{"id":"59810724-556c-487c-bc66-d8e1d3e6a9c5","arxiv_id":"2501.09205","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The discrimination power of any state in quantum channel discrimination equals one plus its robustness to noise, for all finite-dimensional convex resource theories.","lead":"This paper proves that in any quantum resource theory, the maximum advantage a resourceful state gives in a single-shot channel discrimination task is exactly one plus its robustness, the amount of noise needed to erase the resource. The result gives an operational meaning to robustness and unifies prior results for entanglement, coherence, and asymmetry.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Bound-resource-state characterization is false for non-closed free sets; the central equality remains sound but the advertised full characterization needs revision.","rationale":"The reader's weakest_assumption concerns strong duality in Problem (5). That is not the most serious issue: for closed convex F containing a positive-definite state, the dual feasible set is compact, and the equality between the generalized robustness and the value of Problem (5) is a standard conic-duality result, so the proof of Theorem 1 is on solid ground. The real defect is the bound-resource equivalence stated after Corollary 2. It is asserted for arbitrary subsets F, but it fails for non-closed convex F: states in cl(coF)\\coF have zero robustness and hence G=1, yet they are not in coF. The qubit example above settles the matter. Corollary 2 itself, and the main equalities G=1+R, are unaffected; the error is in interpreting these equalities as a full characterization of resourcefulness for non-closed free sets. The reader's CONDITIONAL verdict is therefore appropriate, but the emphasized reason should be the closure issue rather than strong duality.","tokens_in":20694,"tokens_out":35889,"duration_ms":389764,"concrete_test":"Verify the qubit counterexample: set F = DenA \\ {|0><0|} for a two-dimensional system. Check that F is convex and that coF=F. Compute R_coF(|0><0|) directly from Eq. (3) using tau=|1><1| and lambda tending to 0, showing the infimum is 0. Confirm that |0><0| is not in coF. If this example is accepted, revise the equivalence in Section III to use the closure, and adjust the abstract's full-characterization claim accordingly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section III the paper states that for any subset F, a state rho not in F is a bound resource state (G(rho,F)=1) iff rho is in coF. This is false when coF is not closed. Counterexample: let F be the qubit state space minus the pure state |0><0|. F is convex (removing an extreme point of the Bloch ball preserves convexity), contains positive-definite states, and coF=F. For rho=|0><0|, rho is not in F and rho is not in coF. Yet for every lambda>0 and tau=|1><1|, (rho+lambda*tau)/(1+lambda) is a non-extreme mixed state lying in F, so R_coF(rho)=0. By Corollary 2, G(rho,F)=1+R_coF(rho)=1, making rho bound by the paper's own definition. The correct condition is rho in cl(coF)\\F, not rho in coF. This does not invalidate Theorem 1 or Theorem 3, but it invalidates the abstract's claim that resources are fully characterized for arbitrary (possibly non-closed or non-convex) free sets. The strong-duality step used in the proof of Theorem 1 is standard for closed convex F containing a positive-definite state and is not where the paper fails.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies single-shot quantum channel discrimination and asks how much the maximum average success probability can be improved by using a resource state ρ instead of the best free state. The main quantity is the discrimination power G(ρ,F), defined as the supremum over channel discrimination ensembles of the ratio between the success probability with ρ and that with the best free state. Theorem 1 states that for any convex free set F, G(ρ,F)=1+R_F(ρ), where R_F is the generalized robustness. Corollary 2 extends this to arbitrary (possibly nonconvex) F via G(ρ,F)=G(ρ,coF)=1+R_coF(ρ). Theorem 3 extends the equality to protocols with an auxiliary system C when the free set preserves operations on C, and Corollary 4 gives the general auxiliary-system formula GC(ρ,F)=1+R_coFC(ρ). The paper also characterizes 'bound resource states' (resource states with G=1) and discusses the difference between G and the previously studied quantity H defined in Eq. (8). The proofs are detailed and mostly self-contained, with appendices covering group-covariant measurements, the auxiliary-system construction, and degeneracies when no positive-definite free state