OctoInformBeta

OpenAI preprint, September 2026 · §5. From a cone map with a full corner to a dimension-drop map

5.3 Full cone map yields Rordam-Winter elements and a dimension-drop map: Unital **-homomorphisms into finite maximal tensor powers via amplified Cuntz comparison for a Pauli matrix system of rank at least two

Dimension-drop maps from cone maps with full corners.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

Take an integer rank N≥2N \ge 2, a unital C*C^*-algebra PP, and a unital **-homomorphism from the cone over N×NN \times N matrices into PP sending the upper-left corner to a full element. Does such a map produce a unital **-homomorphism from the dimension-drop algebra for sizes NN and N+1N + 1 into a finite maximal tensor power of PP? These notes go through that theorem and the constructions and comparison results it is built on.

Pauli Systems and Dimension-drop Algebras

A positive dimension NN is a natural number satisfying N>0N > 0.✦ For an index set ι\iota, the complex matrix algebra Mι(ℂ)M_\iota(\mathbb{C}) consists of square complex matrices indexed by ι\iota✦, and for indices i,j∈ιi, j \in \iota, the algebraic matrix unit ei,j∈Mι(ℂ)e_{i,j} \in M_\iota(\mathbb{C}) is the matrix with entry 11 in row ii and column jj and 00 elsewhere.✦ Given a positive dimension NN, the matrix C*C^*-algebras MNM_N, MN+1M_{N+1}, and MN(N+1)M_{N(N+1)} are M{0,…,N−1}(ℂ)M_{\{0, \dots, N - 1\}}(\mathbb{C}), M{0,…,N}(ℂ)M_{\{0, \dots, N\}}(\mathbb{C}), and M{0,…,N−1}×{0,…,N}(ℂ)M_{\{0, \dots, N - 1\} \times \{0, \dots, N\}}(\mathbb{C}), with identity matrices 1N∈MN1_N \in M_N and 1N+1∈MN+11_{N+1} \in M_{N+1}.

When N≥2N \ge 2, a Pauli matrix system of rank NN consists of a finite index set LL of cardinality N2N^2 with a distinguished zero label 0∈L0 \in L and a family (ea)a∈L(e_a)_{a \in L} in MNM_N satisfying e0=1Ne_0 = 1_N, ea*=eae_a^* = e_a, and ea2=1Ne_a^2 = 1_N for every a∈La \in L, the trace relation Tr⁡(eaeb)=N\operatorname{Tr}(e_a e_b) = N if a=ba = b and 00 if a≠ba \neq b for all a,b∈La, b \in L, and the condition that for every a≠0a \neq 0 in LL there exists b∈MNb \in M_N with b2=1Nb^2 = 1_N and bea=−eabb e_a = -e_a b.✦

The left matrix inclusion ιL:MN→MN(N+1)\iota_L : M_N \to M_{N(N+1)} is the unital **-homomorphism sending a∈MNa \in M_N to the Kronecker product a⊗1N+1a \otimes 1_{N+1}✦, and the right matrix inclusion ιR:MN+1→MN(N+1)\iota_R : M_{N+1} \to M_{N(N+1)} is the unital **-homomorphism sending b∈MN+1b \in M_{N+1} to 1N⊗b1_N \otimes b.✦ For t∈[0,1]t \in [0, 1], point evaluation evt:C([0,1],MN(N+1))→MN(N+1)\mathrm{ev}_t : C([0, 1], M_{N(N+1)}) \to M_{N(N+1)} on the C*C^*-algebra C([0,1],MN(N+1))C([0, 1], M_{N(N+1)}) of continuous MN(N+1)M_{N(N+1)}-valued paths on [0,1][0, 1] is the unital **-homomorphism sending each path ff to f(t)f(t).✦ The dimension-drop algebra I(N,N+1)I(N, N + 1) is the unital **-subalgebra ev0−1(ιL(MN))∩ev1−1(ιR(MN+1))\mathrm{ev}_0^{-1}(\iota_L(M_N)) \cap \mathrm{ev}_1^{-1}(\iota_R(M_{N+1})) of C([0,1],MN(N+1))C([0, 1], M_{N(N+1)}), recovering I(2,3)I(2, 3) when N=2N = 2.✦

