{"id":"dcb283f0-e653-44ca-aac4-735b543a2348","arxiv_id":"1909.01336","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For any resource measure satisfying invariance, continuity, and additivity, a resource-changing unitary cannot be perfectly realized with free unitaries and a finite-dimensional ancilla, and the achievable error shrinks only as the ancilla dimension grows.","lead":"This paper proves a general trade-off: any unitary that changes a quantum resource, such as energy, coherence, entanglement, or magic, cannot be implemented perfectly using only free unitaries and a finite-sized helper system. The smaller the allowed error, the larger the helper must be, and perfect implementation would require an infinite-dimensional helper.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Qubit magic application relies on unproven full additivity of log stabilizer extent; the cited three-qubit multiplicativity does not imply Property 3'.","rationale":"The reader's stated weakest assumption is the imported no-correlation lemma (Lemma 7), but the reader's rationale identifies the qubit-magic multiplicativity gap as the main reason for a conditional verdict. I agree with the latter: the paper explicitly asserts that three-qubit multiplicativity of the stabilizer extent implies Property 3', which is an unsupported logical jump. The no-correlation lemma is a verification dependency rather than a demonstrated error, since it is stated in full and cited to [8]; a paper may legitimately rely on an external lemma, though the reliance is a reason for caution. The central theorem itself appears structurally sound: it uses only unitary invariance, continuity, additivity, and the no-correlation lemma, and the energy, coherence, entanglement, and quopit-mana applications satisfy the stated hypotheses. The magic gap does not invalidate Theorem 1, so a revision that either proves full multiplicativity for arbitrary pure products or restricts the qubit-magic claim would resolve the issue. I therefore recommend keeping the reader's conditional verdict: the concern is real but localized, and the proposed SDP check can settle whether it is merely a missing proof or an actual false claim.","tokens_in":18695,"tokens_out":16797,"duration_ms":174170,"concrete_test":"Use the SDP definition of stabilizer extent to compute ξ(|T⟩ ⊗ |φ⟩) for an entangled 4-qubit pure state |φ⟩ that cannot be reduced to a tensor product of ≤3-qubit factors, and compare the result with ξ(|T⟩) · ξ(|φ⟩). If the two values differ, log ξ is not additive on pure product states, Property 3' fails, and the qubit-magic application must be withdrawn. If equality is found for a suitable suite of such states, provide an independent proof of multiplicativity for arbitrary pure products before retaining the claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section V.D.1 applies Theorem 4 to qubit magic by taking R(|ψ⟩⟨ψ|) = log ξ(|ψ⟩) and asserting that because [48] proved the stabilizer extent is multiplicative for tensor products of states supported on up to three qubits, Property 3' holds. This inference is not licensed. Property 3' requires additivity for arbitrary pure product states, including products of the form ρ_S^{(i)} ⊗ ρ_E, where ρ_E may be an arbitrary pure state of an N-qubit environment. A theorem that only covers tensor products in which each factor is supported on at most three qubits does not extend by induction: once two such factors are tensored, the resulting state is supported on up to six qubits, and subsequent multiplicativity steps would require the statement for larger factors, which the cited result does not provide. Consequently, the bound on G^p_{U_NC} + L^p_{U_NC} and the Ω(log(...)) scaling for non-Clifford qubit gates are not established as written. The central Theorem 1 and the quopit/mana application are unaffected, so this is a localized but genuine gap in the claimed universality over qubit magic.