{"id":"57b93bf6-3e10-4978-914c-8f9155d3dc67","arxiv_id":"2506.19792","paper_version":2,"verdict":"REJECT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper's claimed collapses of quantum-classical PCPs and the quantum polynomial hierarchy rest on invalid reductions, so the main theorems are unsupported.","lead":"A quantum complexity paper claims to prove three collapse results: unique proofs do not weaken quantum-classical PCPs, quantum advice collapses the quantum polynomial hierarchy, and a new entanglement-bounded hierarchy collapses. The proofs contain a demonstrably false reduction and an unjustified collapse step, so the central results are not established.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2.9 is false: a clause's monomial product is 1 only when all its literals are true, so satisfying assignments can map to UniqueMTP NO instances, breaking the UniqueQCPCP = QCPCP reduction chain.","rationale":"The reader's weakest assumption identifies exactly the load-bearing error. My independent check confirms that the polynomial in Lemma 2.9 computes the AND of each clause's literals, not the OR, so a satisfied clause containing a false literal contributes 0. The formula φ=(x1∨x2)∧(¬x1) is a decisive counterexample: it lies in UniqueSAT with unique assignment (0,1), but the constructed P evaluates to 1 < m=2, making the UniqueMTP instance a NO instance. This invalidates the central reduction chain and the paper's first headline result. I also note independent serious gaps in Section 3.2: the Karp–Lipton proof defines P_x as a QMA problem with quantum inputs (ρ1,ρ2), a model not defined and not covered by the assumption QMA⊆BQP/qpoly, and it invokes QΣ2=QΠ2 without proof. These reinforce rejection but are not needed for this verdict. The reader's REJECT verdict is therefore unchanged.","tokens_in":18159,"tokens_out":8223,"duration_ms":86396,"concrete_test":"Evaluate Lemma 2.9 on φ=(x1∨x2)∧(¬x1): compute P(0,1)=x1·x2+(1−x1)=1 while m=2 and the unique satisfying assignment is (0,1). If P(0,1)<m, the UniqueSAT YES instance is mapped to a UniqueMTP NO instance, refuting the lemma and the reduction chain of Theorem 2.11.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 2.9 is the UniqueSAT→UniqueMTP link in the reduction chain (2.13), and it is mathematically false. The construction P(x)=Σ_{j=1}^m ∏_{ℓ∈C_j} x̃_ℓ, with x̃_ℓ=x_i for ℓ=x_i and x̃_ℓ=1−x_i for ℓ=¬x_i, gives a clause contribution of 1 exactly when every literal in the clause is true. A clause in a CNF formula is satisfied when at least one literal is true, so a satisfying assignment can have P(x)<m. Example: φ=(x1∨x2)∧(¬x1) has unique satisfying assignment (0,1), yet P(0,1)=x1·x2+(1−x1)=0+1=1<2=m. Thus a YES instance of UniqueSAT is mapped to a NO instance of the constructed UniqueMTP instance. Since Theorem 2.11 composes QCPCP≤BQ MTP≤p SAT≤RP UniqueSAT≤p UniqueMTP≤BQ UniqueQCPCP, the claimed collapse UniqueQCPCP=QCPCP rests on this broken step. The proof's assertion that a satisfied clause makes the product 1 is simply incorrect.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims three main results. First, it argues that restricting quantum-classical probabilistically checkable proofs to unique accepting proofs does not reduce power: UniqueQCPCP = QCPCP under BQ-operator and randomized reductions, via the chain QCPCP ≤ BQ MTP ≤ p SAT ≤ RP UniqueSAT ≤ p UniqueMTP ≤ BQ UniqueQCPCP. Second, it claims a non-uniform quantum Karp-Lipton theorem: if QMA ⊆ BQP/qpoly, then QPH ⊆ QΣ2/qpoly. Third, it introduces bounded-entanglement and separable variants BEQPH and SepQPH, claims that each level reduces to a two-round convex optimization problem, and that BEQPH (and SepQPH) collapse above the fourth level. The paper is clearly structured and credits prior work on QPCP reductions, the Valiant-Vazirani theorem, and the collapse of QEPH, but the central arguments contain serious technical flaws.","tokens_in":18504,"tokens_out":9077,"duration_ms":98355,"significance":"If correct, the results would be substantial: a uniqueness collapse for quantum-classical PCPs, a non-uniform Karp-Lipton statement with quantum advice, and a new family of tractable hierarchy variants. The paper also has useful expository elements, such as the explicit reduction-chain diagram and the separation of convexity-based collapse