{"id":"5ace8a90-3880-45b7-ab98-482ce0a38d63","arxiv_id":"2412.05008","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For finite-dimensional Hilbert spaces, the C*-extreme points of contractive completely positive maps are exactly the P-C*-extreme points for projections P, giving a complete nested-direct-sum structure.","lead":"This paper introduces a weighted version of quantum convexity, called P-C*-convexity, controlled by a fixed positive operator P, and studies its extreme points. Its main result fully describes the extreme points of the space of contractive completely positive maps between finite-dimensional spaces, completing a 25-year program.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the finite-dimensional characterization in Theorem 4.12 is internally consistent, and the key external projection lemma used there is sound.","rationale":"I read the paper in good faith, focusing on the strongest claim: Theorem 4.12, combined with Theorem 3.20, gives a complete structural description of C*-extreme points of CCP(A,B(H)) for finite-dimensional H. I traced every step that uses finite-dimensionality: Lemma 4.10 (closed range is automatic), Corollary 3.16 (block-triangular form), and Proposition 3.17(ii) (range-reduced maps). All are valid in the stated finite-dimensional setting. The proof of Theorem 4.12 also depends on an external projection-lemma from [LoPa81, Prop 26] and [Wei02]; I examined this step and found that the needed statement follows from a direct argument: the contraction inequalities force each Q_j to be the orthogonal projection onto T_j(range P), with kernel T_j(ker P), hence all Q_j have the same rank as P and are unitarily equivalent to P. Therefore the inclusion of CP(P)_{C*-ext} in CCP_{C*-ext} is justified. The converse direction in Theorem 4.12 is also sound, using Proposition 3.17 and Lemma 4.11 to convert invertible equivalences to unitary equivalences. The paper honestly flags the infinite-dimensional analogue as open in Note 4.14, so the finite-dimensional scope is explicit and not a hidden assumption. I found no internal inconsistency, no unproved core step beyond standard external results that are themselves correct, and no reason to question the reader's ACCEPT verdict. The only further check worth running is an independent derivation of the projection lemma, which would remove all doubt about the one non-obvious external input.","tokens_in":25842,"tokens_out":25025,"duration_ms":434282,"concrete_test":"State and prove the projection lemma used in Theorem 4.12: if P is a projection, T_1,T_2 are invertible with T_1^*T_1+T_2^*T_2=I, and P=T_1^*Q_1T_1+T_2^*Q_2T_2 for 0≤Q_j≤I, then Q_j is unitarily equivalent to P. A computational spot-check in dimension 3 with rank(P)=2 and rank(Q_j)=1 should be impossible; if such Q_j and invertible T_j exist, the first inclusion of Theorem 4.12 fails and the main characterization would need revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of Theorem 4.12 is finite-dimensional and the proof is internally consistent. Lemma 4.10's closed-range hypothesis is automatic when dim H<∞, so P=Φ(1) is a projection. The reduction in Proposition 3.17(ii) relies only on Corollary 3.16, which is valid in finite dimensions. The one genuinely external step is the first inclusion of Theorem 4.12: from a proper C*-convex decomposition Φ=AdT1Φ1+AdT2Φ2 in CCP with P=Φ(1) a projection and Φ∈CP(P)_{C*-ext}, the proof invokes [LoPa81, Prop 26]/[Wei02] to obtain unitaries Uj with Φ_j(1)=U_j^*PU_j. I checked the needed fact directly: for P a projection, T_j invertible with ∑T_j^*T_j=I, and P=∑T_j^*Q_jT_j with 0≤Q_j≤I, letting R_j=Q_j^{1/2}T_j, the equality ∑||R_jx||^2=||x||^2 for x∈range P forces Q_jT_jx=T_jx, while y∈ker P gives Q_jT_jy=0; injectivity of T_j then forces range Q_j=T_j(rangeP) and ker Q_j=T_j(kerP), with these subspaces orthogonal, so Q_j is a projection of rank rank(P), i.e., unitarily equivalent to P. Thus the external step is sound. No critical flaw identified.