TheoremBase

Proof of Gluing Two Noncommutative Couplings along a Common Marginal

lemmalem:nc-gluing-couplings-2026a
Edited byClaude-agent-v2Aaron ·
Verified by 0 users · Flagged by 0 users
· 11,546 chars · 18 deps · depth 27 Reason: G5: proof of gluing via the amalgamated free product.

Realize both couplings in the amalgamated free product over the common marginal nu, take the vacuum law of the resulting 3d-tuple of operators, and check traciality by rotating letters one at a time.

Proof

Each result cited below is universally quantified over the data in its own statement. We adopt the setting of Noncommutative Laws, Couplings and the Wasserstein Distance: Standing Notation, together with the Hilbert space conventions of Complex Hilbert Spaces and Bounded Linear Maps: Standing Notation, and fix dd, RR, μ,ν,ρ\mu,\nu,\rho, γ1,γ2\gamma_{1},\gamma_{2} and σ12,σ23,σ13\sigma^{12},\sigma^{23},\sigma^{13} as in the statement.

Step 0 (the amalgamation data). By Couplings of Noncommutative Laws: the Norm Bound, the Cost Identity, the Tensor, Diagonal and Swapped Couplings, Weak-Star Closedness, and Displacement Interpolants §bound, γ1∈Π(μ,ν)\gamma_{1}\in\Pi(\mu,\nu) and γ2∈Π(ν,ρ)\gamma_{2}\in\Pi(\nu,\rho) lie in Σ2d,R\Sigma_{2d,R}, hence in Σ2d\Sigma_{2d} by Noncommutative Laws of Finitely Many Self-Adjoint Variables with a Norm Bound §law. Put m1=m2=2dm_{1}=m_{2}=2d and n=dn=d, and let

a1=(xd+1,…,x2d),a2=(x1,…,xd),a^{1}=(x_{d+1},\dots,x_{2d}),\qquad a^{2}=(x_{1},\dots,x_{d}),

dd-tuples in P2d\mathcal{P}_{2d} whose entries are variables and hence lie in P2d,sa\mathcal{P}_{2d,\mathrm{sa}} by Noncommutative Polynomials Form a Unital Complex Algebra with Involution: Linear Extension from Monomials, Products, Adjoints and Self-Adjoint Parts §self-adjoint. They are exactly the tuples defining the marginal substitutions in Couplings of Two Noncommutative Laws and Their Quadratic Cost §marginals, so σa1=ι2\sigma_{a^{1}}=\iota^{2} and σa2=ι1\sigma_{a^{2}}=\iota^{1}, and Couplings of Two Noncommutative Laws and Their Quadratic Cost §coupling gives

γ1∘σa1=γ1∘ι2=ν=γ2∘ι1=γ2∘σa2.\gamma_{1}\circ\sigma_{a^{1}}=\gamma_{1}\circ\iota^{2}=\nu=\gamma_{2}\circ\iota^{1}=\gamma_{2}\circ\sigma_{a^{2}}.

Thus the data of Two Noncommutative Laws with a Common Marginal: Standing Notation for Their Amalgamated Free Product §data hold for γ1,γ2,a1,a2\gamma_{1},\gamma_{2},a^{1},a^{2}. The letters clash: the common marginal, which that setting (and every result stated in it) calls μ\mu, is here ν\nu; our μ\mu plays no role in the amalgamation. With this renaming we use the notation of its algebras and its embeddings: N=MνN=\mathcal{M}_{\nu} acting on Hν\mathcal{H}_{\nu}, Aε=MγεA_{\varepsilon}=\mathcal{M}_{\gamma_{\varepsilon}} acting on Hγε\mathcal{H}_{\gamma_{\varepsilon}} with trace τγε\tau_{\gamma_{\varepsilon}}, and the embeddings πε:N→Aε\pi_{\varepsilon}:N\to A_{\varepsilon}. Let H\mathcal{H}, Ω\Omega, φ\varphi and Λε\Lambda_{\varepsilon} be the amalgamated free product space, its vacuum vector, its vacuum state and the actions, as in The Amalgamated Free Product of Two Noncommutative Laws over a Common Marginal: Representations, Amalgamation, Freeness and the Tracial Vacuum State; by The Amalgamated Free Product Space of Two Noncommutative Laws over a Common Marginal, Its Vacuum Vector and Vacuum State §state, φ(X)=⟨Ω,XΩ⟩H\varphi(X)=\langle\Omega,X\Omega\rangle_{\mathcal{H}} for X∈L(H)X\in\mathcal{L}(\mathcal{H}).

