Each result cited is universally quantified over the data in its own statement.
Conventions and elementary facts. For m , n ∈ N m,n\in\mathbb{N} m , n ∈ N write ι m , n 1 = σ ( x 1 , … , x m ) : P m → P m + n \iota^{1}_{m,n}=\sigma_{(x_{1},\dots,x_{m})}:\mathcal{P}_{m}\to\mathcal{P}_{m+n} ι m , n 1 = σ ( x 1 , … , x m ) : P m → P m + n and ι m , n 2 = σ ( x m + 1 , … , x m + n ) : P n → P m + n \iota^{2}_{m,n}=\sigma_{(x_{m+1},\dots,x_{m+n})}:\mathcal{P}_{n}\to\mathcal{P}_{m+n} ι m , n 2 = σ ( x m + 1 , … , x m + n ) : P n → P m + n ; these are the maps ι 1 , ι 2 \iota^{1},\iota^{2} ι 1 , ι 2 of Freeness of Two Groups of Variables under a Noncommutative Law §free for the split of m + n m+n m + n variables into the first m m m and the last n n n , and a free product α ⋆ β \alpha\star\beta α ⋆ β for that split is taken as in The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §product . For m = n = d m=n=d m = n = d they are the marginal substitutions ι 1 = ι d , d 1 \iota^{1}=\iota^{1}_{d,d} ι 1 = ι d , d 1 and ι 2 = ι d , d 2 \iota^{2}=\iota^{2}_{d,d} ι 2 = ι d , d 2 . Every tuple of variables is self-adjoint by Noncommutative Polynomials Form a Unital Complex Algebra with Involution: Linear Extension from Monomials, Products, Adjoints and Self-Adjoint Parts §self-adjoint , and the tuple of an affine datum is self-adjoint by Affine Substitutions of Noncommutative Laws: Self-Adjointness, Composition, Moment Formulas and Positivity, and the Coordinate Data §self-adjoint ; the same clause shows that ω ∘ σ T \omega\circ\sigma_{T} ω ∘ σ T is a tracial state (a law if ω \omega ω is a law) for every tracial state ω \omega ω and affine datum T T T .
(C) Composition rule. Let a a a be an n n n -tuple in P m \mathcal{P}_{m} P m and b b b an m m m -tuple in P l \mathcal{P}_{l} P l . By Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §composition , σ b ∘ σ a = σ c \sigma_{b}\circ\sigma_{a}=\sigma_{c} σ b ∘ σ a = σ c with c j = σ b ( a j ) c_{j}=\sigma_{b}(a_{j}) c j = σ b ( a j ) , so ( ω ∘ σ b ) ∘ σ a = ω ∘ σ c (\omega\circ\sigma_{b})\circ\sigma_{a}=\omega\circ\sigma_{c} ( ω ∘ σ b ) ∘ σ a = ω ∘ σ c for every linear ω : P l → C \omega:\mathcal{P}_{l}\to\mathbb{C} ω : P l → C . If a j = ∑ k A j k x k a_{j}=\sum_{k}A_{jk}x_{k} a j = ∑ k A jk x k with real A j k A_{jk} A jk , then c j = ∑ k A j k b k c_{j}=\sum_{k}A_{jk}b_{k} c j = ∑ k A jk b k , because σ b \sigma_{b} σ b is linear (Substitution of Noncommutative Polynomials into the Variables §substitution ) with σ b ( x k ) = b k \sigma_{b}(x_{k})=b_{k} σ b ( x k ) = b k (Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §values ). Substitutions are moreover multiplicative (Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §homomorphism ) and σ b ( 1 ) = 1 \sigma_{b}(1)=1 σ b ( 1 ) = 1 (Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §values ).
(Q) If a , b ≥ 0 a,b\ge0 a , b ≥ 0 are real with a 2 ≤ b 2 a^{2}\le b^{2} a 2 ≤ b 2 , then a ≤ b a\le b a ≤ b : otherwise 0 ≤ b < a 0\le b<a 0 ≤ b < a gives b 2 ≤ a b < a 2 b^{2}\le ab<a^{2} b 2 ≤ ab < a 2 . (S) For real u , v ≥ 0 u,v\ge0 u , v ≥ 0 , u v = u v \sqrt{u}\sqrt{v}=\sqrt{uv} u v = uv and 0 = 0 \sqrt{0}=0 0 = 0 , since u v ≥ 0 \sqrt{u}\sqrt{v}\ge0 u v ≥ 0 and ( u v ) 2 = u v (\sqrt{u}\sqrt{v})^{2}=uv ( u v ) 2 = uv , and 0 2 = 0 0^{2}=0 0 2 = 0 , by the uniqueness in Existence and Uniqueness of the Nonnegative Square Root . (B) If 0 < R ≤ R ′ 0<R\le R' 0 < R ≤ R ′ then Σ n , R ⊆ Σ n , R ′ \Sigma_{n,R}\subseteq\Sigma_{n,R'} Σ n , R ⊆ Σ n , R ′ , because R k ≤ R ′ k R^{k}\le R'^{k} R k ≤ R ′ k for every k ∈ N k\in\mathbb{N} k ∈ N (Noncommutative Laws of Finitely Many Self-Adjoint Variables with a Norm Bound §norm-bound ).
(SC) The semicircular law. s c d \mathrm{sc}_{d} sc d satisfies the Schwinger-Dyson equation (The Semicircular Law and Its Scalings §standard ), so by The Vacuum Law of the Semicircular Operators is the Unique Noncommutative Law Satisfying the Schwinger-Dyson Equation §unique it equals the vacuum law of that theorem, which lies in Σ d , 2 \Sigma_{d,2} Σ d , 2 by The Vacuum Law of the Semicircular Operators is the Unique Noncommutative Law Satisfying the Schwinger-Dyson Equation §vacuum-law ; thus s c d ∈ Σ d , 2 \mathrm{sc}_{d}\in\Sigma_{d,2} sc d ∈ Σ d , 2 . By the Schwinger-Dyson equation and the definition of the free difference quotient (with x ∅ = 1 x_{\varnothing}=1 x ∅ = 1 and s c d ( 1 ) = 1 \mathrm{sc}_{d}(1)=1 sc d ( 1 ) = 1 ), for i , k ∈ [ d ] i,k\in[d] i , k ∈ [ d ] : s c d ( x i ) = s c d ( x i 1 ) = ∂ i s c d ( 1 ) = 0 \mathrm{sc}_{d}(x_{i})=\mathrm{sc}_{d}(x_{i}1)=\partial^{\mathrm{sc}_{d}}_{i}(1)=0 sc d ( x i ) = sc d ( x i 1 ) = ∂ i sc d ( 1 ) = 0 and s c d ( x i x k ) = ∂ i s c d ( x k ) = δ i k \mathrm{sc}_{d}(x_{i}x_{k})=\partial^{\mathrm{sc}_{d}}_{i}(x_{k})=\delta_{ik} sc d ( x i x k ) = ∂ i sc d ( x k ) = δ ik . The tuple of T t T_{t} T t of The Semicircular Law and Its Scalings §scaled is ( t x 1 , … , t x d ) (\sqrt{t}\,x_{1},\dots,\sqrt{t}\,x_{d}) ( t x 1 , … , t x d ) , and s c d , t = s c d ∘ σ T t \mathrm{sc}_{d,t}=\mathrm{sc}_{d}\circ\sigma_{T_{t}} sc d , t = sc d ∘ σ T t .
