{"id":"cd96230f-fba0-46cf-91e6-b38dd832d212","arxiv_id":"2608.01438","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A SAT-assisted construction proves that O_N-tile decompositions of hypercubes induce UPBs, yielding new UPB sizes 13 through 23 in C^3⊗C^3⊗C^3.","lead":"This paper builds a SAT-based pipeline that turns decompositions of a multidimensional grid into provably unextendible product bases, sets of orthogonal product quantum states that cannot be extended by any further product state. The pipeline produces previously unknown examples in small quantum systems, including every UPB size from 13 to 23 in three qutrits.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2's new UPB sizes in C^3⊗C^3⊗C^3 rest on unverified O_3-tile decompositions in Appendix B; direct enumeration of Definition 2 would settle it.","rationale":"The theoretical core (Theorem 1) is sound: the proof that any product state orthogonal to the constructed set yields a union of tiles that is a Cartesian product, and the level-set argument excluding the full-support case, are correct. The Section 4 verifier is also logically correct assuming Lemma 1, and would be a valid independent check of the final UPBs. The single load-bearing risk is the finite computational data: the new cardinalities 13–23 in Theorem 2 are derived only from the O_3-tile decompositions in Appendix B. These decompositions are explicit and checkable, so the risk can be eliminated by direct enumeration. The reader's weakest_assumption already identifies exactly this point, so I agree. No new objection is raised; the appropriate disposition remains conditional on the validity of the Appendix B data (or on the external verification certificates).","tokens_in":17911,"tokens_out":20808,"duration_ms":182022,"concrete_test":"Write a script that, for each Appendix B decomposition (s=5,...,15): (1) verifies the s tiles are pairwise disjoint and cover Z_3^3; (2) for every subset J with 2≤|J|≤s−1, forms U=∪_{j∈J} t_j and checks whether U equals A×B×C for some nonempty A,B,C⊆Z_3 by comparing U with the product of its coordinate projections. If all eleven decompositions pass, the O_3-tile condition is certified and the construction path of Theorem 2 is sound; if any fails, the corresponding UPB size is not established by Theorem 1.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 2 asserts UPBs of sizes 13–23 in C^3⊗C^3⊗C^3 solely through Theorem 1 plus the eleven O_3-tile decompositions listed in Appendix B (s=5,...,15). The load-bearing condition is that each listed family is a genuine O_3-tile decomposition: a disjoint cover of Z_3^3 by admissible tiles such that for every proper subset J with 1<|J|<s, ∪_{j∈J} t_j is not a Cartesian product. The paper argues that the SAT encoding enforces this exactly, but the appendix contains only the tile lists; no machine-checkable certificate or independent verification of the non-combinability condition is included in the text. A transcription error or solver error in any one decomposition would silently destroy the corresponding cardinality claim (s=5 gives size 23; s=15 gives size 13). The exact UPB verifier described in Section 4 and the promised Zenodo certificates could rescue the claims, but those certificates are external to the manuscript and are not part of the proof of Theorem 2 as written. Because the condition is finite and easy to check, this is a concrete, closable gap rather than a demonstrated failure.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript introduces a SAT-assisted method for constructing unextendible product bases (UPBs) from decompositions of the N-dimensional grid Z_{d1}×...×Z_{dN}. It defines O_N-tile decompositions and proves Theorem 1: any such decomposition with s tiles yields, via tile-wise Fourier product bases and a global stopper state, a UPB of cardinality ∏d_i − s + 1. It then encodes the search for O_3-tile decompositions of Z_3^3 as a SAT problem, implements an exact UPB verifier based on local orthogonality graphs and maximal unsaturated sets, and reports eleven explicit decompositions with s=5,...,15 tiles. Combining Theorem 1 with these lists, the paper claims UPBs of every size 13 through 23 in C^3⊗C^3⊗C^3, together with additional sizes in other tripartite and quadripartite systems. The data and code are deposited