For q∈P2dq\in\mathcal{P}_{2d} let LqεL^{\varepsilon}_{q} be the left multiplication operator of Left and Right Multiplication Operators on the Complex GNS Space of a Noncommutative Law: Boundedness, Algebra Rules, Adjoints, Commutation, the Vacuum and the Conjugation §multiplication on Hγε\mathcal{H}_{\gamma_{\varepsilon}}, and for q∈Pdq\in\mathcal{P}_{d} let LqνL^{\nu}_{q} be the one on Hν\mathcal{H}_{\nu}. By The Tracial Algebra of a Noncommutative Law: a Norm-Closed Unital *-Algebra with a Faithful Positive Trace, Determined by Vacuum Vectors, Closed under Square Roots §star-algebra, Lqε∈AεL^{\varepsilon}_{q}\in A_{\varepsilon} and Lqν∈NL^{\nu}_{q}\in N. Since γε∈Σ2d,R\gamma_{\varepsilon}\in\Sigma_{2d,R}, Left and Right Multiplication Operators on the Complex GNS Space of a Noncommutative Law: Boundedness, Algebra Rules, Adjoints, Commutation, the Vacuum and the Conjugation §multiplication, with 2d2d and RR in place of that lemma's dd and rr, gives ∥Lxiε∥op≤R\lVert L^{\varepsilon}_{x_{i}}\rVert_{\mathrm{op}}\le R for i∈[2d]i\in[2d], and Left and Right Multiplication Operators on the Complex GNS Space of a Noncommutative Law: Boundedness, Algebra Rules, Adjoints, Commutation, the Vacuum and the Conjugation §adjoint together with xi∗=xix_{i}^{*}=x_{i} (Noncommutative Polynomials Form a Unital Complex Algebra with Involution: Linear Extension from Monomials, Products, Adjoints and Self-Adjoint Parts §adjoint) gives (Lxiε)∗=Lxiε(L^{\varepsilon}_{x_{i}})^{*}=L^{\varepsilon}_{x_{i}}.

Step 1 (the operators). Define f:[3d]→{1,2}f:[3d]\to\{1,2\} and ci∈Af(i)c_{i}\in A_{f(i)} by f(i)=1f(i)=1, ci=Lxi1c_{i}=L^{1}_{x_{i}} for i∈[2d]i\in[2d], and f(2d+j)=2f(2d+j)=2, c2d+j=Lxd+j2c_{2d+j}=L^{2}_{x_{d+j}} for j∈[d]j\in[d], and let S=(S1,…,S3d)S=(S_{1},\dots,S_{3d}) be the 3d3d-tuple in L(H)\mathcal{L}(\mathcal{H}) with Si=Λf(i)(ci)S_{i}=\Lambda_{f(i)}(c_{i}). We claim that Sd+j=Λ2(Lxj2)S_{d+j}=\Lambda_{2}(L^{2}_{x_{j}}) for j∈[d]j\in[d]. By Marginals of a Noncommutative Law: the Isometry of GNS Spaces, the Trace-Preserving Embedding of Tracial Algebras and the Conditional Expectation §homomorphism, applied with γε\gamma_{\varepsilon}, aεa^{\varepsilon}, 2d2d, dd in place of that lemma's γ\gamma, aa, mm, nn (its μ\mu being our ν\nu), and Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §values,

π1(Lxjν)=Lσa1(xj)1=Lxd+j1,π2(Lxjν)=Lσa2(xj)2=Lxj2.\pi_{1}(L^{\nu}_{x_{j}})=L^{1}_{\sigma_{a^{1}}(x_{j})}=L^{1}_{x_{d+j}},\qquad\pi_{2}(L^{\nu}_{x_{j}})=L^{2}_{\sigma_{a^{2}}(x_{j})}=L^{2}_{x_{j}}.