For λ ∈ Σ d \lambda\in\Sigma_{d} λ ∈ Σ d put γ = λ ⋆ s c d ∈ Σ 2 d \gamma=\lambda\star\mathrm{sc}_{d}\in\Sigma_{2d} γ = λ ⋆ sc d ∈ Σ 2 d (split d + d d+d d + d ). By The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §product , γ \gamma γ is free, γ ∘ ι 1 = λ \gamma\circ\iota^{1}=\lambda γ ∘ ι 1 = λ and γ ∘ ι 2 = s c d \gamma\circ\iota^{2}=\mathrm{sc}_{d} γ ∘ ι 2 = sc d . Write g t = ( x 1 + t x d + 1 , … , x d + t x 2 d ) g^{t}=(x_{1}+\sqrt{t}\,x_{d+1},\dots,x_{d}+\sqrt{t}\,x_{2d}) g t = ( x 1 + t x d + 1 , … , x d + t x 2 d ) for the tuple of Γ t \Gamma_{t} Γ t , so σ Γ t = σ g t \sigma_{\Gamma_{t}}=\sigma_{g^{t}} σ Γ t = σ g t by Affine Data and Affine Substitutions of Noncommutative Polynomials §substitution .
Clause 1 (Realisation). By Free Additive Convolution of Two Noncommutative Laws §convolution , λ ⊞ s c d , t = ( λ ⋆ s c d , t ) ∘ σ a d d \lambda\boxplus\mathrm{sc}_{d,t}=(\lambda\star\mathrm{sc}_{d,t})\circ\sigma_{\mathrm{add}} λ ⊞ sc d , t = ( λ ⋆ sc d , t ) ∘ σ add , where a d d \mathrm{add} add has tuple ( x 1 + x d + 1 , … , x d + x 2 d ) (x_{1}+x_{d+1},\dots,x_{d}+x_{2d}) ( x 1 + x d + 1 , … , x d + x 2 d ) . Apply The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §substitution with m = n = m ′ = n ′ = d m=n=m'=n'=d m = n = m ′ = n ′ = d , a = ( x 1 , … , x d ) a=(x_{1},\dots,x_{d}) a = ( x 1 , … , x d ) in P d \mathcal{P}_{d} P d and b = ( t x 1 , … , t x d ) b=(\sqrt{t}\,x_{1},\dots,\sqrt{t}\,x_{d}) b = ( t x 1 , … , t x d ) , the tuple of T t T_{t} T t . By (C), ι 1 ( x j ) = x j \iota^{1}(x_{j})=x_{j} ι 1 ( x j ) = x j and ι 2 ( t x j ) = t x d + j \iota^{2}(\sqrt{t}\,x_{j})=\sqrt{t}\,x_{d+j} ι 2 ( t x j ) = t x d + j , so the tuple of that clause is c = ( x 1 , … , x d , t x d + 1 , … , t x 2 d ) c=(x_{1},\dots,x_{d},\sqrt{t}\,x_{d+1},\dots,\sqrt{t}\,x_{2d}) c = ( x 1 , … , x d , t x d + 1 , … , t x 2 d ) , and
γ ∘ σ c = ( λ ∘ σ a ) ⋆ ( s c d ∘ σ T t ) = λ ⋆ s c d , t , \gamma\circ\sigma_{c}=(\lambda\circ\sigma_{a})\star(\mathrm{sc}_{d}\circ\sigma_{T_{t}})=\lambda\star\mathrm{sc}_{d,t}, γ ∘ σ c = ( λ ∘ σ a ) ⋆ ( sc d ∘ σ T t ) = λ ⋆ sc d , t ,
using σ a ( p ) = p \sigma_{a}(p)=p σ a ( p ) = p from Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §identity . By (C), ( γ ∘ σ c ) ∘ σ a d d = γ ∘ σ c ′ (\gamma\circ\sigma_{c})\circ\sigma_{\mathrm{add}}=\gamma\circ\sigma_{c'} ( γ ∘ σ c ) ∘ σ add = γ ∘ σ c ′ with c i ′ = σ c ( x i + x d + i ) = x i + t x d + i c'_{i}=\sigma_{c}(x_{i}+x_{d+i})=x_{i}+\sqrt{t}\,x_{d+i} c i ′ = σ c ( x i + x d + i ) = x i + t x d + i , that is c ′ = g t c'=g^{t} c ′ = g t . Hence λ ⊞ s c d , t = γ ∘ σ Γ t \lambda\boxplus\mathrm{sc}_{d,t}=\gamma\circ\sigma_{\Gamma_{t}} λ ⊞ sc d , t = γ ∘ σ Γ t .
Clause 2 (Norm bound). By (S), 0 = 0 \sqrt{0}=0 0 = 0 , so Γ 0 \Gamma_{0} Γ 0 has the same maps as the marginal datum p r 1 \mathrm{pr}^{1} pr 1 of Affine Data and Affine Substitutions of Noncommutative Polynomials §coordinate , and σ p r 1 = ι 1 \sigma_{\mathrm{pr}^{1}}=\iota^{1} σ pr 1 = ι 1 by Affine Substitutions of Noncommutative Laws: Self-Adjointness, Composition, Moment Formulas and Positivity, and the Coordinate Data §coordinate . By Clause 1, λ ⊞ s c d , 0 = γ ∘ ι 1 = λ \lambda\boxplus\mathrm{sc}_{d,0}=\gamma\circ\iota^{1}=\lambda λ ⊞ sc d , 0 = γ ∘ ι 1 = λ .