exists.","tokens_in":20862,"tokens_out":7286,"duration_ms":74023,"significance":"If the central equalities hold, the paper resolves a natural open question from Ref. [41] by giving the generalized robustness an exact operational meaning as the maximum attainable advantage in one-shot channel discrimination, both with and without auxiliary systems. This is a clean and broadly applicable result: it unifies earlier special-case results for entanglement, coherence, asymmetry, and steering, and it provides a concrete recipe for constructing discrimination ensembles that achieve the bound. The auxiliary-system result is particularly valuable because such exact quantifications were previously known only in special cases. The paper also usefully clarifies the distinction between the discrimination power G and the earlier measure H. The mathematical core of Theorems 1 and 3 appears sound and the proofs are carefully structured. However, the paper's advertised 'full characterization' of resource states is stated too broadly: the bound-resource-state characterization in Section III is false for non-closed free sets, and this needs correction before the paper can be accepted.","major_comments":[{"comment":"The statement 'ρ ∉ F being a bound resource state is equivalent to ρ ∈ co F' is false when co F is not closed. Counterexample: let F be the qubit state space minus the pure state |0⟩⟨0|. This F is convex, contains positive-definite states, and satisfies co F = F. For ρ = |0⟩⟨0| we have ρ ∉ F and ρ ∉ co F. However, for every λ > 0 the state (ρ + λ|1⟩⟨1|)/(1+λ) is a non-extreme mixed state lying in F, so R_coF(ρ) = 0. Corollary 2 then gives G(ρ,F) = 1, making ρ a bound resource state by the paper's own definition, even though ρ ∉ co F. The correct equivalence is ρ ∈ cl(co F) \\ F. This error propagates to the abstract's claim that 'resources can be fully characterized' for arbitrary (possibly non-closed) free sets and to Figure 1, whose gray areas should be cl(co F) \\ F and cl(co FC) \\ F rather than co F \\ F and co FC \\ F. The main equalities of Theorem 1 and Corollary 2 are not affected, but the characterization claim needs to be revised.","section":"Section III, paragraph after Corollary 2"},{"comment":"The paper claims that 'there are no bound resource states if and only if F is convex and closed.' This biconditional is in fact correct, but the reasoning given in Section III is incomplete: the proof uses the erroneous equivalence ρ ∈ co F instead of ρ ∈ cl(co F). Since the 'if' direction is immediate and the 'only if' direction follows from R_coF(ρ) = 0 iff ρ ∈ cl(co F), the statement should be derived from the closure-corrected characterization. The current text, as written, gives a false intermediate statement that could mislead readers about the role of topological closure in nonconvex resource theories.","section":"Abstract and Section III, bound-state characterization"}],"minor_comments":[{"comment":"The proof assumes the existence of an optimal solution x⋆ of the dual problem (5) and uses the equality Tr(x⋆ρ) = 1 + R_F(ρ). This is a standard strong-duality result for closed convex F containing a positive-definite state, but it is cited from Ref. [49] rather than proved. Since the proof has already reduced to closed F, a brief explicit statement that Slater's condition guarantees attainment would make the argument self-contained at this point.","section":"Theorem 1 proof, Eq. (5)"},{"comment":"The identities G(ρ,F) = G(ρ,cl F) and R_F(ρ) = R_clF(ρ) are asserted without proof. They are true because the success probability and the feasibility condition in Eq. (3) are continuous in the state, but a one-sentence justification would improve readability.","section":"Theorem 1 proof, first paragraph"},{"comment":"The symbol C is used both for the auxiliary system in the discrimination problem and for the output system of the subchannels Λ̃_n in Lemma 6 and Lemma 7. This overloading makes the proof of Theorem 3 significantly harder to follow; renaming one of the two systems (e.g., calling the channel output system D) would clarify the construction.","section":"Theorem 3 and Appendix C, notation"},{"comment":"The channels in Eq. (7) and the subchannels in Lemma 6 involve denominators N−1. The proof chooses N large enough (N ≥ 2), but this is not stated explicitly at the point of definition; adding 'with N ≥ 2' would avoid a momentary division-by-zero concern.","section":"Eq. (7) and Lemma 6"},{"comment":"In the proof that Π^sym is a measurement, the line 'U_h(I_NB) = I_NB' appears after summing over g; the text is correct but could be compressed. More importantly, the equality