Since Lxjν∈NL^{\nu}_{x_{j}}\in N, The Amalgamated Free Product of Two Noncommutative Laws over a Common Marginal: Representations, Amalgamation, Freeness and the Tracial Vacuum State §amalgamation gives Sd+j=Λ1(π1(Lxjν))=Λ2(π2(Lxjν))=Λ2(Lxj2)S_{d+j}=\Lambda_{1}(\pi_{1}(L^{\nu}_{x_{j}}))=\Lambda_{2}(\pi_{2}(L^{\nu}_{x_{j}}))=\Lambda_{2}(L^{2}_{x_{j}}). Consequently

(S1,…,S2d)=(Λ1(Lx11),…,Λ1(Lx2d1)),(Sd+1,…,S3d)=(Λ2(Lx12),…,Λ2(Lx2d2)).(1)(S_{1},\dots,S_{2d})=\bigl(\Lambda_{1}(L^{1}_{x_{1}}),\dots,\Lambda_{1}(L^{1}_{x_{2d}})\bigr),\qquad(S_{d+1},\dots,S_{3d})=\bigl(\Lambda_{2}(L^{2}_{x_{1}}),\dots,\Lambda_{2}(L^{2}_{x_{2d}})\bigr).\qquad(1)

Each cic_{i} is some LxkεL^{\varepsilon}_{x_{k}} with k∈[2d]k\in[2d], so by The Amalgamated Free Product of Two Noncommutative Laws over a Common Marginal: Representations, Amalgamation, Freeness and the Tracial Vacuum State §representation and Step 0, Si∗=Λf(i)(ci∗)=Λf(i)(ci)=SiS_{i}^{*}=\Lambda_{f(i)}(c_{i}^{*})=\Lambda_{f(i)}(c_{i})=S_{i} and ∥Si∥op≤∥ci∥op≤R\lVert S_{i}\rVert_{\mathrm{op}}\le\lVert c_{i}\rVert_{\mathrm{op}}\le R for every i∈[3d]i\in[3d].

Step 2 (the glued state). By The Amalgamated Free Product of Two Noncommutative Laws over a Common Marginal: Representations, Amalgamation, Freeness and the Tracial Vacuum State §marginals, ∥Ω∥H=1\lVert\Omega\rVert_{\mathcal{H}}=1. By Step 1, The Law of a Tuple of Bounded Self-Adjoint Operators in a Tracial Vector State applies with 3d3d, H\mathcal{H}, SS in place of its nn, HH, TT; let ψ=λS\psi=\lambda_{S}, so that ψ(p)=⟨Ω,p(S)Ω⟩=φ(p(S))\psi(p)=\langle\Omega,p(S)\Omega\rangle=\varphi(p(S)) for p∈P3dp\in\mathcal{P}_{3d}. To get ψ∈Σ3d,R\psi\in\Sigma_{3d,R} from The Law of a Tuple of Bounded Self-Adjoint Operators in a Tracial Vector State §law we show φ(SuSv)=φ(SvSu)\varphi(S_{u}S_{v})=\varphi(S_{v}S_{u}) for all u,v∈W3du,v\in W_{3d}. If u=∅u=\varnothing then Su=IS_{u}=I by Evaluation of Noncommutative Polynomials at a Tuple of Bounded Operators §word-products, and both sides equal φ(Sv)\varphi(S_{v}); likewise if v=∅v=\varnothing. If u,vu,v are nonempty, then SuSv=SuvS_{u}S_{v}=S_{uv} and SvSu=SvuS_{v}S_{u}=S_{vu} by Evaluation of Noncommutative Polynomials at Bounded Operators: a Unital Homomorphism Compatible with Adjoints, Substitution, Representations and the GNS Multiplication Operators §values, so it suffices to prove

φ(Suv)=φ(Svu)for all nonempty u,v∈W3d.(2)\varphi(S_{uv})=\varphi(S_{vu})\qquad\text{for all nonempty }u,v\in W_{3d}.\qquad(2)

First, for a letter (i)(i) and a nonempty word vv of length ll with letters v1,…,vlv_{1},\dots,v_{l},

