{"id":"05a09488-d1f3-41c2-bdcb-57c91d22fd36","arxiv_id":"2507.20797","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper proves an equivalence characterizing extreme points of tau-invariant UCP maps from a C*-algebra to B(H) in terms of Radon-Nikodym derivatives and faithful subspaces.","lead":"This paper characterizes the extreme points of unital completely positive maps that stay invariant under a partial group action, using dilation theory. The result completes the Choquet barycentric decomposition picture for these invariant map sets, a tool for decompositions in operator algebras and partial dynamical systems.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 3.5's converse never proves T(Y_g)⊆Y_g; in the trivial-action case this condition is actually false, making the stated domain too small and breaking the proof of (4)⇒(1).","rationale":"The reader's weakest_assumption identifies the same defect we find, so there is substantial agreement; my read is partial because I would state the issue more strongly: the missing invariance is not merely unproved, it is false in the trivial global-action case, so Lemma 3.5's statement needs correction. The main theorem is nevertheless plausible and likely repairable: Lemma 3.5's forward direction uses only intertwining, and the convex-decomposition arguments in Theorem 4.2 only need C^P_φ defined without T(Y_g)⊆Y_g. The paper otherwise gives a careful extension of the global-action result, and the Stinespring version (Lemma 3.3) is essentially correct because T(K_g)⊆K_g follows from the intertwining since U_g is onto K_g. No formal verification is present, and the proof is parameter-free; the issue is internal and specific. For these reasons the conditional verdict is appropriate, not rejection.","tokens_in":18319,"tokens_out":26039,"duration_ms":268890,"concrete_test":"Run the following specialization of Lemma 3.5: let G={e}, A=C, H=C^2, φ(λ)=λ I, and take the Paschke dilation constructed in Theorem 2.7 (X=A⊗B(H) modulo the null space, e=1⊗I, σ(λ)=left multiplication by λ). Compute eσ(A)' and Y=eσ(A)ehat, and list all T with 0≤T≤1, T(Y)⊆Y. If, as the standard construction gives, Y=C I and eσ(A)'≅B(C^2), then T(Y)⊆Y forces T scalar, while [0,φ]∩CPGτ contains λ↦λ diag(1,0). This disproves the stated surjectivity of Lemma 3.5; if instead the construction gives a larger Y, show directly from φ_T∈CPGτ that T(Y_g)⊆Y_g.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The main theorem hinges on Lemma 3.5, which claims an affine order isomorphism onto [0,φ]∩CPGτ(A,B(H)) from operators T∈eσ(A)' with 0≤T≤1, T(Y_g)⊆Y_g and T Wg=Wg T on Y_{g-1}. The converse direction proves only the intertwining identity. From T Wg=WgT on Y_{g-1} one can conclude T(Y_g)⊆Y_g only if one already knows T(Y_{g-1})⊆Y_{g-1}; that is exactly the kind of invariance being sought and is not a consequence of intertwining, because Y_g is not a Hilbert subspace and Wg is not unitary on all of X'. Consequently the operator T obtained from λφ1 in the proof of (4)⇒(1) need not belong to the set C^P_φ, whose definition uses the unproved condition. The gap is not cosmetic: for the trivial global action on A=C with H=C^2, the Paschke dilation has eσ(A)'≅B(C^2) and Y=C ehat, and the condition T(Y)⊆Y forces T to be scalar, whereas [0,φ]∩CPGτ contains all maps λ↦λa with 0≤a≤I. Thus Lemma 3.5 as stated is false, not merely missing a line. The remedy is to drop T(Y_g)⊆Y_g from the domain and from C^P_φ; the forward direction of Lemma 3.5 and the (4)⇒(1) computation go through using only intertwining. The parallel omission in Lemma 3.3 is harmless, since U_g maps K_{g-1} onto K_g and intertwining gives T(K_g)⊆K_g.