For a unital C*C^*-algebra DD, write ⊗max\otimes_{\max} for the maximal C*C^*-tensor product. For m≥0m \ge 0, the right-nested maximal tensor power D⊗max(m+1)D^{\otimes_{\max}(m+1)} is the unital C*C^*-algebra defined recursively by D⊗max1=DD^{\otimes_{\max} 1} = D and D⊗max(m+2)=D⊗maxD⊗max(m+1)D^{\otimes_{\max}(m+2)} = D \otimes_{\max} D^{\otimes_{\max}(m+1)}.✦

The Matrix Cone and Cut-off Elements

For t∈[0,1]t \in [0, 1], path evaluation πt:C([0,1],MN)→MN\pi_t : C([0, 1], M_N) \to M_N on the C*C^*-algebra C([0,1],MN)C([0, 1], M_N) of continuous MNM_N-valued paths on [0,1][0, 1] is the unital **-homomorphism mapping each path ff to f(t)f(t).✦ The matrix cone C*C^*-algebra 𝒞N\mathcal{C}_N is the unital **-subalgebra π0−1(ℂ⋅1N)\pi_0^{-1}(\mathbb{C} \cdot 1_N) of C([0,1],MN)C([0, 1], M_N), recovering the M2M_2-cone 𝒞2\mathcal{C}_2 when N=2N = 2.✦

The algebra C([0,1])C([0, 1]) is the C*C^*-algebra of continuous complex-valued functions on [0,1][0, 1], and the algebra of functions vanishing at zero, C0((0,1])C_0((0, 1]), is the C*C^*-subalgebra of C([0,1])C([0, 1]) consisting of functions vanishing at 00. The algebra of matrices vanishing at zero, MN(C0((0,1]))M_N(C_0((0, 1])), consists of N×NN \times N matrices indexed by {0,…,N−1}\{0, \dots, N - 1\} with entries in C0((0,1])C_0((0, 1]).✦ The matrix path coefficient map κ:MN(C0((0,1]))→C([0,1],MN)\kappa : M_N(C_0((0, 1])) \to C([0, 1], M_N) is the non-unital **-homomorphism sending h=(hij)h = (h_{ij}) to the path s↦(hij(s))s \mapsto (h_{ij}(s)).✦ The unitized matrix path coefficient map κ̃:MN(C0((0,1]))∼→C([0,1],MN)\tilde{\kappa} : M_N(C_0((0, 1]))^\sim \to C([0, 1], M_N) is the canonical unital **-homomorphic extension of κ\kappa to the unitization MN(C0((0,1]))∼M_N(C_0((0, 1]))^\sim✦, and the cone identification map θ:MN(C0((0,1]))∼→𝒞N\theta : M_N(C_0((0, 1]))^\sim \to \mathcal{C}_N is the unital **-homomorphism obtained by restricting the codomain of κ̃\tilde{\kappa} to 𝒞N\mathcal{C}_N.✦

The coordinate function t∈C([0,1])t \in C([0, 1]) is the continuous function sending each s∈[0,1]s \in [0, 1] to ss✦, and coordinate multiplication μt:C([0,1])→C0((0,1])\mu_t : C([0, 1]) \to C_0((0, 1]) is the ℂ\mathbb{C}-linear map sending each f∈C([0,1])f \in C([0, 1]) to the function s↦sf(s)s \mapsto s f(s).✦ For i,j∈{0,…,N−1}i, j \in \{0, \dots, N - 1\}, the corner element tei,j∈𝒞Nt e_{i,j} \in \mathcal{C}_N is the image under θ\theta of the matrix in MN(C0((0,1]))M_N(C_0((0, 1])) having μt(1)\mu_t(1) in row ii and column jj and 00 elsewhere.✦

For δ∈ℝ\delta \in \mathbb{R}, the cut-off function (t−δ)+∈C([0,1])(t - \delta)_+ \in C([0, 1]) is the continuous function sending each s∈[0,1]s \in [0, 1] to max⁡(s−δ,0)\max(s - \delta, 0).✦ When δ≥0\delta \ge 0, the cut-off element qδ∈𝒞Nq_\delta \in \mathcal{C}_N is the path sending s∈[0,1]s \in [0, 1] to (t−δ)+(s)e0,0(t - \delta)_+(s) e_{0,0}.✦ Evaluating the cut-off element at δ=0\delta = 0 gives q0=te0,0.q_0 = t e_{0,0}.(1)✦