Now let λ ∈ Σ d , R \lambda\in\Sigma_{d,R} λ ∈ Σ d , R . Since γ ∈ Σ 2 d \gamma\in\Sigma_{2d} γ ∈ Σ 2 d , it lies in some Σ 2 d , r \Sigma_{2d,r} Σ 2 d , r (Noncommutative Laws of Finitely Many Self-Adjoint Variables with a Norm Bound §law ), so the multiplication operators L p L_{p} L p on H γ \mathcal{H}_{\gamma} H γ , p ∈ P 2 d p\in\mathcal{P}_{2d} p ∈ P 2 d , are those 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 . Apply Marginals of a Noncommutative Law: the Isometry of GNS Spaces, the Trace-Preserving Embedding of Tracial Algebras and the Conditional Expectation to γ \gamma γ with m = 2 d m=2d m = 2 d , n = d n=d n = d and a = ( x 1 , … , x d ) a=(x_{1},\dots,x_{d}) a = ( x 1 , … , x d ) , so σ a = ι 1 \sigma_{a}=\iota^{1} σ a = ι 1 and the marginal is λ \lambda λ , with embedding π : M λ → M γ \pi:\mathcal{M}_{\lambda}\to\mathcal{M}_{\gamma} π : M λ → M γ . For j ∈ [ d ] j\in[d] j ∈ [ d ] the operator L x j L_{x_{j}} L x j on H λ \mathcal{H}_{\lambda} H λ commutes with every R p R_{p} R p (Left and Right Multiplication Operators on the Complex GNS Space of a Noncommutative Law: Boundedness, Algebra Rules, Adjoints, Commutation, the Vacuum and the Conjugation §commute ), so it lies in M λ \mathcal{M}_{\lambda} M λ (The Tracial Algebra of a Noncommutative Law and Its Trace §algebra ), and has norm at most R R R by 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 . By Marginals of a Noncommutative Law: the Isometry of GNS Spaces, the Trace-Preserving Embedding of Tracial Algebras and the Conditional Expectation §homomorphism its image π ( L x j ) \pi(L_{x_{j}}) π ( L x j ) is L ι 1 ( x j ) = L x j L_{\iota^{1}(x_{j})}=L_{x_{j}} L ι 1 ( x j ) = L x j on H γ \mathcal{H}_{\gamma} H γ , and by Marginals of a Noncommutative Law: the Isometry of GNS Spaces, the Trace-Preserving Embedding of Tracial Algebras and the Conditional Expectation §embedding ∥ L x j ∥ o p ≤ R \lVert L_{x_{j}}\rVert_{\mathrm{op}}\le R ∥ L x j ∥ op ≤ R on H γ \mathcal{H}_{\gamma} H γ . The same argument with a = ( x d + 1 , … , x 2 d ) a=(x_{d+1},\dots,x_{2d}) a = ( x d + 1 , … , x 2 d ) , whose marginal is s c d ∈ Σ d , 2 \mathrm{sc}_{d}\in\Sigma_{d,2} sc d ∈ Σ d , 2 by (SC), gives ∥ L x d + j ∥ o p ≤ 2 \lVert L_{x_{d+j}}\rVert_{\mathrm{op}}\le2 ∥ L x d + j ∥ op ≤ 2 on H γ \mathcal{H}_{\gamma} H γ . By Left and Right Multiplication Operators on the Complex GNS Space of a Noncommutative Law: Boundedness, Algebra Rules, Adjoints, Commutation, the Vacuum and the Conjugation §algebra , L g j t = L x j + t L x d + j L_{g^{t}_{j}}=L_{x_{j}}+\sqrt{t}\,L_{x_{d+j}} L g j t = L x j + t L x d + j , so by the triangle inequality and homogeneity of the operator norm ∥ L g j t ∥ o p ≤ R + 2 t \lVert L_{g^{t}_{j}}\rVert_{\mathrm{op}}\le R+2\sqrt{t} ∥ L g j t ∥ op ≤ R + 2 t . Since g t g^{t} g t is self-adjoint, Marginals of a Noncommutative Law: the Isometry of GNS Spaces, the Trace-Preserving Embedding of Tracial Algebras and the Conditional Expectation §law with C = R + 2 t > 0 C=R+2\sqrt{t}>0 C = R + 2 t > 0 gives γ ∘ σ g t ∈ Σ d , R + 2 t \gamma\circ\sigma_{g^{t}}\in\Sigma_{d,R+2\sqrt{t}} γ ∘ σ g t ∈ Σ d , R + 2 t , which is λ ⊞ s c d , t \lambda\boxplus\mathrm{sc}_{d,t} λ ⊞ sc d , t by Clause 1.
Clause 3 (Moments). First the moments of γ \gamma γ . For i , k ∈ [ d ] i,k\in[d] i , k ∈ [ d ] , by (C) and the marginals of γ \gamma γ : m i ( γ ) = γ ( ι 1 ( x i ) ) = m i ( λ ) \mathrm{m}_{i}(\gamma)=\gamma(\iota^{1}(x_{i}))=\mathrm{m}_{i}(\lambda) m i ( γ ) = γ ( ι 1 ( x i )) = m i ( λ ) , m i k ( γ ) = γ ( ι 1 ( x i x k ) ) = m i k ( λ ) \mathrm{m}_{ik}(\gamma)=\gamma(\iota^{1}(x_{i}x_{k}))=\mathrm{m}_{ik}(\lambda) m ik ( γ ) = γ ( ι 1 ( x i x k )) = m ik ( λ ) , m d + i ( γ ) = s c d ( x i ) = 0 \mathrm{m}_{d+i}(\gamma)=\mathrm{sc}_{d}(x_{i})=0 m d + i ( γ ) = sc d ( x i ) = 0 and m d + i , d + k ( γ ) = s c d ( x i x k ) = δ i k \mathrm{m}_{d+i,d+k}(\gamma)=\mathrm{sc}_{d}(x_{i}x_{k})=\delta_{ik} m d + i , d + k ( γ ) = sc d ( x i x k ) = δ ik by (SC). For the mixed moments, p 1 = x i − m i ( λ ) 1 ∈ P d p_{1}=x_{i}-\mathrm{m}_{i}(\lambda)1\in\mathcal{P}_{d} p 1 = x i − m i ( λ ) 1 ∈ P d and p 2 = x k ∈ P d p_{2}=x_{k}\in\mathcal{P}_{d} p 2 = x k ∈ P d satisfy γ ( ι 1 ( p 1 ) ) = λ ( x i ) − m i ( λ ) = 0 \gamma(\iota^{1}(p_{1}))=\lambda(x_{i})-\mathrm{m}_{i}(\lambda)=0 γ ( ι 1 ( p 1 )) = λ ( x i ) − m i ( λ ) = 0 and γ ( ι 2 ( p 2 ) ) = s c d ( x k ) = 0 \gamma(\iota^{2}(p_{2}))=\mathrm{sc}_{d}(x_{k})=0 γ ( ι 2 ( p 2 )) = sc d ( x k ) = 0 , so freeness of γ \gamma γ (Freeness of Two Groups of Variables under a Noncommutative Law §free with ( e 1 , e 2 ) = ( 1 , 2 ) (e_{1},e_{2})=(1,2) ( e 1 , e 2 ) = ( 1 , 2 ) ) gives γ ( ( x i − m i ( λ ) 1 ) x d + k ) = 0 \gamma\bigl((x_{i}-\mathrm{m}_{i}(\lambda)1)x_{d+k}\bigr)=0 γ ( ( x i − m i ( λ ) 1 ) x d + k ) = 0 , hence m i , d + k ( γ ) = m i ( λ ) m d + k ( γ ) = 0 \mathrm{m}_{i,d+k}(\gamma)=\mathrm{m}_{i}(\lambda)\,\mathrm{m}_{d+k}(\gamma)=0 m i , d + k ( γ ) = m i ( λ ) m d + k ( γ ) = 0 ; and m d + k , i ( γ ) = m i , d + k ( γ ) = 0 \mathrm{m}_{d+k,i}(\gamma)=\mathrm{m}_{i,d+k}(\gamma)=0 m d + k , i ( γ ) = m i , d + k ( γ ) = 0 by the symmetry in Affine Substitutions of Noncommutative Laws: Self-Adjointness, Composition, Moment Formulas and Positivity, and the Coordinate Data §moments .