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the compact convex set UCPGτ(A,B(H)) of unital completely positive maps from a unital C*-algebra A to B(H) that are invariant under a partial action τ of a group G on A. The main result, Theorem 4.2, characterizes the extreme points of this set through four equivalent conditions: (1) extremality in UCPGτ(A,B(H)); (2) a uniqueness property for elements of [0,φ]∩tUCPGτ(A,B(H)); (3) faithfulness of the subspace [VH] for a set C^S_φ built from the Stinespring representation; and (4) a null-space condition for a set C^P_φ built from the Paschke dilation. The proofs rely on two Radon-Nikodym type lemmas (Lemmas 3.3 and 3.5) obtained by restricting the classical Arveson and Paschke order isomorphisms to invariant maps. The paper concludes with a remark applying the Choquet and Choquet-Bishop-de Leeuw theorems to obtain barycentric decompositions.","tokens_in":18682,"tokens_out":28179,"duration_ms":274969,"significance":"If the technical gaps can be repaired, the result is a natural and useful generalization of Arveson's extreme point characterization to the setting of partial actions, complementing the earlier global-action work of Bhattacharya and Kulkarni. The paper is self-contained, the Stinespring-side argument is largely convincing, and the equivalence (1)⇔(2) is clean. The main issue is that the Paschke-side lemma on which condition (4) rests is not proved as stated and in fact appears to be false, so the central claim is currently conditional.","major_comments":[{"comment":"In the converse direction of Lemma 3.5 the proof establishes only the intertwining identity T fW_g = fW_g T on Y_{g-1}; it never verifies the domain condition T(Y_g)⊆Y_g. This condition is part of the stated domain of the alleged order isomorphism and of the set C^P_φ, and it is used in the proof of Theorem 4.2. The omission is not cosmetic: for the trivial action on A=C with H=C^2, φ(λ)=λ I, one has Y=C ehat and eσ(A)' is isomorphic to B(C^2), so T(Y)⊆Y forces T to be a scalar, whereas [0,φ]∩CPGτ contains all maps λ↦λa with 0≤a≤I. Thus Lemma 3.5 as stated is false, not merely missing a line of proof, and the proof of (4)⇒(1) in Theorem 4.2 rests on an invalid premise.","section":"Lemma 3.5, Equations (10)-(11), (13); Theorem 4.2"},{"comment":"The lemma and the set C^P_φ mix operators on X' with operators between the dual modules X'_g. The manuscript states Y_{g-1}⊆X'_{g-1} and T∈eσ(A)'⊆P(X'), then forms expressions such as T fW_g and fW_g^* T fW_g on Y_{g-1}. Since fW_g maps X'_{g-1} to X'_g and the paper never proves or even states an embedding of X'_g into X', these compositions are not well-defined as written. This is a load-bearing issue for the statement of Lemma 3.5 and for any corrected version of the argument; the authors need to specify the identification (for example, by using the self-duality of X' to realize X'_g as a complemented submodule of X').","section":"Lemma 3.5, Equations (9)-(11)"},{"comment":"In the converse direction of Lemma 3.3 the proof obtains T U_g = U_g T on K_{g-1} but does not verify the domain condition T(K_g)⊆K_g. Here the missing condition is recoverable: for x=U_g y with y∈K_{g-1}, one has T x = T U_g y = U_g T y ∈ K_g. This verification should be added, since the lemma as written claims an isomorphism whose domain includes T(K_g)⊆K_g.","section":"Lemma 3.3"}],"minor_comments":[{"comment":"The notation g-1 for the inverse g^{-1} is easy to misread; please use g^{-1} consistently.","section":"Throughout"},{"comment":"The set tUCPGτ(A,B(H)) should be defined explicitly, for example as {tψ : ψ∈UCPGτ(A,B(H))}.","section":"Theorem 4.2(2)"},{"comment":"There is a typo in 'becuase' in the paragraph establishing BW-compactness of UCPGτ(A,B(H)); please correct it.","section":"Section 4"},{"comment":"The step replacing fW_g^* T fW_g by fW_g^* fW_g T on Y_{g-1} implicitly assumes T(Y_{g-1}) is contained in the domain of fW_g; this is part of the definitional issue raised in the major comments and should be made explicit in any revision.","section":"Lemma 3.5, forward direction"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope and the Stinespring-side portion is in good shape. The main obstacle is the Paschke-side Lemma 3.5: as stated it is false, and the proof of Theorem 4.2 depends on it. I believe the result may be salvageable with a corrected lemma and a careful treatment of the spaces X'_g, but the current version cannot be accepted as is. I do not see a novelty or attribution problem."