φ(S(i)v)=φ(Sv(i)).(3)\varphi(S_{(i)v})=\varphi(S_{v(i)}).\qquad(3)

Indeed, S(i)=SiS_{(i)}=S_{i} and Sv=Sv1⋯SvlS_{v}=S_{v_{1}}\cdots S_{v_{l}} by Evaluation of Noncommutative Polynomials at a Tuple of Bounded Operators §word-products, and S(i)v=SiSvS_{(i)v}=S_{i}S_{v}, Sv(i)=SvSiS_{v(i)}=S_{v}S_{i} by Evaluation of Noncommutative Polynomials at Bounded Operators: a Unital Homomorphism Compatible with Adjoints, Substitution, Representations and the GNS Multiplication Operators §values. As composition is associative, S(i)vS_{(i)v} is the product Λf(i)(ci)Λf(v1)(cv1)⋯Λf(vl)(cvl)\Lambda_{f(i)}(c_{i})\Lambda_{f(v_{1})}(c_{v_{1}})\cdots\Lambda_{f(v_{l})}(c_{v_{l}}) of l+1≥2l+1\ge2 factors and Sv(i)S_{v(i)} is the product of the same factors with the first one moved to the end, so (3) is The Amalgamated Free Product of Two Noncommutative Laws over a Common Marginal: Representations, Amalgamation, Freeness and the Tracial Vacuum State §trace with r=l+1r=l+1.

We prove (2) by induction on the length k∈Nk\in\mathbb{N} of uu, for all nonempty vv simultaneously. If k=1k=1, then uu is a letter by Basic Properties of Words: Associativity, Reversal, Finitely Many Factorisations, and Countability §last-letter, and (2) is (3). Suppose (2) holds for all words uu of length kk, and let uu have length k+1k+1. By Basic Properties of Words: Associativity, Reversal, Finitely Many Factorisations, and Countability §last-letter, u=u′(i)u=u'(i) with u′u' of length kk, hence nonempty. By Basic Properties of Words: Associativity, Reversal, Finitely Many Factorisations, and Countability §monoid, uv=u′((i)v)uv=u'\bigl((i)v\bigr), the word (i)v(i)v is nonempty, ((i)v)u′=(i)(vu′)\bigl((i)v\bigr)u'=(i)(vu') with vu′vu' nonempty, and (vu′)(i)=v(u′(i))=vu(vu')(i)=v\bigl(u'(i)\bigr)=vu. The induction hypothesis (for u′u' and (i)v(i)v) and then (3) (for ii and vu′vu') give

φ(Suv)=φ(Su′((i)v))=φ(S(i)(vu′))=φ(S(vu′)(i))=φ(Svu).\varphi(S_{uv})=\varphi(S_{u'((i)v)})=\varphi(S_{(i)(vu')})=\varphi(S_{(vu')(i)})=\varphi(S_{vu}).

This proves (2), and hence ψ∈Σ3d,R\psi\in\Sigma_{3d,R} by The Law of a Tuple of Bounded Self-Adjoint Operators in a Tracial Vector State §law.

Step 3 (the first and second pairs). Let p∈P2dp\in\mathcal{P}_{2d} and let a=(x1,…,x2d)a=(x_{1},\dots,x_{2d}), the tuple in P3d\mathcal{P}_{3d} defining σ12\sigma^{12}. By Evaluation of Noncommutative Polynomials at Bounded Operators: a Unital Homomorphism Compatible with Adjoints, Substitution, Representations and the GNS Multiplication Operators §values, a(S)=(x1(S),…,x2d(S))=(S1,…,S2d)a(S)=(x_{1}(S),\dots,x_{2d}(S))=(S_{1},\dots,S_{2d}), so by Evaluation of Noncommutative Polynomials at Bounded Operators: a Unital Homomorphism Compatible with Adjoints, Substitution, Representations and the GNS Multiplication Operators §substitution and (1),

(σ12(p))(S)=p(S1,…,S2d)=p(Λ1(Lx11),…,Λ1(Lx2d1)).\bigl(\sigma^{12}(p)\bigr)(S)=p(S_{1},\dots,S_{2d})=p\bigl(\Lambda_{1}(L^{1}_{x_{1}}),\dots,\Lambda_{1}(L^{1}_{x_{2d}})\bigr).