By Clause 1, λ ⊞ s c d , t = γ ∘ σ Γ t \lambda\boxplus\mathrm{sc}_{d,t}=\gamma\circ\sigma_{\Gamma_{t}} λ ⊞ sc d , t = γ ∘ σ Γ t , and Affine Substitutions of Noncommutative Laws: Self-Adjointness, Composition, Moment Formulas and Positivity, and the Coordinate Data §moments with T = Γ t T=\Gamma_{t} T = Γ t (whose translation part is 0 0 0 , and whose row i i i has the entries 1 1 1 at i i i and t \sqrt{t} t at d + i d+i d + i ) gives
m i ( γ ∘ σ Γ t ) = m i ( γ ) + t m d + i ( γ ) = m i ( λ ) , \mathrm{m}_{i}(\gamma\circ\sigma_{\Gamma_{t}})=\mathrm{m}_{i}(\gamma)+\sqrt{t}\,\mathrm{m}_{d+i}(\gamma)=\mathrm{m}_{i}(\lambda), m i ( γ ∘ σ Γ t ) = m i ( γ ) + t m d + i ( γ ) = m i ( λ ) ,
m i k ( γ ∘ σ Γ t ) = m i k ( γ ) + t m i , d + k ( γ ) + t m d + i , k ( γ ) + t m d + i , d + k ( γ ) = m i k ( λ ) + t δ i k , \mathrm{m}_{ik}(\gamma\circ\sigma_{\Gamma_{t}})=\mathrm{m}_{ik}(\gamma)+\sqrt{t}\,\mathrm{m}_{i,d+k}(\gamma)+\sqrt{t}\,\mathrm{m}_{d+i,k}(\gamma)+t\,\mathrm{m}_{d+i,d+k}(\gamma)=\mathrm{m}_{ik}(\lambda)+t\,\delta_{ik}, m ik ( γ ∘ σ Γ t ) = m ik ( γ ) + t m i , d + k ( γ ) + t m d + i , k ( γ ) + t m d + i , d + k ( γ ) = m ik ( λ ) + t δ ik ,
using t t = t \sqrt{t}\sqrt{t}=t t t = t (Existence and Uniqueness of the Nonnegative Square Root ). Summing over i = k ∈ [ d ] i=k\in[d] i = k ∈ [ d ] gives M ( λ ⊞ s c d , t ) = M ( λ ) + t d M(\lambda\boxplus\mathrm{sc}_{d,t})=M(\lambda)+td M ( λ ⊞ sc d , t ) = M ( λ ) + t d , since M = ∑ i m i i M=\sum_{i}\mathrm{m}_{ii} M = ∑ i m ii by the statement of Affine Substitutions of Noncommutative Laws: Self-Adjointness, Composition, Moment Formulas and Positivity, and the Coordinate Data .
Clause 4 (Distance to the initial law). Let e = ( g 1 t , … , g d t , x 1 , … , x d ) e=(g^{t}_{1},\dots,g^{t}_{d},x_{1},\dots,x_{d}) e = ( g 1 t , … , g d t , x 1 , … , x d ) , the tuple of the affine datum from 2 d 2d 2 d to 2 d 2d 2 d variables with translation 0 0 0 whose row i ∈ [ d ] i\in[d] i ∈ [ d ] has entries 1 1 1 at i i i and t \sqrt{t} t at d + i d+i d + i , and whose row d + i d+i d + i has entry 1 1 1 at i i i , all other entries 0 0 0 . Then π = γ ∘ σ e \pi=\gamma\circ\sigma_{e} π = γ ∘ σ e is a tracial state on P 2 d \mathcal{P}_{2d} P 2 d . By (C), π ∘ ι 1 = γ ∘ σ g t = λ ⊞ s c d , t \pi\circ\iota^{1}=\gamma\circ\sigma_{g^{t}}=\lambda\boxplus\mathrm{sc}_{d,t} π ∘ ι 1 = γ ∘ σ g t = λ ⊞ sc d , t (Clause 1) and π ∘ ι 2 = γ ∘ σ ( x 1 , … , x d ) = γ ∘ ι 1 = λ \pi\circ\iota^{2}=\gamma\circ\sigma_{(x_{1},\dots,x_{d})}=\gamma\circ\iota^{1}=\lambda π ∘ ι 2 = γ ∘ σ ( x 1 , … , x d ) = γ ∘ ι 1 = λ , so π ∈ Π ( λ ⊞ s c d , t , λ ) \pi\in\Pi(\lambda\boxplus\mathrm{sc}_{d,t},\lambda) π ∈ Π ( λ ⊞ sc d , t , λ ) (Couplings of Two Noncommutative Laws and Their Quadratic Cost §coupling ). By (C), σ e ( Δ d ) = ∑ j = 1 d ( e j − e d + j ) 2 = ∑ j = 1 d ( t x d + j ) 2 = t ι 2 ( ∑ j = 1 d x j x j ) \sigma_{e}(\Delta_{d})=\sum_{j=1}^{d}(e_{j}-e_{d+j})^{2}=\sum_{j=1}^{d}(\sqrt{t}\,x_{d+j})^{2}=t\,\iota^{2}\bigl(\sum_{j=1}^{d}x_{j}x_{j}\bigr) σ e ( Δ d ) = ∑ j = 1 d ( e j − e d + j ) 2 = ∑ j = 1 d ( t x d + j ) 2 = t ι 2 ( ∑ j = 1 d x j x j ) , so by Couplings of Two Noncommutative Laws and Their Quadratic Cost §cost and (SC)
I ( π ) = t ∑ j = 1 d s c d ( x j x j ) = t d . I(\pi)=t\sum_{j=1}^{d}\mathrm{sc}_{d}(x_{j}x_{j})=td. I ( π ) = t j = 1 ∑ d sc d ( x j x j ) = t d .
The infimum defining W 2 ( λ ⊞ s c d , t , λ ) 2 W_{2}(\lambda\boxplus\mathrm{sc}_{d,t},\lambda)^{2} W 2 ( λ ⊞ sc d , t , λ ) 2 in The Noncommutative Quadratic Wasserstein Distance and Optimal Couplings §distance is at most I ( π ) = t d = ( t d ) 2 I(\pi)=td=(\sqrt{td})^{2} I ( π ) = t d = ( t d ) 2 , and (Q) gives W 2 ( λ ⊞ s c d , t , λ ) ≤ t d W_{2}(\lambda\boxplus\mathrm{sc}_{d,t},\lambda)\le\sqrt{td} W 2 ( λ ⊞ sc d , t , λ ) ≤ t d .
Clause 5 (Contraction). By Noncommutative Laws of Finitely Many Self-Adjoint Variables with a Norm Bound §law there are reals R 1 , R 2 > 0 R_{1},R_{2}>0 R 1 , R 2 > 0 with μ ∈ Σ d , R 1 \mu\in\Sigma_{d,R_{1}} μ ∈ Σ d , R 1 and ν ∈ Σ d , R 2 \nu\in\Sigma_{d,R_{2}} ν ∈ Σ d , R 2 ; with R = max ( R 1 , R 2 ) R=\max(R_{1},R_{2}) R = max ( R 1 , R 2 ) , (B) gives μ , ν ∈ Σ d , R \mu,\nu\in\Sigma_{d,R} μ , ν ∈ Σ d , R . By The Noncommutative Wasserstein Distance: Existence of Optimal Couplings, Symmetry, Separation, a Moment Bound, Weak-Star Lower Semicontinuity, and Displacement Interpolation §attained there is an optimal coupling ρ ∈ Π ( μ , ν ) \rho\in\Pi(\mu,\nu) ρ ∈ Π ( μ , ν ) , and ρ ∈ Σ 2 d , R ⊆ Σ 2 d \rho\in\Sigma_{2d,R}\subseteq\Sigma_{2d} ρ ∈ Σ 2 d , R ⊆ Σ 2 d by Couplings of Noncommutative Laws: the Norm Bound, the Cost Identity, the Tensor, Diagonal and Swapped Couplings, Weak-Star Closedness, and Displacement Interpolants §bound . Let Θ = ρ ⋆ s c d ∈ Σ 3 d \Theta=\rho\star\mathrm{sc}_{d}\in\Sigma_{3d} Θ = ρ ⋆ sc d ∈ Σ 3 d for the split 3 d = 2 d + d 3d=2d+d 3 d = 2 d + d , so Θ ∘ ι 2 d , d 1 = ρ \Theta\circ\iota^{1}_{2d,d}=\rho Θ ∘ ι 2 d , d 1 = ρ by The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §product . Let f f f be the 2 d 2d 2 d -tuple in P 3 d \mathcal{P}_{3d} P 3 d with f j = x j + t x 2 d + j f_{j}=x_{j}+\sqrt{t}\,x_{2d+j} f j = x j + t x 2 d + j and f d + j = x d + j + t x 2 d + j f_{d+j}=x_{d+j}+\sqrt{t}\,x_{2d+j} f d + j = x d + j + t x 2 d + j (j ∈ [ d ] j\in[d] j ∈ [ d ] ), the tuple of an affine datum from 3 d 3d 3 d to 2 d 2d 2 d variables with translation 0 0 0 ; so π = Θ ∘ σ f \pi=\Theta\circ\sigma_{f} π = Θ ∘ σ f is a tracial state on P 2 d \mathcal{P}_{2d} P 2 d .