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper generalizes the global-action extreme point characterization of [4] to partial actions, and the main theorem is probably right. The construction of a partial action on the Stinespring and Paschke dilation spaces, together with the Radon–Nikodym isomorphisms adapted to the invariant subcollection, is the right toolset. The equivalences (1)⇔(2) and (1)⇔(3) are handled carefully and go through. The citation pattern is honest: the dependence on [3,4] is explicit and the novelty is not oversold.\n\nThe problem is Lemma 3.5. Its converse direction proves only the intertwining identity T fW_g = fW_g T on Y_{g-1}; it never establishes the stated domain condition T(Y_g)⊆Y_g. This is not a harmless omission. For the trivial action on A=C with H=C^2, the Paschke dilation has Y = C ehat, and the condition T(Y)⊆Y forces T to be scalar, while [0,φ]∩CPGτ contains all maps λ↦λa for 0≤a≤I. So Lemma 3.5 as stated is false, not merely missing a line.\n\nThe good news is that the fix is local and easy. The forward direction of Lemma 3.5 does not use T(Y_g)⊆Y_g. The computations in Theorem 4.2, including the crucial (4)⇒(1) step, only use the intertwining condition. So drop T(Y_g)⊆Y_g from the domain of Lemma 3.5 and from the definition of C_P^φ. The identity operator intertwines, so T−λ1 still lands in the corrected C_P^φ, and the (4)⇒(1) proof goes through. The parallel-looking condition in Lemma 3.3 is genuinely harmless, because U_g maps K_{g-1} onto K_g, so the intertwining already implies T(K_g)⊆K_g.\n\nThe Choquet framing is a little padded—they spend a page reminding us of Choquet's theorem before getting to the actual characterization—but that is cosmetic. The substance is a solid, if incremental, contribution to the partial-action literature. I would send it to peer review, with the instruction that the authors must restate Lemma 3.5 and C_P^φ without the Y-invariance condition and re-verify the main theorem. Once that is done, the result is believable and useful.","headline":"A natural partial-action extension of the global extreme-point result, with the right proof strategy but a false Lemma 3.5; the fix is to drop the Y-invariance condition, after which the main theorem holds.","tokens_in":19250,"tokens_out":9542,"would_cite":false,"duration_ms":93291,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46A55","46B22","46L55","46L08","47L07"],"pacs":[],"model":"deepseek-v4-flash","headline":"For unital completely positive maps invariant under a partial group action, the paper proves four equivalent operator-level tests for being an extreme point, completing the barycentric-decomposition picture.","keywords":["extreme points","completely positive maps","partial actions","barycentric decomposition","minimal Hilbert-space dilation","Hilbert-module dilation","operator-derivative order isomorphisms","operator algebras"],"falsifier":"The decisive check is the converse half of Lemma 3.5. Build a partial action, for instance $G=\\mathbb{Z}/2$ acting through a pair of ideals on a $C^*$-algebra, and look for $T\\in\\widetilde{\\sigma}(\\mathcal{A})'$ with $0\\le T\\le 1_{\\mathcal{X}'}$ and $T\\widetilde{W}_g=\\widetilde{W}_gT$ on $\\mathcal{Y}_{g-1}$ but $T(\\mathcal{Y}_g)\\not\\subseteq\\mathcal{Y}_g$. If the induced map $\\varphi_T$ is still invariant under the partial action, Lemma 3.5 is not an order isomorphism and Theorem 4.2(4) cannot be derived through it; in a two-dimensional example this is a finite matrix calculation.","tokens_in":18081,"feed_emoji":"🧮","tokens_out":18571,"duration_ms":148109,"temperature":0.7,"pith_summary":"The paper pinpoints exactly which unital completely positive maps from a $C^*$-algebra to the operators on a Hilbert space, invariant under a partial action of a group, are extreme points of that compact convex set. This matters because the classical barycentric-decomposition theorem decomposes any element of such a set into a measure supported on extreme points; knowing the extreme points makes that decomposition concrete. The main theorem states four equivalent characterizations: extremality, a scalar-multiple property for dominated invariant maps, a faithfulness condition on the cyclic subspace of the minimal Hilbert-space dilation, and a Hilbert-module annihilator condition. If the proof is right, the same abstract decomposition picture that works for global group actions now works when the action is only partial.","feed_headline":"Four equivalent tests