PS(σ,Π;·) = PS(σ,Π^sym;·) is derived for equal priors, which is used later; this restriction should be mentioned in the lemma statement for clarity.","section":"Appendix B, Lemma 5"}],"recommendation":"major_revision","confidential_remarks":"The central mathematical results (Theorems 1 and 3, Corollaries 2 and 4) appear sound, and the paper makes a genuine contribution by giving an operational interpretation of robustness in arbitrary finite-dimensional convex resource theories and in auxiliary-assisted discrimination. My main concern is the false characterization of bound resource states for non-closed free sets. This is a localized error in the interpretation of Corollary 2, and it can be fixed by replacing co F with cl(co F) throughout the characterization statements and in Figure 1. The strong-duality invocation in Eq. (5) is standard and not a serious flaw. I therefore recommend major revision rather than rejection: the paper's main contribution is defensible, but the advertised 'full characterization' claim is currently incorrect as stated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper's central result is solid: Theorem 1 proves G(rho,F)=1+R_F(rho) for all finite-dimensional convex resource theories, and the proof is clean. The upper bound is elementary, and the lower bound uses an explicit covariant channel construction from the dual optimum. This is a real step beyond previous special-case results for entanglement, coherence, and asymmetry, and it gives the robustness measure a direct operational meaning in single-shot channel discrimination. The auxiliary-system extension (Theorem 3 and Corollary 4) is also well argued and plausible.\n\nThe main flaw is the bound-resource-state characterization in Section III. The paper claims that for any subset F, rho not in F is a bound resource state iff rho is in coF. That is false when coF is not closed. The stress-test counterexample is correct: take F to be the qubit Bloch ball minus |0><0|. F is convex, coF=F, and for rho=|0><0| we have R_coF(rho)=0, so G(rho,F)=1 by Corollary 2, yet rho is not in coF. The correct condition is rho in cl(coF)\\F, not coF\\F. This does not damage Theorems 1 or 3, but it invalidates the abstract's claim that resources are \"fully characterized\" for arbitrary (possibly non-closed) free sets, and it makes Figure 1 misleading as drawn. The authors should fix this and strike the overbroad language.\n\nTwo minor points. First, the proof relies on the Lagrange dual in Eq. (5) attaining its optimum, citing Ref. [49] for strong duality. That is a standard tool for closed convex F with a positive-definite free state, so it is not a real weakness, but a one-line justification would make the proof self-contained. Second, the paper should explicitly relate its result to Ref. [51] (Kuroiwa et al., \"Every quantum helps\"), since that work already considered operational advantage beyond convexity; the reader needs to know exactly which of the present statements are new relative to that paper.\n\nOverall, this is a serious contribution with one localized but real error in a side characterization. The main equality survives. It deserves a careful referee, not a desk rejection, and the revision should be minor once the bound-resource statement is corrected.","headline":"The main equality G=1+R is correct and significant, but the bound-resource characterization in Section III is wrong for non-closed free sets and the abstract overclaims as a result.","tokens_in":21439,"tokens_out":3316,"would_cite":true,"duration_ms":34016,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P50","81P45"],"pacs":[],"model":"deepseek-v4-flash","headline":"In any resource theory of quantum states, the generalized robustness of a state is exactly the maximum factor by which it improves one-shot quantum channel discrimination over the best free state.","keywords":["quantum channel discrimination","resource theories of quantum states","generalized robustness","operational resource quantification","one-shot discrimination","bound resource states","auxiliary systems in quantum discrimination","convex resource theories"],"falsifier":"Compute $R_F(\\rho)$ for a small finite-dimensional convex free set and state by solving the dual program, construct the paper's channel ensemble from the optimizer, and check the measured success-probability ratio; a ratio strictly larger than $1+R_F(\\rho)$ would refute Theorem 1, while the theorem predicts the constructed ensemble reaches the bound.","tokens_in":20417,"feed_emoji":"⚛️","tokens_out":10371,"duration_ms":99644,"temperature":0.7,"pith_summary":"The paper aims to prove that, in any resource theory