OctoInform Beta

OpenAI preprint, September 2026 · §3. Reduced HNN adjunctions · §3

§3 (setting): words and reduced words: Words vi1a1⋯vimamv_{i_1}a_1\cdots v_{i_m}a_m over AA are reduced, for bridge data on AA, when E3−ik(ak)=0E_{3-i_k}(a_k)=0 wherever ik+1≠iki_{k+1}\neq i_k

Reduced words: centered coefficients at each change of letter.

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

Let AA be a C*C^*-algebra with a faithful tracial state τ\tau and two decreasing sequences of projections p1=(pn1)p^1 = (p^1_n) and p2=(pn2)p^2 = (p^2_n). Their reduced Higman-Neumann-Neumann (HNN) extension H(A)H(A) adjoins to AA a unitary vv with vpn1v*=pn2v p^1_n v^* = p^2_n for every nn. One computes in H(A)H(A) with products in which each letter vv or v*v^* is followed by a coefficient from AA, such as va1v*a2va3v a_1 v^* a_2 v a_3. Which of them are reduced is decided at the coefficients standing between two different letters.

The library defines words over AA, whose evaluations are these products, and reducedness for any condition imposed on such a coefficient; the two sequences supply the condition through two expectations of AA onto the algebra of convergent sequences.

Peaking Sequences and Bridge Data

A state of AA is a bounded linear functional on AA of norm one that is real and nonnegative on every a*aa^*a.✦ A positive functional τ\tau on AA is tracial when τ(ab)=τ(ba)\tau(ab) = \tau(ba) for all a,b∈Aa, b \in A.✦

A sequence (pn)(p_n) in AA peaks to a bounded functional ff on AA when the pnp_n are decreasing projections, ff is a state with f(pn)=1f(p_n) = 1 for every nn, and every state gg with g(pn)=1g(p_n) = 1 for every nn equals ff.✦ The peak atom of index nn of (pn)(p_n) is the element dn=pn−pn+1d_n = p_n - p_{n+1}(1)✦ of AA.

Bridge data on AA consist of a positive functional τ\tau on AA that is tracial, normalized by τ(1)=1\tau(1) = 1 and faithful, with τ(a*a)=0\tau(a^*a) = 0 only for a=0a = 0, and of two sequences p1p^1, p2p^2 in AA peaking to states f1f_1, f2f_2, such that, with dnid^i_n the peak atoms of pip^i, for i=1,2i = 1, 2 and every nn p0i=1,τ(dni)>0,τ(pn1)=τ(pn2),limn→∞τ(pni)=0.p^i_0 = 1, \qquad \tau(d^i_n) > 0, \qquad \tau(p^1_n) = \tau(p^2_n), \qquad \lim_{n \to \infty} \tau(p^i_n) = 0.(2)✦

Atomic trace data on AA, with atoms indexed by a set VV, consist of a positive functional τ\tau on AA that is tracial, normalized and faithful, and of pairwise orthogonal projections eve_v, v∈Vv \in V, with τ(ev)>0\tau(e_v) > 0 and ∑vτ(ev)=τ(1)\sum_v \tau(e_v) = \tau(1).✦ Let τ\tau be a positive functional on AA that is tracial, normalized and faithful, and let (pn)(p_n) be a sequence in AA peaking to a state, with p0=1p_0 = 1, τ(dn)>0\tau(d_n) > 0 and τ(pn)→0\tau(p_n) \to 0. The peak atomic trace data of (pn)(p_n) for τ\tau are the atomic trace data on AA with trace τ\tau and atoms the peak atoms dnd_n.✦

Diagonals and Expectations

The algebra cc of convergent sequences is the C*C^*-algebra of continuous functions on the one-point compactification ℕ∪{∞}\mathbb{N} \cup \{\infty\} of ℕ\mathbb{N}; an element gg of cc has terms gng_n and limit g(∞)g(\infty).✦ For a sequence (pn)(p_n) in AA, a complex number λ\lambda and a complex sequence (bn)(b_n), the peak diagonal element of (pn)(p_n) is λ1+∑n(bn−λ)dn,\lambda 1 + \sum_n (b_n - \lambda)\, d_n,(3)✦ and the sequence diagonal map δ\delta sends g∈cg \in c to (3) with λ=g(∞)\lambda = g(\infty) and bn=gnb_n = g_n.✦ Let (pn)(p_n) peak to a state, with its peak atoms the atoms of atomic trace data on AA. The sequence diagonal embedding of (pn)(p_n) is the map δ:c→A\delta\colon c \to A defined above, a unital **-homomorphism.