identify extreme partially invariant UCP maps","feed_subtitle":"Barycentric decomposition of partially invariant unital completely positive maps gets a boundary test.","key_machinery":"The load-bearing machinery is a pair of derivative-type order isomorphisms adapted to partial actions. Lemma 3.3 works in the minimal Hilbert-space dilation and sends an admissible operator $T\\in\\pi(\\mathcal{A})'$ with $0\\le T\\le 1_{\\mathcal{K}}$, $T(\\mathcal{K}_g)\\subseteq\\mathcal{K}_g$, and $TU_g=U_gT$ on $\\mathcal{K}_{g-1}$ to the dominated invariant map $\\varphi_T(\\cdot)=V^*\\pi(\\cdot)TV$. Lemma 3.5 works in the Hilbert-module dilation and sends $T\\in\\widetilde{\\sigma}(\\mathcal{A})'$ with $0\\le T\\le 1_{\\mathcal{X}'}$, $T(\\mathcal{Y}_g)\\subseteq\\mathcal{Y}_g$, and $T\\widetilde{W}_g=\\widetilde{W}_gT$ on $\\mathcal{Y}_{g-1}$ to $\\varphi_T(a)=\\langle T\\widetilde{\\sigma}(a)\\hat e,\\hat e\\rangle$. The sets $\\mathcal{C}^S_\\varphi$ and $\\mathcal{C}^P_\\varphi$ are the real spans of these admissible operators, and conditions (3) and (4) of Theorem 4.2 assert that the only such operator that annihilates the cyclic subspace $[V\\mathcal{H}]$, respectively the cyclic vector $\\hat e$, is the zero operator.","core_discovery":"On the paper's own terms, the discovery is Theorem 4.2: for $\\varphi\\in\\mathrm{UCP}_G^\\tau(\\mathcal{A},\\mathcal{B}(\\mathcal{H}))$ with minimal Hilbert-space dilation $(\\pi,V,\\mathcal{K})$ and Hilbert-module dilation $(\\mathcal{X},\\sigma,e)$, the following are equivalent: (1) $\\varphi$ is extreme in the compact convex set $\\mathrm{UCP}_G^\\tau(\\mathcal{A},\\mathcal{B}(\\mathcal{H}))$; (2) every $\\varphi_0$ in $[0,\\varphi]\\cap t\\,\\mathrm{UCP}_G^\\tau(\\mathcal{A},\\mathcal{B}(\\mathcal{H}))$ with $0<t<1$ equals $t\\varphi$; (3) the cyclic subspace $[V\\mathcal{H}]$ is faithful for the operator set $\\mathcal{C}^S_\\varphi$; (4) every $T$ in $\\mathcal{C}^P_\\varphi$ satisfying $\\langle T\\hat e,\\hat e\\rangle=0$ must be zero. Here $\\mathcal{C}^S_\\varphi$ and $\\mathcal{C}^P_\\varphi$ are the real linear spans of the admissible derivative operators from Lemmas 3.3 and 3.5. Extremality of an invariant UCP map is thus reduced to checking operator conditions in the commutant of the dilating representation.","pith_inferences":["Editorial inference: the unproved step $T(\\mathcal{Y}_g)\\subseteq\\mathcal{Y}_g$ in Lemma 3.5 means Theorem 4.2(4) should be read as conditional on that invariance; a repair would either prove the inclusion from the other hypotheses or explicitly add it as a hypothesis in the characterization.","Editorial inference: if that gap is closed, the same four-condition format should transfer to invariant completely positive maps valued in a general von Neumann algebra, since both operator-derivative theorems are module-theoretic.","Editorial inference: a concrete two-cycle example ($G=\\mathbb{Z}/2$ with a nontrivial partial action) would settle whether the missing invariance is automatic; the paper provides no such test."],"forward_implications":["If Theorem 4.2 is correct, the barycentric decomposition in $\\mathrm{UCP}_G^\\tau(\\mathcal{A},\\mathcal{B}(\\mathcal{H}))$ is explicit: every invariant UCP map is an integral of extreme invariant UCP maps whenever $\\mathcal{A}$ is separable.","The nonseparable case inherits the same boundary description through the non-metrizable form of the decomposition theorem: the decomposition measure vanishes on every Baire set disjoint from the extreme points.","Extremality can be tested in either dilation: faithfulness of $[V\\mathcal{H}]$ in the Hilbert-space dilation is equivalent to the annihilator condition in the Hilbert-module dilation.","The scalar-multiple condition (2) gives a direct criterion for checking extremality of a concrete invariant map by looking only at dominated invariant maps."],"supporting_citations":[{"why":"Supplies the operator-derivative order isomorphism for the Hilbert-space dilation and the definition of faithful subspace used in condition (3).","marker":"[2]"},{"why":"Supplies the Hilbert-module dilation and its operator-derivative isomorphism used to build condition (4).","marker":"[14]"},{"why":"Supplies the minimal dilation theorem