For k ∈ { 1 , 2 } k\in\{1,2\} k ∈ { 1 , 2 } apply The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §substitution with m = 2 d m=2d m = 2 d , n = d n=d n = d , m ′ = n ′ = d m'=n'=d m ′ = n ′ = d , b = ( x 1 , … , x d ) b=(x_{1},\dots,x_{d}) b = ( x 1 , … , x d ) in P d \mathcal{P}_{d} P d , and a = ( x 1 , … , x d ) a=(x_{1},\dots,x_{d}) a = ( x 1 , … , x d ) for k = 1 k=1 k = 1 , a = ( x d + 1 , … , x 2 d ) a=(x_{d+1},\dots,x_{2d}) a = ( x d + 1 , … , x 2 d ) for k = 2 k=2 k = 2 , in P 2 d \mathcal{P}_{2d} P 2 d ; then σ a = ι k \sigma_{a}=\iota^{k} σ a = ι k (Couplings of Two Noncommutative Laws and Their Quadratic Cost §marginals ) and σ b \sigma_{b} σ b is the identity (Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §identity ). By (C) the tuple of that clause is c k = ( x ( k − 1 ) d + 1 , … , x k d , x 2 d + 1 , … , x 3 d ) c^{k}=(x_{(k-1)d+1},\dots,x_{kd},x_{2d+1},\dots,x_{3d}) c k = ( x ( k − 1 ) d + 1 , … , x k d , x 2 d + 1 , … , x 3 d ) , and
Θ ∘ σ c 1 = ( ρ ∘ ι 1 ) ⋆ s c d = μ ⋆ s c d , Θ ∘ σ c 2 = ( ρ ∘ ι 2 ) ⋆ s c d = ν ⋆ s c d . \Theta\circ\sigma_{c^{1}}=(\rho\circ\iota^{1})\star\mathrm{sc}_{d}=\mu\star\mathrm{sc}_{d},\qquad\Theta\circ\sigma_{c^{2}}=(\rho\circ\iota^{2})\star\mathrm{sc}_{d}=\nu\star\mathrm{sc}_{d}. Θ ∘ σ c 1 = ( ρ ∘ ι 1 ) ⋆ sc d = μ ⋆ sc d , Θ ∘ σ c 2 = ( ρ ∘ ι 2 ) ⋆ sc d = ν ⋆ sc d .
By Clause 1 and (C), μ ⊞ s c d , t = ( Θ ∘ σ c 1 ) ∘ σ g t = Θ ∘ σ h \mu\boxplus\mathrm{sc}_{d,t}=(\Theta\circ\sigma_{c^{1}})\circ\sigma_{g^{t}}=\Theta\circ\sigma_{h} μ ⊞ sc d , t = ( Θ ∘ σ c 1 ) ∘ σ g t = Θ ∘ σ h with h j = σ c 1 ( x j + t x d + j ) = x j + t x 2 d + j = f j h_{j}=\sigma_{c^{1}}(x_{j}+\sqrt{t}\,x_{d+j})=x_{j}+\sqrt{t}\,x_{2d+j}=f_{j} h j = σ c 1 ( x j + t x d + j ) = x j + t x 2 d + j = f j , while π ∘ ι 1 = Θ ∘ σ ( f 1 , … , f d ) \pi\circ\iota^{1}=\Theta\circ\sigma_{(f_{1},\dots,f_{d})} π ∘ ι 1 = Θ ∘ σ ( f 1 , … , f d ) by (C); so π ∘ ι 1 = μ ⊞ s c d , t \pi\circ\iota^{1}=\mu\boxplus\mathrm{sc}_{d,t} π ∘ ι 1 = μ ⊞ sc d , t . Likewise σ c 2 ( x j + t x d + j ) = x d + j + t x 2 d + j = f d + j \sigma_{c^{2}}(x_{j}+\sqrt{t}\,x_{d+j})=x_{d+j}+\sqrt{t}\,x_{2d+j}=f_{d+j} σ c 2 ( x j + t x d + j ) = x d + j + t x 2 d + j = f d + j gives π ∘ ι 2 = ν ⊞ s c d , t \pi\circ\iota^{2}=\nu\boxplus\mathrm{sc}_{d,t} π ∘ ι 2 = ν ⊞ sc d , t . Thus π ∈ Π ( μ ⊞ s c d , t , ν ⊞ s c d , t ) \pi\in\Pi(\mu\boxplus\mathrm{sc}_{d,t},\nu\boxplus\mathrm{sc}_{d,t}) π ∈ Π ( μ ⊞ sc d , t , ν ⊞ sc d , t ) . In the cost the noise cancels: by (C), σ f ( Δ d ) = ∑ j = 1 d ( f j − f d + j ) 2 = ∑ j = 1 d ( x j − x d + j ) 2 = ι 2 d , d 1 ( Δ d ) \sigma_{f}(\Delta_{d})=\sum_{j=1}^{d}(f_{j}-f_{d+j})^{2}=\sum_{j=1}^{d}(x_{j}-x_{d+j})^{2}=\iota^{1}_{2d,d}(\Delta_{d}) σ f ( Δ d ) = ∑ j = 1 d ( f j − f d + j ) 2 = ∑ j = 1 d ( x j − x d + j ) 2 = ι 2 d , d 1 ( Δ d ) , so
I ( π ) = Θ ( ι 2 d , d 1 ( Δ d ) ) = ρ ( Δ d ) = I ( ρ ) = W 2 ( μ , ν ) 2 I(\pi)=\Theta\bigl(\iota^{1}_{2d,d}(\Delta_{d})\bigr)=\rho(\Delta_{d})=I(\rho)=W_{2}(\mu,\nu)^{2} I ( π ) = Θ ( ι 2 d , d 1 ( Δ d ) ) = ρ ( Δ d ) = I ( ρ ) = W 2 ( μ , ν ) 2
by Couplings of Two Noncommutative Laws and Their Quadratic Cost §cost and The Noncommutative Quadratic Wasserstein Distance and Optimal Couplings §optimal . Hence W 2 ( μ ⊞ s c d , t , ν ⊞ s c d , t ) 2 ≤ W 2 ( μ , ν ) 2 W_{2}(\mu\boxplus\mathrm{sc}_{d,t},\nu\boxplus\mathrm{sc}_{d,t})^{2}\le W_{2}(\mu,\nu)^{2} W 2 ( μ ⊞ sc d , t , ν ⊞ sc d , t ) 2 ≤ W 2 ( μ , ν ) 2 by The Noncommutative Quadratic Wasserstein Distance and Optimal Couplings §distance , and (Q) gives the claim.