in Zenodo.","tokens_in":18184,"tokens_out":26234,"duration_ms":242380,"significance":"If the computational claims are correct, Theorem 1 is a clean and useful reduction: it converts a class of UPB existence problems into finite combinatorial tiling problems, and the SAT encoding is logically faithful to Definition 2. The proof of Theorem 1 is self-contained and rigorous, and the verification algorithm based on maximal unsaturated sets is a simple, exact tool. The reported new sizes 13–23 in C^3⊗C^3⊗C^3 would substantially improve the previously known set {7,19} for that system. The explicit appendix lists, the permanent data DOI, and the source code are concrete reproducibility assets. The main weakness is that the proof of Theorem 2 currently relies on external certificates for the O_3-tile condition, rather than on in-text machine-checkable verification of the Appendix B lists.","major_comments":[{"comment":"The new cardinalities 13–23 in C^3⊗C^3⊗C^3 are obtained by applying Theorem 1 to the eleven O_3-tile decompositions listed in Appendix B. The O_N condition in Definition 2 requires, for each s, that the union of every proper subset J with 2≤|J|≤s−1 is not a Cartesian product. The appendix lists only the tile sets; the paper does not contain a machine-checkable verification of this condition for any of the eleven lists. The sentence 'the SAT encoding produces O_3-tile decompositions' asserts correctness, and the certificates are relegated to the external Zenodo archive. A single transcription or solver error in one list would silently destroy the corresponding claimed UPB size. Since this is the load-bearing computational step, please include in the paper (or in a supplementary file that is part of the reviewed manuscript) a certificate or a short verification script that checks Definitio","section":"§5.1, Theorem 2; Appendix B"}],"minor_comments":[{"comment":"The phrase 'By Definition 2, |I|=1' is compressed. A reader must supply the argument that any other cardinality of I would already contradict Definition 2, and then that the complementary union of s−1 tiles yields the final contradiction. Consider spelling out this two-step reasoning explicitly.","section":"§2.3, Proof of Theorem 1"},{"comment":"The unmarked entries in Table 2 claim new UPB sizes in several tripartite and quadripartite systems, but only the C^3⊗C^3⊗C^3 decompositions are listed in Appendix B. The other unmarked entries are not proved in the text and are supported only by the external data archive. Please clarify which entries are theorems proved in the paper and which are data-supported claims verified in the Zenodo repository.","section":"§5.1, Table 2"},{"comment":"The quantity N_B counts pairs (T,P) with T a nontrivial Cartesian product and P∈T, not the number of Cartesian products T. The text says this immediately before the formula, but the notation could be misread; a brief sentence restating the counting convention would help.","section":"§3.2.3, Eq. (23)"},{"comment":"Each listed tile should be checked for admissibility (at least two proper coordinates) as part of the O_3 verification. This is implicit in Definition 2 but is not stated in the appendix; adding it to the verification certificate would make the check self-contained.","section":"Appendix B"}],"recommendation":"major_revision","confidential_remarks":"The central mathematical construction appears sound, and the proof of Theorem 1 is rigorous. My recommendation hinges on the completeness of the computational evidence for Theorem 2: the authors should provide in-article or bundled machine-checkable certificates for the Appendix B decompositions, rather than relying solely on the external Zenodo archive. This is a concrete, closable gap, not a demonstrated failure. I would not recommend rejection if the certificates are supplied."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a useful paper and I would not desk-reject it. The main theoretical result holds up, and the computational claim is credible, but the verification gap in Appendix B should be closed before publication.