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a general trade-off relation for implementing a target unitary U_S using only free unitaries acting on the system together with an ancilla. For any resource measure R satisfying invariance under free unitaries, a continuity condition, and additivity for product states, Theorem 1 bounds the resource generating/losing power G_U_S + L_U_S in terms of the implementation error and the ancilla dimension, with the bound vanishing as the error and the inverse ancilla dimension tend to zero. Corollary 2 derives a no-go result: no resource-changing unitary can be perfectly implemented with a finite-dimensional ancilla. Theorem 4 relaxes additivity to pure product states and gives an analogous bound. The paper applies the framework to energy, asymmetry, coherence, entanglement, and magic. For qubit magic, it uses the log stabilizer extent as the resource measure and invokes multiplicativity results from Ref. [48]. The proofs rely on a no-correlation lemma reproduced from Ref. [8], and Proposition 6 provides a continuity bound for the max-relative entropy measure.","tokens_in":18825,"tokens_out":4461,"duration_ms":47142,"significance":"If the main theorem is correct, this is a valuable unifying result: it places previously known conservation-law limits on unitary implementation into a general resource-theoretic framework and yields quantitative bounds with operational error measures. The proof structure is transparent, the axiomatic assumptions are clearly stated, and Proposition 6 is a useful continuity result in its own right. The applications to energy, coherence, entanglement, and quopit magic appear sound. However, the central theorem depends on an externally imported lemma that is not proved here, and the qubit-magic application relies on an additivity inference that is not justified by the cited reference. These issues do not undermine the core theorem but do affect the claimed universality over qubit magic and the self-containedness of the main proof.","major_comments":[{"comment":"Theorem 1 rests entirely on the no-correlation lemma: inequalities (A3) and (A4) are the mechanism that converts the implementation error into closeness of the reduced environment states. The lemma is only restated from Ref. [8], which is a preprint by the same authors, and no proof is included in this manuscript. Since Theorem 1 and Corollaries 2 and 3 collapse if this lemma fails, the main derivation is not self-contained. Please either prove Lemma 7 in the paper or provide a published reference containing a complete proof.","section":"Section III / Appendix A, Lemma 7"},{"comment":"The application of Theorem 4 to qubit magic is not established. The paper claims that R(|ψ⟩⟨ψ|) = log ξ(|ψ⟩) satisfies Property 3' because Ref. [48] showed that the stabilizer extent is multiplicative for tensor products of states supported on up to three qubits. Property 3' requires additivity for arbitrary pure product states, including ρ_S ⊗ ρ_E where ρ_E is an arbitrary N-qubit pure state. The cited three-qubit multiplicativity does not extend by induction, since a tensor product of two such factors is supported on up to six qubits and the cited result does not cover that case. Consequently, the bound on G^p_{U_NC} + L^p_{U_NC} and the claimed Ω(log((G+L)/√ε)) scaling for non-Clifford qubit gates are not supported as written. This is a localized but genuine gap in the claimed universality over qubit magic.","section":"Section V.D.1"},{"comment":"The discussion toward full generality is more limited than it may appear. The statement that exact implementation forces Eq. (29) for any measure with Property 1 is correct, but the subsequent argument only shows that a subadditive monotone would satisfy R(|φ⟩⟨φ|) = R(σ_E'); it does not establish that some measure with Property 1 violates Eq. (29) for a resourceful unitary. Thus the section does not prove a no-go theorem in the absence of additivity; it identifies a condition whose failure would imply impossibility. The authors should make this limitation explicit.","section":"Section VI, Eq. (29)"}],"minor_comments":[{"comment":"The title contains a spacing artifact: \"unitar y evolutions\" should read \"unitary evolutions\".","section":"Title and throughout"},{"comment":"In the coherence application, the continuity functions are listed as f1(x) = x, g1(x) = log x, and h1(x) = (1+x)b(x/(1+x)), but Eq. (22) has the dimension d in the logarithmic term. This is presumably g1(d) = log d; the notation should be clarified to avoid confusion between the distance variable and the dimension.","section":"Section V.B, after Eq. (22)"},{"comment":"The quantity F_e(ρ_S, Λ) is defined as a square root of an entanglement fidelity. This is nonstandard and could be confused with the usual fidelity; a brief comment explaining the choice would improve readability.","section":"Section II, Eq. (9)"},{"comment":"The