Clause 6 (Semigroup). Use the split 3 d = d + 2 d 3d=d+2d 3 d = d + 2 d and write Θ ′ = λ ⋆ s c 2 d ∈ Σ 3 d \Theta'=\lambda\star\mathrm{sc}_{2d}\in\Sigma_{3d} Θ ′ = λ ⋆ sc 2 d ∈ Σ 3 d , so Θ ′ ∘ ι d , 2 d 1 = λ \Theta'\circ\iota^{1}_{d,2d}=\lambda Θ ′ ∘ ι d , 2 d 1 = λ by The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §product . By Free Semicircular Variables are Characterised by the Schwinger-Dyson Equation Relative to the Other Variables; Semicircular Systems and Rotations §systems , s c 2 d = s c d ⋆ s c d \mathrm{sc}_{2d}=\mathrm{sc}_{d}\star\mathrm{sc}_{d} sc 2 d = sc d ⋆ sc d (split d + d d+d d + d ), so s c 2 d ∘ ι 1 = s c d \mathrm{sc}_{2d}\circ\iota^{1}=\mathrm{sc}_{d} sc 2 d ∘ ι 1 = sc d by The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §product .
Step 1: an auxiliary law. Let G G G be the affine datum from 3 d 3d 3 d to 2 d 2d 2 d variables with translation 0 0 0 and matrix H H H given by H i , i = 1 H_{i,i}=1 H i , i = 1 , H i , d + i = t H_{i,d+i}=\sqrt{t} H i , d + i = t , H d + i , 2 d + i = 1 H_{d+i,2d+i}=1 H d + i , 2 d + i = 1 (i ∈ [ d ] i\in[d] i ∈ [ d ] ) and all other entries 0 0 0 ; its tuple is ( x 1 + t x d + 1 , … , x d + t x 2 d , x 2 d + 1 , … , x 3 d ) (x_{1}+\sqrt{t}\,x_{d+1},\dots,x_{d}+\sqrt{t}\,x_{2d},x_{2d+1},\dots,x_{3d}) ( x 1 + t x d + 1 , … , x d + t x 2 d , x 2 d + 1 , … , x 3 d ) . Put κ = Θ ′ ∘ σ G ∈ Σ 2 d \kappa=\Theta'\circ\sigma_{G}\in\Sigma_{2d} κ = Θ ′ ∘ σ G ∈ Σ 2 d . Apply The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §substitution with m = d m=d m = d , n = 2 d n=2d n = 2 d , m ′ = n ′ = d m'=n'=d m ′ = n ′ = d , a = ( x 1 , … , x d ) a=(x_{1},\dots,x_{d}) a = ( x 1 , … , x d ) in P d \mathcal{P}_{d} P d and b = ( x 1 , … , x d ) b=(x_{1},\dots,x_{d}) b = ( x 1 , … , x d ) in P 2 d \mathcal{P}_{2d} P 2 d , so σ a \sigma_{a} σ a is the identity and σ b = ι 1 \sigma_{b}=\iota^{1} σ b = ι 1 : by (C) the tuple of that clause is c = ( x 1 , … , x 2 d ) c=(x_{1},\dots,x_{2d}) c = ( x 1 , … , x 2 d ) in P 3 d \mathcal{P}_{3d} P 3 d and Θ ′ ∘ σ c = λ ⋆ ( s c 2 d ∘ ι 1 ) = λ ⋆ s c d \Theta'\circ\sigma_{c}=\lambda\star(\mathrm{sc}_{2d}\circ\iota^{1})=\lambda\star\mathrm{sc}_{d} Θ ′ ∘ σ c = λ ⋆ ( sc 2 d ∘ ι 1 ) = λ ⋆ sc d . By (C) and Clause 1, κ ∘ ι 1 = Θ ′ ∘ σ ( x 1 + t x d + 1 , … , x d + t x 2 d ) = ( Θ ′ ∘ σ c ) ∘ σ g t = ( λ ⋆ s c d ) ∘ σ Γ t = λ ⊞ s c d , t \kappa\circ\iota^{1}=\Theta'\circ\sigma_{(x_{1}+\sqrt{t}x_{d+1},\dots,x_{d}+\sqrt{t}x_{2d})}=(\Theta'\circ\sigma_{c})\circ\sigma_{g^{t}}=(\lambda\star\mathrm{sc}_{d})\circ\sigma_{\Gamma_{t}}=\lambda\boxplus\mathrm{sc}_{d,t} κ ∘ ι 1 = Θ ′ ∘ σ ( x 1 + t x d + 1 , … , x d + t x 2 d ) = ( Θ ′ ∘ σ c ) ∘ σ g t = ( λ ⋆ sc d ) ∘ σ Γ t = λ ⊞ sc d , t , since σ c ( g j t ) = g j t \sigma_{c}(g^{t}_{j})=g^{t}_{j} σ c ( g j t ) = g j t in P 3 d \mathcal{P}_{3d} P 3 d .
Step 2: κ = ( λ ⊞ s c d , t ) ⋆ s c d \kappa=(\lambda\boxplus\mathrm{sc}_{d,t})\star\mathrm{sc}_{d} κ = ( λ ⊞ sc d , t ) ⋆ sc d . By the forward direction of Free Semicircular Variables are Characterised by the Schwinger-Dyson Equation Relative to the Other Variables; Semicircular Systems and Rotations §characterisation (with m = d m=d m = d , n = 2 d n=2d n = 2 d , α = λ \alpha=\lambda α = λ , applied to Θ ′ \Theta' Θ ′ ), Θ ′ \Theta' Θ ′ satisfies the Schwinger-Dyson equation in the last 2 d 2d 2 d variables . Let j ∈ [ d ] j\in[d] j ∈ [ d ] and q ∈ P 2 d q\in\mathcal{P}_{2d} q ∈ P 2 d . By (C), σ G ( x d + j q ) = x 2 d + j σ G ( q ) \sigma_{G}(x_{d+j}q)=x_{2d+j}\,\sigma_{G}(q) σ G ( x d + j q ) = x 2 d + j σ G ( q ) , so
κ ( x d + j q ) = Θ ′ ( x 2 d + j σ G ( q ) ) = ∂ 2 d + j Θ ′ ( σ G ( q ) ) = ∑ i = 1 2 d H i , 2 d + j ∂ i κ ( q ) = ∂ d + j κ ( q ) , \kappa(x_{d+j}q)=\Theta'\bigl(x_{2d+j}\,\sigma_{G}(q)\bigr)=\partial^{\Theta'}_{2d+j}\bigl(\sigma_{G}(q)\bigr)=\sum_{i=1}^{2d}H_{i,2d+j}\,\partial^{\kappa}_{i}(q)=\partial^{\kappa}_{d+j}(q), κ ( x d + j q ) = Θ ′ ( x 2 d + j σ G ( q ) ) = ∂ 2 d + j Θ ′ ( σ G ( q ) ) = i = 1 ∑ 2 d H i , 2 d + j ∂ i κ ( q ) = ∂ d + j κ ( q ) ,
the third equality by The Chain Rule for Free Difference Quotients under an Affine Substitution §chain-rule (with m = 3 d m=3d m = 3 d , n = 2 d n=2d n = 2 d , T = G T=G T = G , k = 2 d + j k=2d+j k = 2 d + j ) and the last because H i , 2 d + j H_{i,2d+j} H i , 2 d + j is 1 1 1 for i = d + j i=d+j i = d + j and 0 0 0 otherwise. Thus κ \kappa κ satisfies the Schwinger-Dyson equation in its last d d d variables, and since κ ∘ ι 1 = λ ⊞ s c d , t ∈ Σ d \kappa\circ\iota^{1}=\lambda\boxplus\mathrm{sc}_{d,t}\in\Sigma_{d} κ ∘ ι 1 = λ ⊞ sc d , t ∈ Σ d , the reverse direction of Free Semicircular Variables are Characterised by the Schwinger-Dyson Equation Relative to the Other Variables; Semicircular Systems and Rotations §characterisation (with m = n = d m=n=d m = n = d ) gives κ = ( λ ⊞ s c d , t ) ⋆ s c d \kappa=(\lambda\boxplus\mathrm{sc}_{d,t})\star\mathrm{sc}_{d} κ = ( λ ⊞ sc d , t ) ⋆ sc d .