\n\nWhat's new: an N-dimensional tile-to-UPB theorem generalizing the bipartite O-tile idea, a SAT encoding for searching O_N-tile decompositions, an exact local-orthogonality-graph verifier, and explicit UPBs of sizes 13 through 23 in C^3⊗C^3⊗C^3. The central sizes are new relative to the cited literature. I went through the proof of Theorem 1 and it is sound: the support/product argument is clean, and the stopper state does what it needs to. The SAT encoding also looks logically exact, not heuristic.\n\nThe soft spot is in the computational layer. Theorem 2 depends on the tile decompositions in Appendix B, but those lists are not machine-verified in the manuscript. A SAT solver bug or a transcription error in any one of them would silently destroy the corresponding cardinality claim. The finite non-combinability condition is easy to check by direct enumeration, so this is a concrete, closable gap. The authors' Zenodo archive and exact verifier are the right response, but the paper should include an inline certificate or a small script so a reader can reproduce the check without trusting the data link. The verifier itself uses a lemma from an overlapping group; that is not circular, since the lemma is only for verification and not for constructing the claimed sizes, but a referee should look at it.\n\nOne smaller point: the claim that only sizes 7 and 19 were previously known in 3⊗3⊗3 should be double-checked against the full literature, including the Chinese-language survey they cite. It seems plausible, but the text alone doesn't let me confirm. Table 2 also lists many sizes in other systems; those are secondary and partly rest on recursive constructions, so the directly computed cases should be clearly separated.\n\nWho this is for: people working on UPBs, completely entangled subspaces, and bound entanglement will get real value from the construction and the transferable method. With the verification made explicit, I would accept it. As is, it deserves a serious referee.","headline":"Solid tile-to-UPB theorem with a believable SAT pipeline; the remaining work is making the Appendix B decompositions independently checkable.","tokens_in":18675,"tokens_out":3937,"would_cite":true,"duration_ms":39879,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A partition of a finite hypercube into non-combinable tiles always induces an unextendible product basis of known size.","keywords":["unextendible product bases","O_N-tile decomposition","hypercube decomposition","SAT encoding","completely entangled subspaces","bound entanglement","orthogonality graphs","tile-to-UPB theorem"],"falsifier":"Enumerate every subset $J$ of tiles in each Appendix B decomposition with $2\\le |J|\\le s-1$ and test whether $\\bigcup_{j\\in J} t_j$ is a Cartesian product $A\\times B\\times C$; if any such union exists, the corresponding claimed UPB size is not established by Theorem 1. As an independent check, the paper's exact verification algorithm, run on the supplied symbolic product states, would return False for any set that is not actually a UPB.","tokens_in":17813,"feed_emoji":"🧩","tokens_out":8966,"duration_ms":80956,"temperature":0.7,"pith_summary":"This paper bridges discrete geometry and quantum information: a partition of the hypercube $Z_{d_1}\\times\\cdots\\times Z_{d_N}$ into Cartesian-product tiles, called an $O_N$-tile decomposition, is shown to induce an unextendible product basis (UPB) in the multipartite Hilbert space $\\mathbb{C}^{d_1}\\otimes\\cdots\\otimes \\mathbb{C}^{d_N}$. The induced UPB has cardinality $d_1\\cdots d_N - s + 1$, where $s$ is the number of tiles; it is assembled from Fourier product bases on each tile plus a single global stopper state. Since UPBs are the standard raw material for completely entangled subspaces and bound entangled states, the question of which cardinalities occur is long-standing. The paper encodes the search for such tilings as a Boolean satisfiability (SAT) problem and, with an exact verification algorithm, constructs UPBs of every size 13 through 23 in $\\mathbb{C}^3\\otimes\\mathbb{C}^3\\otimes\\mathbb{C}^3$, where previously only sizes 7 and 19 were known. The small instances also act as seeds for recursive constructions in larger systems.","feed_headline":"Cube tilings yield unextendible quantum bases of sizes 13-23","feed_subtitle":"SAT-found tilings of the 3-cube supply UPBs in C^3⊗C^3⊗C^3 and seed larger multipartite systems.","key_machinery":"The load-bearing object is the $O_N$-tile decomposition: a partition of the $N$-cube $C=Z_{d_1}\\times\\cdots\\times Z_{d_N}$ into tiles $t_j=R_1^{(j)}\\times\\cdots\\times R_N^{(j)}$ with the non-combinability condition that every proper sub-union of tiles (with between 2 and $s-1$ members) fails to be a Cartesian product. The construction assigns each tile a Fourier product basis of its supported subspace, removes the all-zero Fourier vector $|\\eta_j\\rangle$ from each tile, and appends the global stopper $|S\\rangle$; the stopper overlaps a tile-wise Fourier product state only when all local Fourier frequencies are zero. The non-combinability condition forces any product state orthogonal to the c","core_discovery":"The paper's central claim is Theorem 1 (tile-to-UPB): if $C=Z_{d_1}\\times\\cdots\\times Z_{d_N}$ is a disjoint union of $s\\ge 3$ Cartesian-product tiles $t_j$ such that no union of a proper subset of tiles with at least two members is itself a Cartesian product, then the set $U$ built from tile-wise Fourier product bases, deleting each tile's all-zero Fourier vector and adding the global stopper $|S\\rangle=\\bigotimes_i\\sum_{r\\in Z_{d_i}}|r\\rangle$, is a UPB of cardinality $\\prod_i d_i - s + 1$. Theorem 2 is the computational companion: a SAT search found $O_3$-tile decompositions of $Z_3^3$ with $s=5,\\ldots,15$ tiles, producing UPBs of every size $13,\\ldots,23$ in $\\mathbb{C}^3\\otimes\\mathbb{C","pith_inferences":["The $O_N$-tile condition is sufficient but likely not necessary for a UPB: product bases built from other mechanisms may realize sizes that no tiling realizes, so a failed SAT search should not be read as non-existence of a UPB.","Extending the same SAT pipeline to larger cubes (e.g. $Z_4^3$ or $Z_3^4$) could enumerate further attainable sizes, with the main bottleneck being the number of non-combinability clauses.","The maximal-unsaturated-set enumeration in the verifier suggests a compact certificate format: a small family of MUSs proving that no $N$-tuple of unsaturated sets covers all indices.","Because these UPBs have a highly structured form—Fourier product bases per tile plus one stopper—they may also serve as natural probes for local distinguishability and strong nonlocality, though the paper only gestures toward recursive applications."],"forward_implications":["UPBs of every size between 13 and 23 exist in $\\mathbb{C}^3\\otimes\\mathbb{C}^3\\otimes\\mathbb{C}^3$, including the previously known size 19; sizes 8–12 remain open.","Every such UPB yields, via its orthogonal complement, a completely entangled subspace, and the normalized projector onto that complement gives a bound entangled state.","Combined with the direct-sum lemma cited as [24], the small instances seed infinite families: for $d=2x+3y$ and $a\\in A$, $b\\in B$, a UPB of size $ax+by$ exists in $\\mathbb{C}^3\\otimes\\mathbb{C}^3\\otimes\\mathbb{C}^d$.","The verification algorithm is exact for arbitrary finite sets of multipartite product states, not only tile-induced sets, and its running time stays under a second on the benchmarked instances.","Solving the same SAT formulation for other cubes would automatically produce UPBs of the corresponding sizes, making the pipeline a general tool for the prescribed-size UPB problem."],"supporting_citations":[{"why":"Supplies the definition of UPBs, the standard shifts example, and the bound-entanglement motivation that the tile-to-UPB construction serves.","marker":"[14]"},{"why":"Introduces the original tile decomposition for the 2-cube and the nonlocality-without-entanglement application that motivates hypercube decompositions.","marker":"[20]"},{"why":"Formalizes the U-tile correspondence for bipartite systems, the direct precursor of the N-dimensional tile-to-UPB theorem.","marker":"[27]"},{"why":"Introduces O-tiles for 2-cube decompositions, the bipartite condition that this paper generalizes to O_N-tiles.","marker":"[28]"},{"why":"Establishes the Alon–Lovász lower bound and the orthogonality-graph framework used for the k=7 case and for verification context.","marker":"[23]"},{"why":"Provides the graph-theoretic UPB criterion (Lemma 1) on which the paper's exact verification algorithm is based.","marker":"[33]"},{"why":"Supplies the direct-sum lemma and 1-factorization methods that extend the new small instances to larger multipartite systems.","marker":"[24]"},{"why":"Records the previously