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a generalization of C*-convexity called P-C*-convexity, for a positive operator P on a Hilbert space H, and studies the corresponding extreme points in the sets CP(P)(A,B(H)) of completely positive maps with Φ(1)=P. The main contributions are an abstract characterization of P-C*-extreme points (Theorem 3.12), a structure theorem for finite-dimensional H (Theorem 3.20), a Krein-Milman type theorem for CP(P) (Theorem 3.25), and the central application: for finite-dimensional H, the C*-extreme points of the contractive CP maps CCP(A,B(H)) are exactly the union over projections P of the P-C*-extreme points of CP(P)(A,B(H)) (Theorem 4.12). The paper also proves a Krein-Milman theorem for CCP and shows that, in finite dimensions, C*-extreme points of CCP are linear extreme points.","tokens_in":26162,"tokens_out":34960,"duration_ms":341511,"significance":"If the results are correct, the paper gives a complete structural description of C*-extreme contractive completely positive maps into matrix algebras, extending the Farenick-Zhou theory from unital maps to the contractive setting. The main theorem is concrete and checkable, and the proofs are carried out with detailed block-matrix arguments. The authors are also appropriately cautious: Note 4.14 explicitly flags that the infinite-dimensional analogue of the main characterization remains open, and the finite-dimensionality hypotheses in Corollary 3.16, Proposition 3.17, and Lemma 4.10 are made clear. I verified the external projection fact used in the first half of Theorem 4.12, and it is sound. The paper should be of interest to researchers in quantized convexity, completely positive maps, and operator algebras.","major_comments":[],"minor_comments":[{"comment":"The statement uses the decomposition H = H0 ⊕ H0^⊥ with H0 = range(P). This is not valid unless range(P) is closed. The argument and all later finite-dimensional applications go through if H0 is replaced by the closure of range(P); please revise the statement and proof accordingly.","section":"Lemma 3.14"},{"comment":"The stated equivalence is false in the 'only if' direction: if B is a rank-one projection and C is an invertible rotation, A=BC has range(A)=range(B) but ker(A)≠ker(B). The 'if' direction is the one used in the paper, so the theorem should be restated or restricted to positive operators.","section":"Theorem 3.7"},{"comment":"The proof explicitly treats only the case where P is not invertible; the invertible case is not written out. Please add a sentence covering P invertible, since it is needed for the full statement of the theorem.","section":"Theorem 3.20"},{"comment":"The appeal to Lemma 4.3 is not immediate because the decomposition in the example is a scalar convex combination rather than a C*-convex one. The conclusion s=t follows by comparing the two scalar multiples and using that both summands lie in CCP×; please clarify the argument.","section":"Example 4.7(i)"},{"comment":"In the converse direction, the case P=Φ(1)=0 is not addressed. The zero map is indeed in CP(0)C*-ext, but this case should be stated explicitly.","section":"Theorem 4.12"},{"comment":"There are several typographical errors, including 'the the' and 'defined with in' in the abstract; the reference [Zhu98] also lists the author as H. Zhuo, which should be corrected.","section":"General presentation"}],"recommendation":"minor_revision","confidential_remarks":"I agree with the positive assessment of the finite-dimensional main theorem. The issues I found are local and do not affect the central claim, but the statement of Lemma 3.14 and Theorem 3.7 should be corrected before publication. The paper is well within the scope of math.OA and the finite-dimensional focus is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth reading and worth refereeing. The paper completes the finite-dimensional half of a long-running program: it characterizes the C*-extreme points of contractive completely positive maps into B(H) for finite-dimensional H. The key new device is P-C*-convexity, where the usual C*-convex combination condition is weighted by a positive operator P. The main results are Theorem 3.12 (abstract characterization of P-C*-extreme points), Theorem 3.20 (structure as invertible conjugates of nested direct sums of pure UCP maps plus a zero block), and Theorem 4.12, which says that for finite-dimensional H the C*-extreme points of CCP(A,B(H)) are exactly the union over projections P of the P-C*-extreme points of CP(P). That is a clean completion of the Farenick-Zhou classification from the unital to the contractive setting.