Step 3: the left side. By Clause 1 for the law λ ⊞ s c d , t \lambda\boxplus\mathrm{sc}_{d,t} λ ⊞ sc d , t and the time s s s , Step 2 and (C),
( λ ⊞ s c d , t ) ⊞ s c d , s = κ ∘ σ Γ s = Θ ′ ∘ σ u , u i = σ G ( x i + s x d + i ) = x i + t x d + i + s x 2 d + i ( i ∈ [ d ] ) . (\lambda\boxplus\mathrm{sc}_{d,t})\boxplus\mathrm{sc}_{d,s}=\kappa\circ\sigma_{\Gamma_{s}}=\Theta'\circ\sigma_{u},\qquad u_{i}=\sigma_{G}(x_{i}+\sqrt{s}\,x_{d+i})=x_{i}+\sqrt{t}\,x_{d+i}+\sqrt{s}\,x_{2d+i}\quad(i\in[d]). ( λ ⊞ sc d , t ) ⊞ sc d , s = κ ∘ σ Γ s = Θ ′ ∘ σ u , u i = σ G ( x i + s x d + i ) = x i + t x d + i + s x 2 d + i ( i ∈ [ d ]) .
Step 4: the right side. If t + s = 0 t+s=0 t + s = 0 , then t = s = 0 t=s=0 t = s = 0 , and both sides equal λ \lambda λ by two applications of Clause 2. Otherwise v = t + s > 0 v=t+s>0 v = t + s > 0 ; let O : [ d ] × [ 2 d ] → R O:[d]\times[2d]\to\mathbb{R} O : [ d ] × [ 2 d ] → R have O i , i = t / v O_{i,i}=\sqrt{t/v} O i , i = t / v , O i , d + i = s / v O_{i,d+i}=\sqrt{s/v} O i , d + i = s / v and all other entries 0 0 0 . For i , k ∈ [ d ] i,k\in[d] i , k ∈ [ d ] , ∑ l = 1 2 d O i l O k l \sum_{l=1}^{2d}O_{il}O_{kl} ∑ l = 1 2 d O i l O k l is 0 0 0 if i ≠ k i\ne k i = k and ( t / v ) 2 + ( s / v ) 2 = t / v + s / v = 1 (\sqrt{t/v})^{2}+(\sqrt{s/v})^{2}=t/v+s/v=1 ( t / v ) 2 + ( s / v ) 2 = t / v + s / v = 1 if i = k i=k i = k , so Free Semicircular Variables are Characterised by the Schwinger-Dyson Equation Relative to the Other Variables; Semicircular Systems and Rotations §rotation (with k = d k=d k = d , n = 2 d n=2d n = 2 d ) gives s c 2 d ∘ σ ( O , 0 ) = s c d \mathrm{sc}_{2d}\circ\sigma_{(O,0)}=\mathrm{sc}_{d} sc 2 d ∘ σ ( O , 0 ) = sc d . Apply The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §substitution with m = d m=d m = d , n = 2 d n=2d n = 2 d , m ′ = n ′ = d m'=n'=d m ′ = n ′ = d , a = ( x 1 , … , x d ) a=(x_{1},\dots,x_{d}) a = ( x 1 , … , x d ) and b b b the tuple of ( O , 0 ) (O,0) ( O , 0 ) , whose entries are b i = t / v x i + s / v x d + i b_{i}=\sqrt{t/v}\,x_{i}+\sqrt{s/v}\,x_{d+i} b i = t / v x i + s / v x d + i : by (C) the tuple of that clause is c ′ = ( x 1 , … , x d , b 1 ′ , … , b d ′ ) c'=(x_{1},\dots,x_{d},b'_{1},\dots,b'_{d}) c ′ = ( x 1 , … , x d , b 1 ′ , … , b d ′ ) with b i ′ = t / v x d + i + s / v x 2 d + i b'_{i}=\sqrt{t/v}\,x_{d+i}+\sqrt{s/v}\,x_{2d+i} b i ′ = t / v x d + i + s / v x 2 d + i , and Θ ′ ∘ σ c ′ = λ ⋆ ( s c 2 d ∘ σ ( O , 0 ) ) = λ ⋆ s c d \Theta'\circ\sigma_{c'}=\lambda\star(\mathrm{sc}_{2d}\circ\sigma_{(O,0)})=\lambda\star\mathrm{sc}_{d} Θ ′ ∘ σ c ′ = λ ⋆ ( sc 2 d ∘ σ ( O , 0 ) ) = λ ⋆ sc d . By Clause 1 and (C), λ ⊞ s c d , v = ( Θ ′ ∘ σ c ′ ) ∘ σ Γ v = Θ ′ ∘ σ u ′ \lambda\boxplus\mathrm{sc}_{d,v}=(\Theta'\circ\sigma_{c'})\circ\sigma_{\Gamma_{v}}=\Theta'\circ\sigma_{u'} λ ⊞ sc d , v = ( Θ ′ ∘ σ c ′ ) ∘ σ Γ v = Θ ′ ∘ σ u ′ with
u i ′ = σ c ′ ( x i + v x d + i ) = x i + v t / v x d + i + v s / v x 2 d + i = u i , u'_{i}=\sigma_{c'}(x_{i}+\sqrt{v}\,x_{d+i})=x_{i}+\sqrt{v}\sqrt{t/v}\,x_{d+i}+\sqrt{v}\sqrt{s/v}\,x_{2d+i}=u_{i}, u i ′ = σ c ′ ( x i + v x d + i ) = x i + v t / v x d + i + v s / v x 2 d + i = u i ,
because v t / v = t \sqrt{v}\sqrt{t/v}=\sqrt{t} v t / v = t and v s / v = s \sqrt{v}\sqrt{s/v}=\sqrt{s} v s / v = s by (S). Comparing with Step 3 gives the claim.