paper says the stabilizer extent is \"multiplicative\" while the resource measure R is the logarithm of the extent; the translation to additivity of R is correct only if the cited multiplicativity holds exactly for the required tensor products, which is the issue raised in the major comment.","section":"Section V.D.1"},{"comment":"The note about related independent work by Chiribella, Yang, and Renner is acknowledged, but the manuscript does not discuss how the results compare to that work. A brief statement of the relation would be helpful to readers.","section":"Note added, Section VII"}],"recommendation":"major_revision","confidential_remarks":"The central Theorem 1 and the energy/quopit applications are likely correct and publishable after revision. The two substantive issues are the unproved no-correlation lemma imported from a preprint by the same authors and the unjustified qubit-magic additivity claim. I would support acceptance once these are resolved, ideally by proving the lemma and by either proving full pure-product additivity for log stabilizer extent or rephrasing the qubit-magic claim without claiming universality over arbitrary qubit products."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a real generalization, not a repackaging. Takagi and Tajima show that if you have any resource measure satisfying invariance, continuity, and additivity, then implementing a resourceful unitary with a free unitary plus a finite-dimensional ancilla forces a quantitative trade-off between implementation error and ancilla dimension, set by the resource generating/losing power of the target. The no-go corollary is clean: perfect implementation is impossible for any resource-changing unitary. The proof structure, built on the no-correlation lemma from their earlier work, is sound as far as I can check, and the applications to energy, coherence, entanglement, and quopit magic all go through. Proposition 6, a continuity bound for max-relative entropy of magic, is a useful standalone piece.\n\nTwo soft spots. First, the no-correlation lemma is imported from a preprint by the same authors and stated without proof in Appendix A. It is load-bearing. I don't see circularity—the lemma is proved in [8]—but a referee should ask for a proof in this paper or a published reference. Second, the qubit magic application overreaches. The paper claims Property 3' follows because the stabilizer extent is multiplicative for tensor products of states supported on up to three qubits [48]. That does not license additivity for arbitrary pure product states: a state of three qubits tensored with an N-qubit environment is not covered by the cited result, and no induction argument is supplied. So the Omega(log(...)) scaling for non-Clifford qubit gates is not established as written. This is a localized gap in Section V.D.1; Theorem 1, Theorem 4, and the other applications stand.\n\nWho should read this: anyone in quantum thermodynamics, resource theories, or magic-state distillation. It deserves a serious referee. The fix for the magic section is straightforward: either cite a stronger multiplicativity result or restrict the claim. I would not desk-reject. Recommended action: send to peer review, with a referee who can check the cited multiplicativity claim.","headline":"A clean general no-go for implementing resourceful unitaries with finite ancillas, with a localized but real overreach in the qubit-magic application that a referee should catch.","tokens_in":19412,"tokens_out":5503,"would_cite":true,"duration_ms":54482,"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":"The paper proves a universal limitation: a unitary that changes a quantum resource cannot be perfectly realized with free unitaries and finite ancilla, and it quantifies the accuracy–ancilla trade-off.","keywords":["resource theories","unitary implementation","no-go theorem","ancilla dimension","resource generating power","quantum coherence","entanglement","magic states"],"falsifier":"Find one counterexample: a resource measure $R$ satisfying Properties 1–3, a finite-dimensional ancilla, and a free unitary $V$ such that $\\mathrm{Tr}_E[V(\\rho_S\\otimes\\rho_E)V^\\dagger]$ exactly equals $U_S\\rho_S U_S^\\dagger$ for all $\\rho_S$ while $G_{U_S}+L_{U_S}>0$; the theorem says none exists, so an explicit construction or numerical search that finds one would settle the