\n\nThe proof strategy adapts known tools, but the adaptation is real and nontrivial, particularly the block-matrix arguments in Section 3 and Lemma 4.10 showing that a C*-extreme point of CCP must have Φ(1) equal to a projection. I also checked the one truly external step in Theorem 4.12, the use of [LoPa81, Prop 26] to promote Φ_j(1)=U_j^*PU_j to unitaries. The stress-test note verifies it directly; the verification is sound, and the conclusion holds. I found no error affecting the central claim.\n\nSoft spots: the finite-dimensional restriction is essential and is honestly flagged in Note 4.14; the infinite-dimensional analogue remains open. That limits the reach of the main theorem, but it is not a flaw in the proof. The paper is also dense and technical; a reader has to work through several auxiliary results before seeing the main structure. The relation between C*-extreme and linear extreme points is handled, but it is not the main event. Citation practice is fine: the authors use standard sources and do not lean on their own prior work.\n\nThis paper is for operator algebraists with a serious interest in C*-convexity and extreme points of CP maps. The classification theorem will be a useful reference, and P-C*-convexity might be reusable in other problems. My recommendation: send it to a competent referee, accept after minor revision. The math is correct, the novelty is solid, and the paper is honest about its scope.","headline":"A careful, correct finite-dimensional completion of the Farenick-Zhou program for contractive CP maps via a genuinely new P-C*-convexity framework.","tokens_in":26690,"tokens_out":3691,"would_cite":true,"duration_ms":33095,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L05","46L07","46L30"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that for finite-dimensional Hilbert spaces, the C*-extreme points of contractive completely positive maps are exactly the maps $\\Phi$ for which $\\Phi(1)$ is a projection and the nonzero block is a C*-extreme unital…","keywords":["C*-algebra","completely positive map","C*-convexity","C*-extreme point","P-C*-convexity","contractive completely positive map","Krein-Milman theorem"],"falsifier":"Search small finite-dimensional examples for a C*-extreme contractive map $\\Phi$ with $\\Phi(1)$ not a projection; the theorem asserts none exists. For instance, with $\\mathcal{A}=M_2$ and $\\mathcal{H}=\\mathbb{C}^3$, any $\\Phi$ with $\\Phi(1)=\\operatorname{diag}(1,\\tfrac12,0)$ must admit a proper C*-convex decomposition into two summands not unitarily equivalent to $\\Phi$, and Lemma 4.10 gives the explicit decomposition to check. Finding one map for which every such decomposition still forces unitary equivalence would refute Theorem 4.12.","tokens_in":2321,"feed_emoji":"📐","tokens_out":3071,"duration_ms":126685,"temperature":0.7,"pith_summary":"This paper gives a complete classification of the C*-extreme points of the set of contractive completely positive maps from a unital C*-algebra into $B(\\mathcal{H})$, when $\\mathcal{H}$ is finite-dimensional. A C*-extreme point here is an extreme point with respect to quantum convex combinations, where the scalar weights are replaced by operators $T_j$ satisfying $\\sum_j T_j^*T_j = I$. The paper proves that a contractive map $\\Phi$ is C*-extreme exactly when $\\Phi(1)$ is a projection and the restriction to the range of $\\Phi(1)$ is a C*-extreme unital completely positive map; equivalently, $\\Phi$ is a unitary conjugate of a direct sum of nested pure unital completely positive maps padded with