Clause 7 (Continuity). Time. Fix p ∈ P d p\in\mathcal{P}_{d} p ∈ P d . For a word w = ( w 1 , … , w k ) ∈ W d w=(w_{1},\dots,w_{k})\in W_{d} w = ( w 1 , … , w k ) ∈ W d and S ⊆ [ k ] S\subseteq[k] S ⊆ [ k ] let w S ∈ W 2 d w^{S}\in W_{2d} w S ∈ W 2 d be the word of length k k k whose l l l -th letter is d + w l d+w_{l} d + w l if l ∈ S l\in S l ∈ S and w l w_{l} w l otherwise. By Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §values , σ g r ( x w ) \sigma_{g^{r}}(x_{w}) σ g r ( x w ) is the product along w w w of the g w l r = x w l + r x d + w l g^{r}_{w_{l}}=x_{w_{l}}+\sqrt{r}\,x_{d+w_{l}} g w l r = x w l + r x d + w l , and an induction on k k k using distributivity (Noncommutative Polynomials Form a Unital Complex Algebra with Involution: Linear Extension from Monomials, Products, Adjoints and Self-Adjoint Parts §algebra ) and x u x u ′ = x u u ′ x_{u}x_{u'}=x_{uu'} x u x u ′ = x u u ′ (Noncommutative Polynomials Form a Unital Complex Algebra with Involution: Linear Extension from Monomials, Products, Adjoints and Self-Adjoint Parts §monomials ) gives
σ g r ( x w ) = ∑ S ⊆ [ k ] ( r ) ∣ S ∣ x w S ( r ≥ 0 ) , \sigma_{g^{r}}(x_{w})=\sum_{S\subseteq[k]}(\sqrt{r})^{|S|}\,x_{w^{S}}\qquad(r\ge0), σ g r ( x w ) = S ⊆ [ k ] ∑ ( r ) ∣ S ∣ x w S ( r ≥ 0 ) ,
and σ g r ( x ∅ ) = 1 \sigma_{g^{r}}(x_{\varnothing})=1 σ g r ( x ∅ ) = 1 . Writing p p p as the finite sum of p ( w ) x w p(w)x_{w} p ( w ) x w over the words w w w in its support, linearity of σ g r \sigma_{g^{r}} σ g r and of γ \gamma γ and Clause 1 give, for r ≥ 0 r\ge0 r ≥ 0 ,
Re ( λ ⊞ s c d , r ) ( p ) = ∑ w ∑ S ( r ) ∣ S ∣ Re ( p ( w ) γ ( x w S ) ) , \operatorname{Re}(\lambda\boxplus\mathrm{sc}_{d,r})(p)=\sum_{w}\sum_{S}(\sqrt{r})^{|S|}\operatorname{Re}\bigl(p(w)\,\gamma(x_{w^{S}})\bigr), Re ( λ ⊞ sc d , r ) ( p ) = w ∑ S ∑ ( r ) ∣ S ∣ Re ( p ( w ) γ ( x w S ) ) ,
and the same with Im \operatorname{Im} Im , where the complex numbers p ( w ) γ ( x w S ) p(w)\gamma(x_{w^{S}}) p ( w ) γ ( x w S ) do not depend on r r r (for the empty word the inner sum is the single term p ( ∅ ) γ ( 1 ) p(\varnothing)\gamma(1) p ( ∅ ) γ ( 1 ) ). The map r ↦ r r\mapsto\sqrt{r} r ↦ r is continuous on [ 0 , ∞ ) [0,\infty) [ 0 , ∞ ) : for r ≥ r 0 ≥ 0 r\ge r_{0}\ge0 r ≥ r 0 ≥ 0 we have 0 ≤ r − r 0 0\le\sqrt{r}-\sqrt{r_{0}} 0 ≤ r − r 0 (by (Q), as r 0 ≤ r r_{0}\le r r 0 ≤ r ) and ( r − r 0 ) 2 ≤ ( r − r 0 ) ( r + r 0 ) = r − r 0 (\sqrt{r}-\sqrt{r_{0}})^{2}\le(\sqrt{r}-\sqrt{r_{0}})(\sqrt{r}+\sqrt{r_{0}})=r-r_{0} ( r − r 0 ) 2 ≤ ( r − r 0 ) ( r + r 0 ) = r − r 0 , so in all cases ∣ r − r 0 ∣ 2 ≤ ∣ r − r 0 ∣ |\sqrt{r}-\sqrt{r_{0}}|^{2}\le|r-r_{0}| ∣ r − r 0 ∣ 2 ≤ ∣ r − r 0 ∣ ; given ε > 0 \varepsilon>0 ε > 0 , δ = ε 2 \delta=\varepsilon^{2} δ = ε 2 satisfies Continuous Map Between Metric Spaces for the metric of The Absolute Value Metric on the Real Numbers , since if ∣ r − r 0 ∣ < ε 2 |r-r_{0}|<\varepsilon^{2} ∣ r − r 0 ∣ < ε 2 and ε ≤ ∣ r − r 0 ∣ \varepsilon\le|\sqrt{r}-\sqrt{r_{0}}| ε ≤ ∣ r − r 0 ∣ , then ε 2 ≤ ∣ r − r 0 ∣ 2 ≤ ∣ r − r 0 ∣ < ε 2 \varepsilon^{2}\le|\sqrt{r}-\sqrt{r_{0}}|^{2}\le|r-r_{0}|<\varepsilon^{2} ε 2 ≤ ∣ r − r 0 ∣ 2 ≤ ∣ r − r 0 ∣ < ε 2 , which is absurd. By claims 1 and 5 of Continuity of Sums and Products of Real-Valued Functions on a Metric Space and induction on ∣ S ∣ |S| ∣ S ∣ , each power r ↦ ( r ) ∣ S ∣ r\mapsto(\sqrt{r})^{|S|} r ↦ ( r ) ∣ S ∣ is continuous on [ 0 , ∞ ) [0,\infty) [ 0 , ∞ ) , and by claim 5 of the same theorem so are its real multiples and the finite sums displayed above. This proves the first assertion.
Law. Let ( μ k ) (\mu_{k}) ( μ k ) in Σ d , R \Sigma_{d,R} Σ d , R converge weak-star to μ \mu μ , and put R ′ = max ( R , 2 ) R'=\max(R,2) R ′ = max ( R , 2 ) . By (B) and (SC), every μ k \mu_{k} μ k and s c d \mathrm{sc}_{d} sc d lie in Σ d , R ′ \Sigma_{d,R'} Σ d , R ′ , and the constant sequence ( s c d ) (\mathrm{sc}_{d}) ( sc d ) converges weak-star to s c d \mathrm{sc}_{d} sc d by Weak-Star Convergence of Noncommutative Laws §weak-star . By The Free Product of Two Noncommutative Laws: Existence, Uniqueness, Norm Bound, Substitutions and Weak-Star Continuity §continuity , μ k ⋆ s c d → μ ⋆ s c d \mu_{k}\star\mathrm{sc}_{d}\to\mu\star\mathrm{sc}_{d} μ k ⋆ sc d → μ ⋆ sc d weak-star. For p ∈ P d p\in\mathcal{P}_{d} p ∈ P d , Clause 1 gives ( μ k ⊞ s c d , t ) ( p ) = ( μ k ⋆ s c d ) ( σ Γ t ( p ) ) (\mu_{k}\boxplus\mathrm{sc}_{d,t})(p)=(\mu_{k}\star\mathrm{sc}_{d})(\sigma_{\Gamma_{t}}(p)) ( μ k ⊞ sc d , t ) ( p ) = ( μ k ⋆ sc d ) ( σ Γ t ( p )) and ( μ ⊞ s c d , t ) ( p ) = ( μ ⋆ s c d ) ( σ Γ t ( p ) ) (\mu\boxplus\mathrm{sc}_{d,t})(p)=(\mu\star\mathrm{sc}_{d})(\sigma_{\Gamma_{t}}(p)) ( μ ⊞ sc d , t ) ( p ) = ( μ ⋆ sc d ) ( σ Γ t ( p )) ; applying Weak-Star Convergence of Noncommutative Laws §weak-star to the fixed polynomial σ Γ t ( p ) ∈ P 2 d \sigma_{\Gamma_{t}}(p)\in\mathcal{P}_{2d} σ Γ t ( p ) ∈ P 2 d , the real and imaginary parts of the former converge to those of the latter. Hence μ k ⊞ s c d , t → μ ⊞ s c d , t \mu_{k}\boxplus\mathrm{sc}_{d,t}\to\mu\boxplus\mathrm{sc}_{d,t} μ k ⊞ sc d , t → μ ⊞ sc d , t weak-star.