matter.","tokens_in":18415,"feed_emoji":"⚛️","tokens_out":8852,"duration_ms":79945,"temperature":0.7,"pith_summary":"The paper proves a universal limitation on what can be done with a restricted set of quantum operations: if a target unitary can change the amount of some resource, it cannot be implemented exactly with only 'free' unitaries and a finite-dimensional auxiliary system. It quantifies this by a trade-off inequality between implementation error, ancilla dimension, and the resource-generating and resource-losing power of the target unitary. The bound holds for any resource measure that is invariant under free unitaries, continuous, and additive on product states (or, in a relaxed version, additive on pure product states when the ancilla is pure). This makes the no-go result apply broadly, to energy, coherence, entanglement, asymmetry, and magic. If correct, the result identifies a generic cost in quantum engineering: resource-changing unitaries are never free to implement, and their accuracy is fundamentally limited by the size of the available aiding system.","feed_headline":"No finite ancilla can perfectly fake a resourceful quantum gate","feed_subtitle":"A new trade-off ties implementation error and ancilla size to the resource a target unitary creates or destroys.","key_machinery":"The machinery has three parts. The resource generating power $G_U:=\\max_\\rho\\{R(U\\rho U^\\dagger)-R(\\rho)\\}$ and resource losing power $L_U:=-\\min_\\rho\\{R(U\\rho U^\\dagger)-R(\\rho)\\}$ measure how much resource the target unitary can create or destroy. The no-correlation lemma (reproduced from earlier work as Lemma 7) says that an approximate implementation forces the final ancilla state to be nearly independent of the input, with a quantitative bound on how close the environment states for two distinguished inputs must be. The proof combines this lemma with the three properties of $R$—invariance under free unitaries, Lipschitz-type continuity, and additivity on product states—to bound $G_{U_S}+L_{U_S}$ by the error and ancilla dimension. The relaxed theorem uses the pure-state version of additivity plus the fact that a pure ancilla state gives a pure intermediate reference state.","core_discovery":"The central claim is Theorem 1: for any resource measure $R$ satisfying invariance under free unitaries, continuity, and additivity for product states, and for any implementation $I=(H_E,V_{SE},\\rho_E)$ of a unitary $U_S$ using a free unitary $V_{SE}$ and an ancilla of dimension $d_E$, one has $G_{U_S}+L_{U_S}\\le \\alpha_L(\\delta^I_{U_S},d_E)+\\beta_L(\\delta^I_{U_S})$, where $G_{U_S}$ and $L_{U_S}$ are the maximum resource increase and decrease $U_S$ can induce, $\\delta^I_{U_S}$ is the worst-case gate error, and $\\alpha_L,\\beta_L$ vanish as the error and the inverse ancilla dimension go to zero. The immediate corollary is that a unitary with $G_{U_S}+L_{U_S}>0$ cannot be perfectly implemented with a finite-dimensional ancilla. A second theorem relaxes additivity to pure product states and yields the same conclusion for implementations with a pure ancilla state. The paper then verifies the hypotheses for specific measures—energy expectation, Wigner–Yanase skew information, athermality, relative entropy of coherence, squashed entanglement, relative entropy of entanglement, stabilizer extent, and mana—so the no-go conclusion covers those settings.","pith_inferences":["A direct but unstated corollary is that adaptive protocols involving measurement and feedforward can evade the dimensional bound, because the theorem applies only to unitary circuits without intermediate measurements; gate teleportation is the standard example.","The same no-correlation logic could likely be applied to channels rather than unitaries, yielding analogous trade-offs for resource-changing quantum channels; this is a natural testable extension the paper does not explore.","The lower bounds are not shown to be tight; constructing explicit approximate implementations that saturate them would clarify whether the required ancilla dimension is really as large as the theorem suggests.","Because the no-go holds for any additive continuous measure, it suggests that exact resource-changing operations require either an infinite-dimensional reference frame or some non-unitary ingredient, which may have consequences for superselection rules and thermodynamic batteries."],"forward_implications":["Corollary 2: for any resource theory with a measure