Fullness and Cuntz Comparison

An element aa of a unital C*C^*-algebra AA is full if the two-sided ideal of AA generated by aa equals AA.✦ If cc is a positive full element of a unital C*C^*-algebra AA, then there exist a natural number kk and elements v0,…,vk−1∈Av_0, \dots, v_{k-1} \in A such that ∑i=0k−1vi*cvi=1.\sum_{i=0}^{k-1} v_i^* c v_i = 1.(2)✦

For elements aa and bb of a unital C*C^*-algebra AA, aa is Cuntz subequivalent to bb, written a≾ba \precsim b (elementwise in AA without matrix amplifications), if aa lies in the norm closure of {x*bx:x∈A}\{x^* b x : x \in A\}.✦ Let PP, RR, and BB be unital C*C^*-algebras, let φ:P→B\varphi : P \to B and ψ:R→B\psi : R \to B be unital **-homomorphisms with commuting ranges, and let JJ and II be finite index sets. Suppose X∈PX \in P, (yj)j∈J(y_j)_{j \in J} is a family of pairwise orthogonal positive elements in PP with yj≥0y_j \ge 0, (uj,i)j∈J,i∈I(u_{j,i})_{j \in J, i \in I} is a family in PP, e∈Re \in R, and (zi)i∈I(z_i)_{i \in I} and (vj)j∈J(v_j)_{j \in J} are families in RR. If ∑i∈Iuj,i*yjuj,i=X\sum_{i \in I} u_{j,i}^* y_j u_{j,i} = X for every j∈Jj \in J, zi*zl=δilez_i^* z_l = \delta_{il} e for all i,l∈Ii, l \in I, and ∑j∈Jvj*evj=1\sum_{j \in J} v_j^* e v_j = 1 in RR, then φ(X)≾φ(∑j∈Jyj).\varphi(X) \precsim \varphi\Bigl(\sum_{j \in J} y_j\Bigr).(3)✦

The Claim

Theorem 1. Let N≥2N \ge 2 be a positive dimension equipped with a Pauli matrix system, let PP be a unital C*C^*-algebra, and let η:𝒞N→P\eta : \mathcal{C}_N \to P be a unital **-homomorphism such that η(q0)\eta(q_0) is full in PP. Then for some natural number mm, there exists a unital **-homomorphism I(N,N+1)→P⊗max(m+1).I(N, N + 1) \longrightarrow P^{\otimes_{\max}(m+1)}.(4)✦

In the paper, the result reads: if PP is a nonzero unital C*C^*-algebra and η:𝒞2→P\eta : \mathcal{C}_2 \to P is a unital **-homomorphism sending the corner element te0,0t e_{0,0} (written ιe11\iota e_{11} in the paper, with 𝒞2\mathcal{C}_2 written CC) to a full element, then for some positive integer mm the maximal tensor power P⊗maxmP^{\otimes_{\max} m} contains Rørdam-Winter elements b1,b2,vBb_1, b_2, v_B satisfying 1−b1−b2≾(b1−1/2)+1 - b_1 - b_2 \precsim (b_1 - 1/2)_+ and hence admits a unital **-homomorphism I(2,3)→P⊗maxmI(2, 3) \to P^{\otimes_{\max} m}. The theorem concluding in (4) omits both the nonzero condition on PP and the intermediate Rørdam-Winter elements, while generalizing rank 22 to any N≥2N \ge 2 with a Pauli matrix system and writing q0q_0 (via (1)) and P⊗max(m+1)P^{\otimes_{\max}(m+1)} in place of te0,0t e_{0,0} and P⊗maxmP^{\otimes_{\max} m}.

What Is Checked

The main theorem (4) is stated over a positive dimension N≥2N \ge 2 with a Pauli matrix system, the matrix cone 𝒞N\mathcal{C}_N and its cut-off element q0q_0, the dimension-drop algebra I(N,N+1)I(N, N + 1), a unital C*C^*-algebra PP with right-nested maximal tensor powers P⊗max(m+1)P^{\otimes_{\max}(m+1)}, and a unital **-homomorphism η:𝒞N→P\eta : \mathcal{C}_N \to P sending q0q_0 to a full element. Beyond the main claim, the library proves that the cut-off element q0q_0 equals the corner element te0,0t e_{0,0}, that every positive full element in a unital C*C^*-algebra admits elements v0,…,vk−1v_0, \dots, v_{k-1} satisfying ∑i=0k−1vi*cvi=1\sum_{i=0}^{k-1} v_i^* c v_i = 1, and that amplified Cuntz comparison holds.