,\\dots,H_8$ (Lemma 2.1) and $W_0,\\dots,W_4$ (Equation (9)) are chosen so that each contributes a prescribed number of new vectors; Lemma 3.2 carries out the count for order 8, and the same blocks are arranged along the diagonals of the order-$m$ Latin square to cover the whole range. Lemma 3.1's prohibition of cardinality $n+1$ makes the exclusion $4m+1$ necessary, so the theorem gives the complete possible set.","pith_inferences":["For composite orders other than multiples of 4, the same block-substitution strategy would yield the cardinality spectrum if one can find a $q\\times q$ quantum Latin square with controlled overlaps between its translates; the paper's conclusion says this is precisely the difficulty.","Assertions (P1)–(P4) are finite exact claims about explicit vectors, so a short computer verification or an algebraic proof would turn the 'it has been checked' step into a fully explicit part of the argument.","The final union display deletes $c = 16m^2 - 16m + 25$ from the second interval; taken literally this would omit $c = 121$ for $m=3$ and $c=217$ for $m=4$, although the paper's own sum formulas realize those values, so the deletion looks like a typographical slip.","Because every block is written with real coordinates, the construction shows the same cardinality range is attainable if one restricts to real Hilbert-space quantum Latin squares."],"forward_implications":["For $n = 4m$, the set of attainable cardinalities is exactly $[n,n^2] \\setminus \\{n+1\\}$; no other gaps exist for these orders.","All constructed squares live in $\\mathcal{H}_m \\otimes \\mathcal{H}_2^{\\otimes 2}$, so the full cardinality spectrum is realized using only one $m$-dimensional space tensored with two qubit spaces.","Because Lemma 3.1 rules out $n+1$ for every order $n$, the result is the strongest possible statement for orders divisible by 4.","The block-substitution method yields concrete explicit QLS$(4m)$s for each cardinality, not merely an existence proof by counting arguments."],"supporting_citations":[{"why":"Introduces quantum Latin squares and their connection to unitary error bases, giving the object studied.","marker":"[5]"},{"why":"Defines cardinality for QLS and establishes the order-4 possible cardinalities that anchor the block construction.","marker":"[8]"},{"why":"Supplies Lemma 2.2, the tensor-product lemma giving QLS(mn) with cardinality c1c2, and the near-complete maximal-cardinality result that the paper complements.","marker":"[12]"},{"why":"Provides the standard global-phase identification of unit vectors used to count distinct entries.","marker":"[7]"}],"fun_headline_variants":["All but one cardinality for quantum Latin squares of order 4m","Quantum Latin squares realize every cardinality except 4m+1","For order divisible by 4, all sizes but one are possible","Quantum Latin squares of order 4m: all cardinalities except 4m+1","Every order-4m quantum Latin square size exists, except 4m+1"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof relies on unproved computational overlap statements, in particular (P1)–(P4) after Equation (9), which assert that $W_0,W_1,W_2,W_3,W_4$ share exactly the specified numbers of entries and that the other $W$-blocks introduce no accidental overlaps; if any of those intersection counts is incorrect, the block-sum cardinality formulas in Lemma 3.2 and Theorem 1.1 could fail.","fun_headline_variants_meta":{"raw":{"variants":["All but one cardinality for quantum Latin squares of order 4m","Quantum Latin squares realize every cardinality except 4m+1","For order divisible by 4, all sizes but one are possible","Quantum Latin squares of order 4m: all cardinalities except 4m+1","Every order-4m quantum Latin square size exists, except 4m+1"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000903,"raw_usage":{"total_tokens":3938,"prompt_tokens":1051,"completion_tokens":2887,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":667,"completion_tokens_details":{"reasoning_tokens":2786}},"tokens_in":667,"tokens_out":2887,"duration_ms":22848,"temperature":1.0,"reasoning_tokens":2786,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:26:11.499079+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Perform an exact arithmetic check of the 16-vector sets defining $W_0,W_1,W_2,W_3,W_4$ and $H_\\ell$ for $\\ell = 0,\\dots,8$, counting pairs of vectors equal up to global phase; the asserted counts are (P1)–(P4) and the stated numbers of new elements for each $H_\\ell$. If any of these finite intersection counts differs, the cardinality formulas for order 8 and the upper range of Theorem 1.1 collapse.","supporting_citations":[{"cited_title":"Musto and J","cited_arxiv_id":null,"evidence_quote":"Introduces quantum Latin squares and their connection to unitary error bases, giving the object studied."},{"cited_title":"Paczos, M","cited_arxiv_id":null,"evidence_quote":"Defines cardinality for QLS and establishes the order-4 possible cardinalities that anchor the block construction."},{"cited_title":"Zhang and H","cited_arxiv_id":null,"evidence_quote":"Supplies Lemma 2.2, the tensor-product lemma giving QLS(mn) with cardinality c1c2, and the near-complete maximal-cardinality result that the paper complements."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the standard global-phase identification of unit vectors used to count distinct entries."}],"review_version":1}