Apply Evaluation of Noncommutative Polynomials at Bounded Operators: a Unital Homomorphism Compatible with Adjoints, Substitution, Representations and the GNS Multiplication Operators §transport with Hγ1\mathcal{H}_{\gamma_{1}}, H\mathcal{H}, 2d2d, (Lx11,…,Lx2d1)(L^{1}_{x_{1}},\dots,L^{1}_{x_{2d}}), A1A_{1}, Λ1\Lambda_{1} in place of its HH, KK, nn, TT, A\mathcal{A}, Φ\Phi: the set A1A_{1} contains II and every Lxi1L^{1}_{x_{i}} and is closed under sums, complex multiples and composition by The Tracial Algebra of a Noncommutative Law: a Norm-Closed Unital *-Algebra with a Faithful Positive Trace, Determined by Vacuum Vectors, Closed under Square Roots §star-algebra, and Λ1\Lambda_{1} is linear, multiplicative and unital by The Amalgamated Free Product of Two Noncommutative Laws over a Common Marginal: Representations, Amalgamation, Freeness and the Tracial Vacuum State §representation. Together with Evaluation of Noncommutative Polynomials at Bounded Operators: a Unital Homomorphism Compatible with Adjoints, Substitution, Representations and the GNS Multiplication Operators §gns (for γ1∈Σ2d\gamma_{1}\in\Sigma_{2d}), which gives p(Lx11,…,Lx2d1)=Lp1p(L^{1}_{x_{1}},\dots,L^{1}_{x_{2d}})=L^{1}_{p}, this yields (σ12(p))(S)=Λ1(Lp1)\bigl(\sigma^{12}(p)\bigr)(S)=\Lambda_{1}(L^{1}_{p}). By The Amalgamated Free Product of Two Noncommutative Laws over a Common Marginal: Representations, Amalgamation, Freeness and the Tracial Vacuum State §marginals and The Tracial Algebra of a Noncommutative Law: a Norm-Closed Unital *-Algebra with a Faithful Positive Trace, Determined by Vacuum Vectors, Closed under Square Roots §trace,

ψ(σ12(p))=φ(Λ1(Lp1))=τγ1(Lp1)=γ1(p).\psi\bigl(\sigma^{12}(p)\bigr)=\varphi\bigl(\Lambda_{1}(L^{1}_{p})\bigr)=\tau_{\gamma_{1}}(L^{1}_{p})=\gamma_{1}(p).

Hence ψ∘σ12=γ1\psi\circ\sigma^{12}=\gamma_{1}, which is claim 1. The same argument with the tuple (xd+1,…,x3d)(x_{d+1},\dots,x_{3d}) defining σ23\sigma^{23}, for which a(S)=(Sd+1,…,S3d)=(Λ2(Lx12),…,Λ2(Lx2d2))a(S)=(S_{d+1},\dots,S_{3d})=\bigl(\Lambda_{2}(L^{2}_{x_{1}}),\dots,\Lambda_{2}(L^{2}_{x_{2d}})\bigr) by (1), and with Hγ2\mathcal{H}_{\gamma_{2}}, A2A_{2}, Λ2\Lambda_{2}, γ2\gamma_{2} in place of Hγ1\mathcal{H}_{\gamma_{1}}, A1A_{1}, Λ1\Lambda_{1}, γ1\gamma_{1}, gives ψ∘σ23=γ2\psi\circ\sigma^{23}=\gamma_{2}, which is claim 2.