and the complete boundedness facts used throughout the proof.","marker":"[15]"},{"why":"Supplies the two decomposition theorems that make the extreme-point characterization useful.","marker":"[16]"},{"why":"Supplies the definition of partial action on a C*-algebra used to form the invariant sets and the induced partial actions on the dilation spaces.","marker":"[9]"},{"why":"Provides the proof technique for the main theorem via dominated invariant maps.","marker":"[3]"},{"why":"Provides the prior global-action version of the characterization and the uniqueness of the Hilbert-module dilation.","marker":"[4]"}],"fun_headline_variants":["Four conditions characterize extreme UCP maps under partial actions","Extreme points of invariant UCP maps: four equivalent tests","Barycentric boundary: extreme partially invariant UCP maps","Four-way test for extremality of invariant UCP maps","Equivalent criteria for extreme invariant UCP maps"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The main theorem assumes that Lemma 3.5 is an order isomorphism onto the dominated invariant maps, but the converse half of that lemma never proves the required invariance $T(\\mathcal{Y}_g)\\subseteq\\mathcal{Y}_g$ (and Lemma 3.3 similarly omits the proof of $T(\\mathcal{K}_g)\\subseteq\\mathcal{K}_g$, though there it follows from the intertwining relation and unitarity); if such an invariance fails, the operator set used in condition (4) is not the right one.","fun_headline_variants_meta":{"raw":{"variants":["Four conditions characterize extreme UCP maps under partial actions","Extreme points of invariant UCP maps: four equivalent tests","Barycentric boundary: extreme partially invariant UCP maps","Four-way test for extremality of invariant UCP maps","Equivalent criteria for extreme invariant UCP maps"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000602,"raw_usage":{"total_tokens":2811,"prompt_tokens":945,"completion_tokens":1866,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":561,"completion_tokens_details":{"reasoning_tokens":1788}},"tokens_in":561,"tokens_out":1866,"duration_ms":12040,"temperature":1.0,"reasoning_tokens":1788,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:43:44.405513+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The decisive check is the converse half of Lemma 3.5. Build a partial action, for instance $G=\\mathbb{Z}/2$ acting through a pair of ideals on a $C^*$-algebra, and look for $T\\in\\widetilde{\\sigma}(\\mathcal{A})'$ with $0\\le T\\le 1_{\\mathcal{X}'}$ and $T\\widetilde{W}_g=\\widetilde{W}_gT$ on $\\mathcal{Y}_{g-1}$ but $T(\\mathcal{Y}_g)\\not\\subseteq\\mathcal{Y}_g$. If the induced map $\\varphi_T$ is still invariant under the partial action, Lemma 3.5 is not an order isomorphism and Theorem 4.2(4) cannot be derived through it; in a two-dimensional example this is a finite matrix calculation.","supporting_citations":[{"cited_title":"B.Subalgebras of C∗-algebras","cited_arxiv_id":null,"evidence_quote":"Supplies the operator-derivative order isomorphism for the Hilbert-space dilation and the definition of faithful subspace used in condition (3)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Hilbert-module dilation and its operator-derivative isomorphism used to build condition (4)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the minimal dilation theorem and the complete boundedness facts used throughout the proof."},{"cited_title":"Lectures on Choquet’s theorem","cited_arxiv_id":null,"evidence_quote":"Supplies the two decomposition theorems that make the extreme-point characterization useful."},{"cited_title":"Partial Dynamical Systems, Fell Bundles and Applications , Amer","cited_arxiv_id":null,"evidence_quote":"Supplies the definition of partial action on a C*-algebra used to form the invariant sets and the induced partial actions on the dilation spaces."},{"cited_title":"and Kulkarni, C","cited_arxiv_id":null,"evidence_quote":"Provides the proof technique for the main theorem via dominated invariant maps."},{"cited_title":"and Kulkarni, C","cited_arxiv_id":null,"evidence_quote":"Provides the prior global-action version of the characterization and the uniqueness of the Hilbert-module dilation."}],"review_version":2}