mechanisms from entanglement-based ones. However, the main claims are not established. The reduction in Lemma 2.9 is mathematically incorrect, the Karp-Lipton proof applies QMA to promise problems with quantum inputs, and the proof invokes QΣ2 = QΠ2 without justification. The BEQPH collapse argument also uses a minimax swap that does not follow from Sion's theorem. Since these are load-bearing defects in all three advertised contributions, the paper cannot be accepted in its present form.","major_comments":[{"comment":"Lemma 2.9 is false. The polynomial P(x) = Σ_j ∏_{ℓ∈C_j} x̃_ℓ, with x̃_ℓ = x_i for ℓ = x_i and x̃_ℓ = 1−x_i for ℓ = ¬x_i, evaluates to 1 on a clause only when every literal in that clause is true, not when the clause is merely satisfied. A satisfying assignment can therefore have P(x) < m. For example, φ = (x1∨x2) ∧ (¬x1) has the unique satisfying assignment (0,1), but P(0,1) = x1x2 + (1−x1) = 0 + 1 = 1 < 2 = m. Thus this UniqueSAT YES instance is mapped to a UniqueMTP NO instance. Since Lemma 2.9 is the UniqueSAT-to-UniqueMTP link in the composition (2.13) used in Theorem 2.11, the claimed collapse UniqueQCPCP = QCPCP is unsupported.","section":"Section 2.2, Lemma 2.9, Eqs. (2.6)-(2.7) and Eq. (2.13)"},{"comment":"The base case of the Karp-Lipton proof defines a promise problem P_x whose inputs are the quantum states (ρ1,ρ2), and then asserts 'By construction, P_x ∈ QMA.' This is not standard QMA: QMA is a class of promise problems over classical bit strings, with a quantum witness supplied to the verifier, not a class of problems whose instance itself is a quantum state. A circuit receiving x, ρ1, ρ2, and an advice state is not a QMA verifier for a classical language. The same issue appears in the inductive step at Eqs. (3.16)-(3.18), where the supposed language A' has inputs (x,ρ1,ρ2) containing quantum states. The assumption QMA ⊆ BQP/qpoly therefore cannot be applied as written.","section":"Section 3.2, Theorem 3.10, Eqs. (3.9)-(3.12)"},{"comment":"The proof uses the equalities QΣ2 = QΠ2 and QΠ2/qpoly = QΣ2/qpoly as if they were available. The paper does not prove these equalities, and they are not known consequences of the stated definitions; QΣ2 = QΠ2 is itself a collapse statement at the second level of QPH. Using it in a proof that QPH ⊆ QΣ2/qpoly is either circular or relies on an unproven assumption. This step is load-bearing, since it is what reduces the inner quantifier block ∀ρ2∃σ1∀σ2 to a QΣ2/qpoly predicate in the induction step.","section":"Section 3.2, Theorem 3.10, Eqs. (3.20)-(3.21)"},{"comment":"The collapse argument for BEQPH does not justify its minimax swap. The expression v*3 = max_{ρ1} min_{σ1} max_{ρ2∈S(ρ1)} Tr(R(ρ2⊗σ1)) is rewritten as max over a joint set of (ρ1,ρ2) followed by min over σ1, with the explanation that Sion's theorem applies. Sion's theorem swaps an outer max and min over convex compact sets for a function that is concave-convex in the two variables. Here the inner max over ρ2 is constrained by ρ1, so the two quantifiers being interchanged are not over an independent product set. Convexity of S(ρ1) does not by itself permit moving the max over ρ2 past the intervening min over σ1. Consequently the reduction of BEQΣ3 to Eq. (3.28), and the analogous derivation for BEQΣ4 and higher levels, is not established.","section":"Section 3.3, Theorem 3.14, Eqs. (3.25)-(3.28)"}],"minor_comments":[{"comment":"The proof states a 'standard promise structure' whose first condition is 'There exists at least a y such that P(y) ≥ a' and whose second condition is 'For all y, P(y) < a'; these two conditions cannot both be part of the same promise. The YES and NO cases should be stated disjunctively and should match Definition 2.2.","section":"Section 2.2, Lemma 2.8"},{"comment":"Sampling a uniform real r ∈ [0,B) using ⌈log n⌉ random bits is impossible, since a finite number of bits cannot sample a continuous uniform distribution. The proof should specify a fixed-point precision and argue that the gap δD is preserved.","section":"Section 2.2, Lemma 2.10, step (i)"},{"comment":"The sentence containing 'where b_{i−1} < B_{i−1}' is notationally unclear: B_{i−1} is not defined, and the use of B for both a prover register and an entanglement bound is confusing. Please define the maximal entanglement quantity explicitly and use distinct