The corner functional of a bounded functional ff on AA at p∈Ap \in A is the functional a↦f(pap)a \mapsto f(pap)✦, and the corner state of ff at pp is the corner functional divided by Re⁡f(p)\operatorname{Re} f(p).✦ Let (pn)(p_n) peak to ff, and let τ\tau be a state with τ(dn)>0\tau(d_n) > 0 for every nn. The peaking coefficient function of a∈Aa \in A is the convergent sequence E(a)E(a) of the corner states of τ\tau at the peak atoms, evaluated at aa, with limit f(a)f(a): E(a)n=τ(dnadn)τ(dn),E(a)(∞)=f(a).E(a)_n = \frac{\tau(d_n a d_n)}{\tau(d_n)}, \qquad E(a)(\infty) = f(a).(4)✦ The peaking coefficient map of (pn)(p_n) for τ\tau is E:A→cE\colon A \to c, a linear map of norm at most one.✦ For bridge data, δi\delta_i is the sequence diagonal embedding of pip^i and EiE_i its peaking coefficient map for τ\tau.

The Two Factors

Let DD be a C*C^*-algebra and χ\chi a map sending each x∈Dx \in D to a complex function χ(x)\chi(x) on a set VV. A matrix diagonal input on AA over a finite index set II, with coordinates χ\chi, consists of a positive functional τ\tau on AA that is tracial, normalized and faithful and, for each i∈Ii \in I, of atomic trace data on AA with trace τ\tau and atoms evie^i_v, a unital **-homomorphism ιi:D→A\iota_i\colon D \to A, its diagonal, and a bounded linear map Fi:A→DF_i\colon A \to D, its expectation, such that for all x∈Dx \in D, a∈Aa \in A and v∈Vv \in V eviιi(x)=χ(x)(v)evi,∥Fi(a)∥≤∥a∥,Fi(ιi(x))=x,Fi(ιi(x)a)=xFi(a),e^i_v \iota_i(x) = \chi(x)(v)\, e^i_v, \quad \|F_i(a)\| \le \|a\|, \quad F_i(\iota_i(x)) = x, \quad F_i(\iota_i(x) a) = x F_i(a),(5)✦ and τ(eviaevi)=0\tau(e^i_v a e^i_v) = 0 whenever Fi(a)=0F_i(a) = 0.

Let τ\tau be a positive functional on AA that is tracial, normalized and faithful, and let pip^i, i∈Ii \in I, be finitely many sequences in AA peaking to states, with p0i=1p^i_0 = 1, τ(dni)>0\tau(d^i_n) > 0 and τ(pni)→0\tau(p^i_n) \to 0. The peaking matrix input of the pip^i for τ\tau is the matrix diagonal input on AA over II and D=cD = c, with coordinates χ(x)(v)=xv\chi(x)(v) = x_v, whose iith atomic trace data, diagonal and expectation are the peak atomic trace data, the sequence diagonal embedding and the peaking coefficient map for τ\tau of pip^i.✦

Let φ:D→A\varphi\colon D \to A be an injective unital **-homomorphism, let atomic trace data on AA have trace τ\tau and atoms eve_v, v∈Vv \in V, and let qv∈Dq_v \in D with φ(qv)=ev\varphi(q_v) = e_v. The pullback of these atomic trace data along φ\varphi is the atomic trace data on DD with trace τ∘φ\tau \circ \varphi, a positive functional on DD, and atoms qvq_v.✦ Let (pn)(p_n) peak to ff and let τ\tau be a state with τ(dn)>0\tau(d_n) > 0 for every nn. The sequence atom of index vv is E(dv)∈cE(d_v) \in c. The sequence atomic trace data on cc are the pullback, along the sequence diagonal embedding and with the sequence atoms as the qvq_v, of the atomic trace data on AA whose atoms are the peak atoms.✦ For a nonempty finite index set II, the sequence matrix input over II is the matrix diagonal input on cc over II whose trace and atomic trace data are those of the sequence atomic trace data and whose diagonals and expectations are the identity.✦