known size-19 UPB in C^3⊗C^3⊗C^3 that the new 13–23 family extends.","marker":"[26]"}],"fun_headline_variants":["SAT tiles produce quantum UPBs of sizes 13-23","Cube tilings yield UPBs in C^3⊗C^3⊗C^3","Hypercube tiles give unextendible bases, sizes 13-23","Tiling to UPB: SAT search finds bases 13-23","Automated tiling constructs unextendible product bases"],"cache_read_input_tokens":2816,"weakest_assumption_plain":"The claimed sizes 13–23 in $\\mathbb{C}^3\\otimes\\mathbb{C}^3\\otimes\\mathbb{C}^3$ rest on the Appendix B tile lists being genuine $O_3$-tile decompositions; the paper depends on the SAT solver's output and its transcription into the appendix, without machine-checkable certificates for the non-combinability condition.","fun_headline_variants_meta":{"raw":{"variants":["SAT tiles produce quantum UPBs of sizes 13-23","Cube tilings yield UPBs in C^3⊗C^3⊗C^3","Hypercube tiles give unextendible bases, sizes 13-23","Tiling to UPB: SAT search finds bases 13-23","Automated tiling constructs unextendible product bases"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00066,"raw_usage":{"total_tokens":2925,"prompt_tokens":886,"completion_tokens":2039,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":630,"completion_tokens_details":{"reasoning_tokens":1943}},"tokens_in":630,"tokens_out":2039,"duration_ms":15365,"temperature":1.0,"reasoning_tokens":1943,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T00:12:44.578617+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Enumerate every subset $J$ of tiles in each Appendix B decomposition with $2\\le |J|\\le s-1$ and test whether $\\bigcup_{j\\in J} t_j$ is a Cartesian product $A\\times B\\times C$; if any such union exists, the corresponding claimed UPB size is not established by Theorem 1. As an independent check, the paper's exact verification algorithm, run on the supplied symbolic product states, would return False for any set that is not actually a UPB.","supporting_citations":[{"cited_title":"Unextendible product bases and bound entanglement.Physical Review Letters, 82(26):5385, 1999","cited_arxiv_id":null,"evidence_quote":"Supplies the definition of UPBs, the standard shifts example, and the bound-entanglement motivation that the tile-to-UPB construction serves."},{"cited_title":"Quantum nonlocality without entanglement.Physical Review A, 59(2):1070, 1999","cited_arxiv_id":null,"evidence_quote":"Introduces the original tile decomposition for the 2-cube and the nonlocality-without-entanglement application that motivates hypercube decompositions."},{"cited_title":"Unextendible product bases from tile structures and their local entanglement-assisted distinguishability.Physical Review A, 101(6):062329, 2020","cited_arxiv_id":null,"evidence_quote":"Formalizes the U-tile correspondence for bipartite systems, the direct precursor of the N-dimensional tile-to-UPB theorem."},{"cited_title":"Unextendible product bases from tile structures in bipartite systems.Journal of Physics A: Mathematical and Theoretical, 56(1):015303, 2023","cited_arxiv_id":null,"evidence_quote":"Introduces O-tiles for 2-cube decompositions, the bipartite condition that this paper generalizes to O_N-tiles."},{"cited_title":"Unextendible product bases.Journal of Combinatorial Theory","cited_arxiv_id":null,"evidence_quote":"Establishes the Alon–Lovász lower bound and the orthogonality-graph framework used for the k=7 case and for verification context."},{"cited_title":"Graph-theoretic characterization of unextendible product bases.Physical Review Research, 5(3):033144, 2023","cited_arxiv_id":null,"evidence_quote":"Provides the graph-theoretic UPB criterion (Lemma 1) on which the paper's exact verification algorithm is based."},{"cited_title":"Unextendible product bases and 1-factorization of complete graphs.Discrete Applied Mathematics, 154(6):942–949, 2006","cited_arxiv_id":null,"evidence_quote":"Supplies the direct-sum lemma and 1-factorization methods that extend the new small instances to larger multipartite systems."},{"cited_title":"Genuinely entangled subspace with all-encompassing distillable entanglement across every bipartition.Physical Review A, 99:032335, 2019","cited_arxiv_id":null,"evidence_quote":"Records the previously known size-19 UPB in C^3⊗C^3⊗C^3 that the new 13–23 family extends."}],"review_version":1}