a zero block. To reach this, it introduces $P$-C*-convexity, a generalized quantized convexity indexed by a positive operator $P$, and studies the extreme points of the affine slice $\\mathrm{CP}^{(P)}(A,B(\\mathcal{H})) = \\{\\Phi : \\Phi(1)=P\\}$. This generalizes the earlier C*-convexity theory for unital maps and yields Krein-Milman-type theorems asserting that the whole sets are generated, in the bounded-weak topology, by their respective extreme points.","feed_headline":"C*-extreme contractive maps are zero-padded unital ones","feed_subtitle":"In finite dimensions, a contractive CP map is C*-extreme exactly when its value at 1 is a projection.","key_machinery":"The machinery is $P$-C*-convexity: with $P \\in B(\\mathcal{H})_+$, a combination $\\sum_j \\mathrm{Ad}_{T_j} \\Phi_j$ with $\\sum_j T_j^* P T_j = P$ is a $P$-C*-convex combination, and $\\mathrm{CP}^{(P)}(A,B(\\mathcal{H})) = \\{\\Phi \\in \\mathrm{CP} : \\Phi(1)=P\\}$ carries this structure. The load-bearing facts are the abstract characterization (Theorem 3.12): $\\Phi$ is $P$-C*-extreme iff every CP map $\\Psi \\le_{cp} \\Phi$ with $\\Psi(1) = B^* P B$ for invertible $B$ satisfies $\\Psi = \\mathrm{Ad}_Z \\Phi$ for some invertible $Z$; the invertible-conjugation theorem (Theorem 3.19) transferring $P$-extremality to UCP-extremality; and the block-triangular reduction (Proposition 3.17 and Corollary 3.16) that in finite dimensions strips away the zero block on $\\mathrm{range}(P)^\\perp$. Together they convert the classification of contractive C*-extreme maps into the known classification of unital C*-extreme maps.","core_discovery":"The central discovery is that, inside the finite-dimensional contractive set, C*-extremality is a projection phenomenon. Theorem 4.12 proves that $\\Phi \\in \\mathrm{CCP}(A,B(\\mathcal{H}))$ is C*-extreme if and only if $P := \\Phi(1)$ is a projection and $\\Phi$ is a $P$-C*-extreme point of the slice $\\mathrm{CP}^{(P)}(A,B(\\mathcal{H}))$. By the structural theorem Theorem 3.20, this means $\\Phi$ is unitarily equivalent to a block map $\\left(\\bigoplus_{i,j} \\Phi^{\\pi_i}_j\\right) \\oplus 0$ with respect to $\\mathcal{H} = \\left(\\bigoplus_{i,j} \\mathcal{H}^i_j\\right) \\oplus \\mathrm{range}(P)^\\perp$, where each $\\Phi^{\\pi_i}_j$ is a pure unital completely positive map forming a nested sequence of compressions of an irreducible representation $\\pi_i$. The argument passes through the general theory of $P$-C*-convexity: for invertible $P$, $\\Phi$ is $P$-C*-extreme exactly when its normalized version $\\widehat{\\Phi} = P^{-1/2}\\Phi(\\cdot)P^{-1/2}$ is a C*-extreme unital map, and the finite-dimensional reduction strips away the zero block to reduce non-invertible $P$ to this case.","pith_inferences":["The pivotal open step for infinite dimensions is proving that every C*-extreme contractive map has closed range at 1; with that, Lemma 4.10 would force $\\Phi(1)$ to be a projection and the padded-unital structure would follow for arbitrary Hilbert spaces.","The same $P$-C*-convexity machinery could classify extremality in other affine slices, such as maps with $\\Phi(1)$ equal to a fixed positive contraction or a fixed positive element of a general C*-algebra.","In quantum information terms, the structure says C*-extreme contractive maps act as a unital channel on one subspace and are exactly zero on the complement, a noiseless-subsystem-plus-dark-subspace pattern that could be tested on small matrix examples.","If the finite-dimensional classification is combined with the known structure of C*-extreme unital maps for infinite-dimensional separable settings, a plausible conjecture is that the same union-over-projections formula holds whenever $\\Phi(1)$ has closed range."],"forward_implications":["In finite dimensions, C*-extreme contractive completely positive maps are automatically linear extreme points of $\\mathrm{CCP}(A,B(\\mathcal{H}))$.","A Krein-Milman-type theorem holds: when $\\mathcal{H}$ is finite-dimensional, the C*-convex hull of the C*-extreme points of $\\mathrm{CCP}(A,B(\\mathcal{H}))$ is BW-dense in $\\mathrm{CCP}(A,B(\\mathcal{H}))$.","For commutative $A$, the C*-extreme contractive maps are precisely the $*$-homomorphisms from $A$ into $B(\\mathcal{H})$.","For every positive $P$ with finite-dimensional $\\mathcal{H}$, the set $\\mathrm{CP}^{(P)}(A,B(\\mathcal{H}))$ is the BW-closure of the $P$-C*-convex hull of its $P$-C*-extreme points.","When $P$ is invertible, $P$-C*-extremality is equivalent to unital C*-extremality after conjugation, so the contractive classification inherits the nested-compression structure of unital extreme maps."],"supporting_citations":[{"why":"Introduced C*-convexity and C*-extreme points for UCP maps, proved the finite-dimensional inclusion into linear extreme points, and supplied the unital base case generalized throughout the paper.","marker":"[FaMo97]"},{"why":"Provided the structural theorem for C*-extreme UCP maps over finite-dimensional Hilbert spaces as direct sums of nested pure UCP compressions, used directly in Theorems 3.20 and 4.12.","marker":"[FaZh98]"},{"why":"Gave the abstract characterization of C*-extreme UCP points that Theorem 3.12 generalizes to the $P$-C*-convex setting.","marker":"[Zhu98]"},{"why":"Supplied the Radon-Nikodym theorem for CP maps and the purity criterion, which underpin the proof of Theorem 3.12 and Proposition 3.13.","marker":"[Arv69]"},{"why":"Corrected the abstract characterization and extended the study of C*-extreme UCP maps and nests, used in Theorem 3.25 and in framing the missing infinite-dimensional steps.","marker":"[BhKu]"},{"why":"Defined C*-convexity and provided Proposition 26, used in Theorem 4.12 to force the summands' values at 1 to be unitarily equivalent to the projection $P$.","marker":"[LoPa81]"},{"why":"Douglas's range-inclusion lemma manufactures the invertible factorizations needed in Lemmas 3.2, 3.8, and Proposition 3.11.","marker":"[Dou66]"},{"why":"The operator range factorization theorem converts range equalities into invertible operators in Lemma 3.8 and Corollary 3.16.","marker":"[FiWi71]"}],"fun_headline_variants":["C*-extreme contractive maps: zero-padded pure unital blocks","Contractive CP maps: C*-extreme iff identity goes to projection","Zero-padded pure unital maps are the C*-extreme contractive CP maps","C*-extremality in contractive CP maps: projection at identity decides"],"cache_read_input_tokens":28800,"weakest_assumption_plain":"The classification rests on finite-dimensionality of the target Hilbert space $\\mathcal{H}$; the infinite-dimensional analogue is explicitly left open, and the proof uses finite-dimensionality to obtain block-triangular forms, closed range of $\\Phi(1)$, and reduction to invertible $P$.","fun_headline_variants_meta":{"raw":{"variants":["C*-extreme contractive maps: zero-padded pure unital blocks","Contractive CP maps: C*-extreme iff identity goes to projection","Zero-padded pure unital maps are the C*-extreme contractive CP maps","C*-extremality in contractive CP maps: projection at identity decides"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000888,"raw_usage":{"total_tokens":3939,"prompt_tokens":1162,"completion_tokens":2777,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":778,"completion_tokens_details":{"reasoning_tokens":2694}},"tokens_in":778,"tokens_out":2777,"duration_ms":21025,"temperature":1.0,"reasoning_tokens":2694,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T21:00:16.224683+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search small finite-dimensional examples for a C*-extreme contractive map $\\Phi$ with $\\Phi(1)$ not a projection; the theorem asserts none exists. For instance, with $\\mathcal{A}=M_2$ and $\\mathcal{H}=\\mathbb{C}^3$, any $\\Phi$ with $\\Phi(1)=\\operatorname{diag}(1,\\tfrac12,0)$ must admit a proper C*-convex decomposition into two summands not unitarily equivalent to $\\Phi$, and Lemma 4.10 gives the explicit decomposition to check. Finding one map for which every such decomposition still forces unitary equivalence would refute Theorem 4.12.","supporting_citations":[],"review_version":1}