The HNN factors over AA are AA, of index 11, and cc, of index 22.✦ The HNN matrix input of bridge data on AA, also called their pair matrix input, gives each factor a matrix diagonal input over the index set {1,2}\{1, 2\} with D=cD = c and coordinates χ(x)(v)=xv\chi(x)(v) = x_v: on AA the peaking matrix input of p1p^1 and p2p^2, on cc the sequence matrix input for the peak atomic trace data of p1p^1.✦

The Definition

Definition 1. A word ww over AA is a finite nonempty sequence of pairs (ik,ak)(i_k, a_k) with ik∈{1,2}i_k \in \{1, 2\} and ak∈Aa_k \in A, read as the letter vikv_{i_k}, one of two letters v1v_1, v2v_2, followed by the coefficient aka_k: w=(i1,a1)(i2,a2)⋯(im,am),m≥1.w = (i_1, a_1)(i_2, a_2) \cdots (i_m, a_m), \qquad m \ge 1.(6)✦ For a map φ\varphi of AA into an algebra CC and a pair v=(v1,v2)v = (v_1, v_2) of elements of CC, the evaluation of ww is ev⁡φ,v(w)=vi1φ(a1)vi2φ(a2)⋯vimφ(am).\operatorname{ev}_{\varphi, v}(w) = v_{i_1} \varphi(a_1)\, v_{i_2} \varphi(a_2) \cdots v_{i_m} \varphi(a_m).(7)✦ For a condition KK on pairs (i,a)(i, a) with i∈{1,2}i \in \{1, 2\} and a∈Aa \in A, the word ww is reduced for KK when K(ik,ak) holds for every k with 1≤k<m and ik+1≠ik.K(i_k, a_k) \text{ holds for every } k \text{ with } 1 \le k < m \text{ and } i_{k+1} \ne i_k.(8)✦ For bridge data on AA, a turn after the letter viv_i, where viv_i is followed by a coefficient a∈Aa \in A and then by the other letter, is centered when E3−i(a)=0,E_{3-i}(a) = 0,(9)✦ A word ww over AA is reduced for the bridge data when ww is reduced for the condition (9) on pairs (i,a)(i, a).

In the paper, the result reads: a word is a0vε1a1⋯vεmama_0 v^{\varepsilon_1} a_1 \cdots v^{\varepsilon_m} a_m with each εt=±1\varepsilon_t = \pm 1, and it is reduced when E1(at)=0E_1(a_t) = 0 at each turn (εt,εt+1)=(+1,−1)(\varepsilon_t, \varepsilon_{t+1}) = (+1, -1) and E2(at)=0E_2(a_t) = 0 at each turn (−1,+1)(-1, +1), where v+1=vv^{+1} = v and v−1=v*v^{-1} = v^* are the page’s letters v2v_2 and v1v_1.

The library’s words begin with a letter: they are the paper’s words with a0=1a_0 = 1.

What Is Checked

Words and their evaluation are stated over any set AA of coefficients, any map φ\varphi of AA into an algebra and any two of its elements v1v_1, v2v_2, and reducedness over any condition KK on pairs (i,a)(i, a); the centered turns of (9) are stated over bridge data on a C*C^*-algebra AA and the expectations E1E_1, E2E_2 of their pair matrix input on the factor AA.

The library proves that the peak atoms of a peaking sequence are pairwise orthogonal projections. A positive functional ff on AA is bounded, with |f(a)|≤Re⁡f(1)∥a∥|f(a)| \le \operatorname{Re} f(1)\, \|a\|✦, and is a state when f(1)=1f(1) = 1.✦ The sequence diagonal embedding is injective✦, and the atoms eve_v of the sequence atomic trace data satisfy evx=xveve_v x = x_v e_v for every x∈cx \in c.✦