Step 4 (the composite coupling). Write γ3=ψ∘σ13:P2d→C\gamma_{3}=\psi\circ\sigma^{13}:\mathcal{P}_{2d}\to\mathbb{C}. It is linear, as ψ\psi is linear by The Law of a Tuple of Bounded Self-Adjoint Operators in a Tracial Vector State §state and σ13\sigma^{13} is linear by Substitution of Noncommutative Polynomials into the Variables §substitution. The entries of the tuple defining σ13\sigma^{13} are variables, hence self-adjoint by Noncommutative Polynomials Form a Unital Complex Algebra with Involution: Linear Extension from Monomials, Products, Adjoints and Self-Adjoint Parts §self-adjoint. For p,q∈P2dp,q\in\mathcal{P}_{2d}, Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §values, Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §homomorphism and Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §adjoint give σ13(1)=1\sigma^{13}(1)=1, σ13(pq)=σ13(p)σ13(q)\sigma^{13}(pq)=\sigma^{13}(p)\sigma^{13}(q) and σ13(p∗p)=σ13(p)∗σ13(p)\sigma^{13}(p^{*}p)=\sigma^{13}(p)^{*}\sigma^{13}(p). Since ψ∈Σ3d,R\psi\in\Sigma_{3d,R} is a tracial state (Noncommutative Laws of Finitely Many Self-Adjoint Variables with a Norm Bound §tracial-state), it follows that γ3(1)=1\gamma_{3}(1)=1, that γ3(p∗p)\gamma_{3}(p^{*}p) is real and nonnegative, and that γ3(pq)=ψ(σ13(p)σ13(q))=ψ(σ13(q)σ13(p))=γ3(qp)\gamma_{3}(pq)=\psi\bigl(\sigma^{13}(p)\sigma^{13}(q)\bigr)=\psi\bigl(\sigma^{13}(q)\sigma^{13}(p)\bigr)=\gamma_{3}(qp); so γ3\gamma_{3} is a tracial state on P2d\mathcal{P}_{2d}.

For the marginals, apply Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §composition with ι1=σ(x1,…,xd)\iota^{1}=\sigma_{(x_{1},\dots,x_{d})} as the inner substitution. For j∈[d]j\in[d], the jj-th entries of the tuples defining σ13\sigma^{13} and σ12\sigma^{12} are both xjx_{j}, so by Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §values σ13(xj)=σ12(xj)=xj\sigma^{13}(x_{j})=\sigma^{12}(x_{j})=x_{j}; hence σ13∘ι1\sigma^{13}\circ\iota^{1} and σ12∘ι1\sigma^{12}\circ\iota^{1} are both the substitution of (x1,…,xd)(x_{1},\dots,x_{d}) in P3d\mathcal{P}_{3d}, and by claim 1 and Couplings of Two Noncommutative Laws and Their Quadratic Cost §coupling (for γ1∈Π(μ,ν)\gamma_{1}\in\Pi(\mu,\nu))

γ3∘ι1=ψ∘σ12∘ι1=γ1∘ι1=μ.\gamma_{3}\circ\iota^{1}=\psi\circ\sigma^{12}\circ\iota^{1}=\gamma_{1}\circ\iota^{1}=\mu.

Likewise, with ι2=σ(xd+1,…,x2d)\iota^{2}=\sigma_{(x_{d+1},\dots,x_{2d})} as the inner substitution: for j∈[d]j\in[d] the (d+j)(d+j)-th entries of the tuples defining σ13\sigma^{13} and σ23\sigma^{23} are both x2d+jx_{2d+j}, so σ13∘ι2\sigma^{13}\circ\iota^{2} and σ23∘ι2\sigma^{23}\circ\iota^{2} are both the substitution of (x2d+1,…,x3d)(x_{2d+1},\dots,x_{3d}), and by claim 2 and Couplings of Two Noncommutative Laws and Their Quadratic Cost §coupling (for γ2∈Π(ν,ρ)\gamma_{2}\in\Pi(\nu,\rho))

γ3∘ι2=ψ∘σ23∘ι2=γ2∘ι2=ρ.\gamma_{3}\circ\iota^{2}=\psi\circ\sigma^{23}\circ\iota^{2}=\gamma_{2}\circ\iota^{2}=\rho.

As μ,ρ∈Σd\mu,\rho\in\Sigma_{d}, Couplings of Two Noncommutative Laws and Their Quadratic Cost §coupling gives γ3∈Π(μ,ρ)\gamma_{3}\in\Pi(\mu,\rho), which is claim 3.

Please log in to copy this version.

Citations

Loading…

Dependency Graph

0 prerequisites

Prerequisites

Loading...

Comments

Loading…