notation.","section":"Section 3.3, Definition 3.12"}],"recommendation":"reject","confidential_remarks":"The reader's assessment matches my reading of the manuscript. The UniqueQCPCP result rests on a demonstrably false polynomial construction; the Karp-Lipton proof uses nonstandard quantum-input QMA and an unproven hierarchy equality; and the BEQPH collapse relies on an unjustified minimax swap. These are not local presentation issues but defects in the central claims, and fixing them would require replacing the core arguments. I therefore recommend rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper's headline result, UniqueQCPCP = QCPCP, rests on a reduction that is simply wrong. Lemma 2.9 maps UniqueSAT to UniqueMTP via the multilinear polynomial P(x) = Σ_j ∏_{ℓ∈C_j} x̃_ℓ, and claims a clause contributes 1 exactly when it is satisfied. That is false: the product is 1 only when every literal in the clause is true. A satisfying assignment can thus make P(x) < m, so a YES instance can map to a NO instance of UniqueMTP. The example φ = (x1∨x2)∧(¬x1) with unique assignment (0,1) gives P = 1 < 2. Because the whole chain QCPCP ≤_BQ MTP ≤_p SAT ≤_RP USAT ≤_p UniqueMTP ≤_BQ UniqueQCPCP depends on this step, the first main result is unsupported.\n\nThe Karp–Lipton half has its own troubles. The proof defines a promise problem P_x whose input is a pair of quantum states (ρ1, ρ2) and claims P_x ∈ QMA. Standard QMA takes a classical input x plus a quantum witness; feeding quantum states as the instance is not a well-defined use of the class. Later the proof invokes QΣ2 = QΠ2 as if it were known; it is not proved here and is itself equivalent to a collapse statement. So the second main result is also not established.\n\nWhat is worth keeping? The definitions of BEQPH and SepQPH are new, and the idea that convexity, not just entanglement, drives the collapse is worth exploring. The writing is clear and the authors engage seriously with the literature. But those assets do not compensate for a false central lemma and an invalid Karp–Lipton proof.\n\nThis paper deserves a serious referee rather than a desk reject, because the questions are important and the mistakes are identifiable and fixable in principle. But as it stands, I would not accept it. The authors need to either repair the UniqueMTP reduction or retract the uniqueness result, and to rework the Karp–Lipton proof with a legitimate treatment of quantum inputs and a proof or explicit avoidance of QΣ2 = QΠ2. If those parts cannot be fixed, the BEQPH material alone might support a shorter paper.\n\nRead it if you work on quantum proof complexity; otherwise skip.","headline":"The UniqueQCPCP=QCPCP collapse rests on a false polynomial reduction, and the Karp–Lipton proof has a serious gap; the paper needs major revision.","tokens_in":19008,"tokens_out":2892,"would_cite":false,"duration_ms":30041,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["68Q12","68Q15","81P68"],"pacs":["03.67.-a","03.67.Lx"],"model":"deepseek-v4-flash","headline":"The paper claims that unique quantum-classical probabilistically checkable proofs are as powerful as general ones under randomized reductions, and that a non-uniform quantum Karp–Lipton theorem plus a bounded-entanglement hierarchy…","keywords":["quantum complexity theory","probabilistically checkable proofs","unique proofs","quantum polynomial hierarchy","Karp-Lipton theorem","bounded entanglement","quantum advice","convex optimization"],"falsifier":"Test the reduction on the formula $\\varphi = x_1 \\wedge (x_1 \\vee \\neg x_1)$, whose unique satisfying assignment is $x_1=1$. The lemma constructs $P(x)=x_1+x_1(1-x_1)$; at $x_1=1$, $P=1$, while the threshold is $a=m=2$, so the unique satisfying assignment is not accepted—directly contradicting the completeness condition and hence the claimed reduction chain.","tokens_in":17974,"feed_emoji":"⚛️","tokens_out":17130,"duration_ms":149630,"temperature":0.7,"pith_summary":"This paper sets out to show that several constraints on quantum proof systems do not constrain as much as they seem. It claims that requiring a quantum-classical PCP to have a unique accepting proof leaves its power unchanged, $\\mathsf{UniqueQCPCP} = \\mathsf{QCPCP}$ under $\\mathrm{BQ}$-operator and randomized reductions. It also proves a quantum analogue of the Karp–Lipton theorem—if $\\mathsf{QMA}\\subseteq \\mathsf{BQP}/\\mathsf{qpoly}$, then $\\mathsf{QPH}\\subseteq \\mathsf{Q\\Sigma}_2/\\mathsf{qpoly}$—and introduces a bounded-entanglement hierarchy $\\mathsf{BEQPH}$ that collapses unconditionally at level four, with each level reducible to a convex optimization problem. If these results are right, the computational border between 'hard' and 'easy' quantum proof systems is not drawn by uniqueness, advice, or entanglement by itself, but by whether the feasible set of protocols is convex.","feed_headline":"Unique quantum-classical PCPs equal general ones","feed_subtitle":"Same-power result via a reduction chain through SAT; quantum hierarchy collapses at level 4","key_machinery":"The carrying objects are the $\\mathrm{BQ}$-operator—a polynomial-time quantum reduction with success probability at least $2/3$—and convexity of quantum state sets. For the PCP result, the chain $\\mathsf{QCPCP} \\le_{\\mathrm{BQ}} \\mathsf{MTP} \\le_p \\mathsf{SAT} \\le_{\\mathrm{RP}} \\mathsf{UniqueSAT} \\le_p \\mathsf{UniqueMTP} \\le_{\\mathrm{BQ}} \\mathsf{UniqueQCPCP}$ transforms a general quantum-classical PCP into one with a unique proof, with uniqueness preserved by the Valiant–Vazirani hashing reduction. For the hierarchy results, the collapse is driven by Sion's minimax theorem applied to bilinear acceptance probabilities over convex, compact sets of density matrices with bounded relative entropy of entanglement; the consistent extension property ties each round to the previous one, and the bounded-entanglement sets $T^{(m)}$ have convex feasible regions, so alternating quantifiers reorder and every level becomes a two-round max-min problem.","core_discovery":"The paper's central discovery is that uniqueness and entanglement bounds do not carry the computational weight one might expect. In the PCP setting, it proves $\\mathsf{UniqueQCPCP} = \\mathsf{QCPCP}$ by composing a $\\mathrm{BQ}$ reduction from $\\mathsf{QCPCP}$ to a multilinear threshold problem, a deterministic reduction to $\\mathsf{SAT}$, the Valiant–Vazirani randomized reduction from $\\mathsf{SAT}$ to $\\mathsf{UniqueSAT}$, a deterministic reduction back to a unique multilinear threshold problem, and a $\\mathrm{BQ}$ reduction from that problem to a unique quantum-classical verifier. In the hierarchy setting, it proves that the assumption $\\mathsf{QMA}\\subseteq \\mathsf{BQP}/\\mathsf{qpoly}$ forces $\\mathsf{QPH}\\subseteq \\mathsf{Q\\Sigma}_2/\\mathsf{qpoly}$, a non-uniform quantum Karp–Lipton theorem. It then defines $\\mathsf{BEQPH}$, whose proof states must satisfy a consistent extension property and an upper bound on relative entropy of entanglement, and shows that every level of $\\mathsf{BEQPH}$ is equivalent to a two-round convex max-min problem, yielding an unconditional collapse $\\mathsf{BEQ\\Sigma}_k = \\mathsf{BEQ\\Sigma}_4$ for all $k\\ge 4$; the zero-entanglement variant $\\mathsf{SepQPH}$ inherits the same collapse. The stated reason is that the feasible sets are convex and compact, so Sion's minimax theorem lets the alternating quantifiers be reordered, whereas the product-state sets in $\\mathsf{QPH}$ are non-convex.","pith_inferences":["I infer that the convexity mechanism is generic: any entanglement measure with convex and compact sublevel sets—not just relative entropy of entanglement—should yield the same fourth-level collapse, which is testable by rerunning the proof with a different measure.","I infer that the UniqueQCPCP result may be repairable if the flawed polynomial encoding in Lemma 2.9 is replaced by one in which each satisfied clause contributes exactly 1, for example by summing over all subsets of literals in a clause; the rest of the reduction chain would then go through unchanged.","I infer that the non-uniform Karp–Lipton proof, which only invokes QMA containment at the base level, likely extends to QCMA or to classical advice, giving $\\mathsf{QPH}\\subseteq \\mathsf{Q\\Sigma}_2/\\mathsf{poly}$ under the analogous assumption.","I infer that the inability of SepQPH to simulate QPH suggests a separation result: if SepQPH is eventually proved to be properly contained in QPH, it would pin convexity of the proof set as the exact boundary between tractable and potentially intractable quantum hierarchies."],"forward_implications":["If $\\mathsf{UniqueQCPCP} = \\mathsf{QCPCP}$ holds, uniqueness is not a resource limitation in quantum-classical PCPs, and any hardness or completeness result for $\\mathsf{QCPCP}$ carries over to its unique variant.","The quantum Karp–Lipton theorem implies that if QMA has efficient quantum advice, then the whole quantum polynomial hierarchy can be decided by a $\\mathsf{Q\\Sigma}_2$ protocol with that same advice, matching the classical $\\mathsf{PH} = \\mathsf{\\Sigma}_2$ collapse.","The BEQPH collapse means that imposing bounded-entanglement and consistency constraints across rounds adds no power beyond level four; higher levels reduce to two-round convex optimization and are therefore tractable in the semidefinite-programming sense.","Since SepQPH with zero entanglement also collapses but cannot simulate QPH, the hardness of QPH cannot be blamed on entanglement alone—the non-convexity of product states is essential.","The results place BEQPH and SepQPH inside $\\mathsf{QRG(1)}$, the class of constant-round quantum refereed games, and support the paper's conjecture that both are properly contained in QPH."],"supporting_citations":[{"why":"Supplies the BQ-operator definition and the base QCPCP-to-MTP reduction that the unique chain starts from.","marker":"[BGW24]"},{"why":"Provides the randomized reduction from SAT to UniqueSAT that creates the unique witness in the middle of the chain.","marker":"[VV86]"},{"why":"Establishes the QEPH collapse and the consistent-extension/convexity framework that BEQPH and SepQPH are modeled on.","marker":"[GY24]"},{"why":"Formulates QPH and QCPH and proves earlier partial Karp–Lipton results that this paper's quantum Karp–Lipton theorem extends.","marker":"[GSS+22]"},{"why":"Provides the QCPH collapse and error-reduction machinery used in the hierarchy arguments.","marker":"[AGKR24]"},{"why":"Supplies the equivalence between constant-round quantum refereed games and second-level quantum interactive proofs used in Corollary 3.15.","marker":"[JW09]"},{"why":"Supplies Sion's minimax theorem, the tool that swaps alternating quantifiers over convex state sets.","marker":"[Sio58]"},{"why":"Supplies the standard fact that density matrices form a convex compact set, grounding the convexity premise.","marker":"[Wat18]"},{"why":"Provides the UniqueQCMA = QCMA randomized reduction that the paper explicitly parallels in the PCP setting.","marker":"[ABOBS22]"},{"why":"Supplies the classical Karp–Lipton theorem being generalized to quantum advice.","marker":"[KL80]"}],"fun_headline_variants":["Unique quantum PCPs equal general; hierarchy collapses at 4","Quantum proof systems: uniqueness free, hierarchy collapses at 4","Quantum Karp-Lipton: QMA advice collapses QPH; uniqueness no loss","Bounded-entanglement quantum hierarchy collapses at level 4","Unique QCPCP = QCPCP; BEQPH collapses at level 4"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The reduction chain to unique quantum-classical PCPs assumes that a satisfied clause in the CNF formula contributes a full $1$ to the constructed multilinear polynomial; this is only true when every literal in the clause is true.","fun_headline_variants_meta":{"raw":{"variants":["Unique quantum PCPs equal general; hierarchy collapses at 4","Quantum proof systems: uniqueness free, hierarchy collapses at 4","Quantum Karp-Lipton: QMA advice collapses QPH; uniqueness no loss","Bounded-entanglement quantum hierarchy collapses at level 4","Unique QCPCP = QCPCP; BEQPH collapses at level 4"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001551,"raw_usage":{"total_tokens":6319,"prompt_tokens":1181,"completion_tokens":5138,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":797,"completion_tokens_details":{"reasoning_tokens":5042}},"tokens_in":797,"tokens_out":5138,"duration_ms":35751,"temperature":1.0,"reasoning_tokens":5042,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:25:21.593251+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Test the reduction on the formula $\\varphi = x_1 \\wedge (x_1 \\vee \\neg x_1)$, whose unique satisfying assignment is $x_1=1$. The lemma constructs $P(x)=x_1+x_1(1-x_1)$; at $x_1=1$, $P=1$, while the threshold is $a=m=2$, so the unique satisfying assignment is not accepted—directly contradicting the completeness condition and hence the claimed reduction chain.","supporting_citations":[],"review_version":2}