of quantum states, the generalized robustness of a state is exactly the operational advantage that state provides in one-shot quantum channel discrimination. Concretely, the maximum ratio between the success probability achievable with a given state and the success probability achievable with the best free state equals $1+R_F(\\rho)$, where $R_F(\\rho)$ is the robustness relative to the free set $F$. The equality holds for every convex free set, extends to nonconvex free sets through their convex hull, and also holds when the discriminator is allowed an auxiliary system, provided the robustness is evaluated against $\\operatorname{co}F_C$, the convex hull of all states reachable from free states by local channels on that system. A reader should care because robustness is a computable convex-programming quantity, so this gives a universal operational meaning to it and closes an open question left by earlier work limited to entanglement, coherence, steering, and asymmetry.","feed_headline":"Robustness equals a quantum state's discrimination edge","feed_subtitle":"In any resource theory, this one number fixes the best success-probability ratio over free states, with or without ancillas.","key_machinery":"The load-bearing object is the generalized robustness $R_F(\\rho)=\\min\\{\\lambda\\ge 0:(\\rho+\\lambda\\tau)/(1+\\lambda)\\in F,\\ \\tau\\in\\mathrm{Den}_A\\}$, together with its Lagrange dual $R_F(\\rho)=\\max\\{\\operatorname{Tr}(x\\rho)-1: x\\ge 0,\\ \\operatorname{Tr}(x\\omega)\\le 1\\ \\forall\\omega\\in F\\}$. The lower-bound direction uses the dual optimizer $x^\\star$ to define an effect $e=x^\\star/\\|x^\\star\\|_\\infty$ and a family of group-covariant channels built from generalized Pauli-$X$ shifts; a symmetrization step ensures the success-probability ratio for this family collapses to $\\operatorname{Tr}(e\\rho)/\\sup_{\\omega\\in F}\\operatorname{Tr}(e\\omega)$, which is at least $1+R_F(\\rho)$ by duality. The auxiliary-system extension replaces the simple measurement by a maximally entangled state and a decomposition into subchannels, with the free-set condition guaranteeing that local operations on the auxiliary system keep free states free.","core_discovery":"On the paper's own terms, the central discovery is an identity: the discrimination power $G(\\rho,F)=\\sup_{\\{p_n,\\Lambda_n\\}}P_S(\\rho;\\{p_n,\\Lambda_n\\})/\\sup_{\\omega\\in F}P_S(\\omega;\\{p_n,\\Lambda_n\\})$ satisfies $G(\\rho,F)=1+R_F(\\rho)$ for every convex free-state set $F$, and $G(\\rho,F)=1+R_{\\operatorname{co}F}(\\rho)$ for every set $F$. When an auxiliary system $C$ is used, the statement becomes $G_C(\\rho,F)=1+R_{\\operatorname{co}F_C}(\\rho)$, where $F_C$ is the set obtained from $F$ by applying arbitrary local channels on $C$. The proof constructs explicit channel ensembles from the optimizer of a dual convex program, showing the bound is attainable, and it identifies the states that give no discrimination advantage, the bound resource states, as precisely the states in the convex hull of the free set.","pith_inferences":["An extension the paper leaves implicit is the multi-shot or adaptive setting: the proof here is one-shot, and a natural conjecture, which is my inference rather than the paper's, is that the corresponding long-run advantage would be governed by a regularized robustness.","The explicit dual-optimizer construction suggests an experimental recipe for saturating the bound with a finite channel ensemble whose size is set by the spectral norm of the optimizer; the paper does not spell this out.","For nonconvex resource theories, the theorem says that only the convex hull of the free set matters for this task, which implies that 'bound' resources in any such theory are exactly the states inside that hull; testing this against particular nonconvex resources such as magic or imaginarity would be a further application beyond the paper's own case studies."],"forward_implications":["In every convex resource theory, the generalized robustness is not merely a mathematical quantifier: it is the exact factor by which a resourceful state can outperform all free states in a one-shot channel-discrimination test.","A non-free state is useless for one-shot channel discrimination exactly when it lies in the convex hull of the free set; if the free set is convex and closed, every resource state is useful in some discrimination task.","Allowing an auxiliary system does not escape the robustness bound: the ancilla-assisted discrimination power is still $1+$ robustness, now computed against $\\operatorname{co}F_C$, so ancillas enlarge the relevant free hull rather than bypass the