satisfying Properties 1–3, no unitary with nonzero resource-generating or resource-losing power can be perfectly implemented with a finite-dimensional ancilla and free unitaries alone.","The trade-off is quantitative: as the allowed error $\\delta$ goes to zero, the ancilla dimension must grow without bound; Theorem 1 gives the explicit functional dependence through $\\alpha_L$ and $\\beta_L$.","The results reproduce the known energy-conservation no-go theorem as a special case, using the energy expectation value as the resource measure.","In the theory of coherence, any coherence-generating unitary is impossible to implement exactly with a finite coherent ancilla; in entanglement theory, any entangling gate is impossible with local unitaries and finite shared entanglement (without classical communication).","For fault-tolerant computation, implementing a non-Clifford gate with Clifford unitaries and magic states requires the number of ancilla qubits to grow at least as $\\Omega(\\log((G^p_{U_{NC}}+L^p_{U_{NC}})/\\sqrt{\\epsilon}))$ for accuracy $\\epsilon$."],"supporting_citations":[{"why":"Supplies the no-correlation lemma (Lemma 7) that the proofs of Theorems 1 and 4 rely on.","marker":"[8]"},{"why":"Establishes the earlier energy-conservation uncertainty bound that this paper generalizes.","marker":"[7]"},{"why":"Frames the general resource-theoretic language of free states and free operations used throughout.","marker":"[9]"},{"why":"Defines the relative entropy of coherence, one of the measures checked against Properties 1–3.","marker":"[13]"},{"why":"Introduces the resource theory of stabilizer computation and the mana measure used for quopits.","marker":"[23]"},{"why":"Provides the stabilizer extent, the pure-state additive measure used for qubit magic in Theorem 4.","marker":"[48]"},{"why":"Defines relative entropy of entanglement, which is additive on pure product states and used in applications.","marker":"[47]"},{"why":"Gives the asymptotic continuity bound used to verify Property 2 for coherence and related measures.","marker":"[63]"}],"fun_headline_variants":["Finite ancilla can't perfectly mimic any resource-changing gate","Universal limit: free unitaries can't perfectly simulate resourceful ops","Resource-changing gates can't be perfectly done with finite ancilla","No perfect finite-ancilla implementation of any resourceful unitary","Trade-off ties ancilla size to error for resourceful gates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire argument hangs on the no-correlation lemma taken from earlier work without proof—that any approximate implementation of a unitary leaves the environment nearly uncorrelated with the system input; if that lemma fails, the trade-off and the no-go theorem collapse.","fun_headline_variants_meta":{"raw":{"variants":["Finite ancilla can't perfectly mimic any resource-changing gate","Universal limit: free unitaries can't perfectly simulate resourceful ops","Resource-changing gates can't be perfectly done with finite ancilla","No perfect finite-ancilla implementation of any resourceful unitary","Trade-off ties ancilla size to error for resourceful gates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000437,"raw_usage":{"total_tokens":2217,"prompt_tokens":936,"completion_tokens":1281,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":552,"completion_tokens_details":{"reasoning_tokens":1194}},"tokens_in":552,"tokens_out":1281,"duration_ms":9810,"temperature":1.0,"reasoning_tokens":1194,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T05:23:41.817693+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find one counterexample: a resource measure $R$ satisfying Properties 1–3, a finite-dimensional ancilla, and a free unitary $V$ such that $\\mathrm{Tr}_E[V(\\rho_S\\otimes\\rho_E)V^\\dagger]$ exactly equals $U_S\\rho_S U_S^\\dagger$ for all $\\rho_S$ while $G_{U_S}+L_{U_S}>0$; the theorem says none exists, so an explicit construction or numerical search that finds one would settle the matter.","supporting_citations":[{"cited_title":"Coherence cost for violating conservation laws","cited_arxiv_id":"1906.04076","evidence_quote":"Supplies the no-correlation lemma (Lemma 7) that the proofs of Theorems 1 and 4 rely on."},{"cited_title":"Skrzypczyk, A","cited_arxiv_id":null,"evidence_quote":"Gives the asymptotic continuity bound used to verify Property 2 for coherence and related measures."}],"review_version":1}