formula.","Because robustness is the optimal value of a convex program, the discrimination power can be obtained by standard convex-optimization methods in any finite-dimensional resource theory, not only in the special theories previously solved.","The equality $G_C(\\rho,F)=G(\\rho,F_C)$ gives a criterion for when ancillas help: the ancilla-assisted and bare discrimination powers coincide for every state exactly when $\\operatorname{co}F=\\operatorname{co}F_C$."],"supporting_citations":[{"why":"Supplies the Lagrange-dual formulation of generalized robustness and the existence of an optimal dual solution used to build the explicit channel family.","marker":"[49]"},{"why":"Shows every resource state helps in some subchannel-discrimination task and poses the open question about quantifying the advantage; this paper's Theorem 1 answers it for channel discrimination.","marker":"[41]"},{"why":"Gives the prior entanglement-specific result and a lower bound for Schmidt-number states with auxiliary systems that Theorem 3 tightens to an exact equality.","marker":"[42]"},{"why":"Provides the steering-specific robustness result for subchannel discrimination that Theorem 3 extends to a direct quantification for channel discrimination with an ancillary system.","marker":"[43]"},{"why":"One of the earlier special-case results identifying robustness with discrimination advantage for coherence, which the general theorem subsumes.","marker":"[44]"},{"why":"Extends the robustness-discrimination identification to asymmetry and coherence, another special case unified by the general result.","marker":"[46]"}],"fun_headline_variants":["Robustness measure sets channel discrimination advantage","Resource robustness gives exact success-probability boost","One robustness number quantifies all discrimination edge","Discrimination power equals robustness plus one","Robustness fixes channel discrimination gain exactly"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the standard strong-duality fact that the convex program defining generalized robustness attains its optimum for every closed convex free set containing a full-rank state, because the explicit channel ensemble is built from that optimizer; the auxiliary-system version also assumes the free set is preserved by local operations on the ancilla, or else the statement is read with the enlarged set $F_C$.","fun_headline_variants_meta":{"raw":{"variants":["Robustness measure sets channel discrimination advantage","Resource robustness gives exact success-probability boost","One robustness number quantifies all discrimination edge","Discrimination power equals robustness plus one","Robustness fixes channel discrimination gain exactly"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000298,"raw_usage":{"total_tokens":1684,"prompt_tokens":864,"completion_tokens":820,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":480,"completion_tokens_details":{"reasoning_tokens":754}},"tokens_in":480,"tokens_out":820,"duration_ms":6664,"temperature":1.0,"reasoning_tokens":754,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T20:11:17.990127+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $R_F(\\rho)$ for a small finite-dimensional convex free set and state by solving the dual program, construct the paper's channel ensemble from the optimizer, and check the measured success-probability ratio; a ratio strictly larger than $1+R_F(\\rho)$ would refute Theorem 1, while the theorem predicts the constructed ensemble reaches the bound.","supporting_citations":[{"cited_title":"Takagi and B","cited_arxiv_id":null,"evidence_quote":"Supplies the Lagrange-dual formulation of generalized robustness and the existence of an optimal dual solution used to build the explicit channel family."},{"cited_title":"Hsieh, B","cited_arxiv_id":null,"evidence_quote":"Gives the prior entanglement-specific result and a lower bound for Schmidt-number states with auxiliary systems that Theorem 3 tightens to an exact equality."},{"cited_title":"Takagi, B","cited_arxiv_id":null,"evidence_quote":"Provides the steering-specific robustness result for subchannel discrimination that Theorem 3 extends to a direct quantification for channel discrimination with an ancillary system."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"One of the earlier special-case results identifying robustness with discrimination advantage for coherence, which the general theorem subsumes."},{"cited_title":"Napoli, T","cited_arxiv_id":null,"evidence_quote":"Extends the robustness-discrimination identification to asymmetry and coherence, another special case unified by the general result."}],"review_version":1}