Each result cited is universally quantified over the data in its own statement.
Notation. For γ ∈ Σ 3 d 2 \gamma\in\Sigma^{2}_{3d} γ ∈ Σ 3 d 2 put Ψ ( γ ) = p ( γ ) − h ( s ( γ ) ) − θ s ( γ ) 2 \Psi(\gamma)=p(\gamma)-h(s(\gamma))-\theta\,s(\gamma)^{2} Ψ ( γ ) = p ( γ ) − h ( s ( γ )) − θ s ( γ ) 2 , where s ( γ ) s(\gamma) s ( γ ) and p ( γ ) p(\gamma) p ( γ ) are the displacement and momentum pairing of Couplings of a Square-Integrable Plan with a Law: Displacement and Momentum Pairing §displacement-pairing ; by that clause a realisation of γ \gamma γ exists and s ( γ ) s(\gamma) s ( γ ) , p ( γ ) p(\gamma) p ( γ ) may be computed from any realisation. Put a π = ∣ π ∣ m o m a_{\pi}=|\pi|_{\mathrm{mom}} a π = ∣ π ∣ mom and K = a π + L + 4 θ r d K=a_{\pi}+L+4\theta r\sqrt{d} K = a π + L + 4 θ r d , a real number with K > 0 K>0 K > 0 . By The Coupling Test Function of a Square-Integrable Plan §test , Ψ ( γ ) ≤ Φ ( λ ) \Psi(\gamma)\le\Phi(\lambda) Ψ ( γ ) ≤ Φ ( λ ) whenever λ ∈ Σ d 2 \lambda\in\Sigma^{2}_{d} λ ∈ Σ d 2 and γ ∈ C ( λ ) \gamma\in\mathcal{C}(\lambda) γ ∈ C ( λ ) ; and since Φ ( λ ) \Phi(\lambda) Φ ( λ ) is a least upper bound in the sense of The Real Numbers: Standing Notation and Background §bounds , for every real ε > 0 \varepsilon>0 ε > 0 some γ ∈ C ( λ ) \gamma\in\mathcal{C}(\lambda) γ ∈ C ( λ ) has Ψ ( γ ) > Φ ( λ ) − ε \Psi(\gamma)>\Phi(\lambda)-\varepsilon Ψ ( γ ) > Φ ( λ ) − ε . For L 2 L^{2} L 2 d d d -tuples of one tracial W*-probability space ( H , M , Ω ) (H,M,\Omega) ( H , M , Ω ) , the pairing is the inner product and ∥ ⋅ ∥ 2 \lVert\cdot\rVert_{2} ∥ ⋅ ∥ 2 the norm of the complex Hilbert space H d H^{d} H d , sums and real multiples are those of H d H^{d} H d , and the pairing is real, by Sums, Real Multiples and the Pairing of Square-Integrable Tuples in a Tracial W*-Probability Space §operations and Sums, Real Multiples and the Pairing of Square-Integrable Tuples in a Tracial W*-Probability Space §pairing . Hence, by the results in force by Plan Jets and Hamiltonians on Square-Integrable Noncommutative Laws: Standing Notation §background , for all L 2 L^{2} L 2 d d d -tuples U 1 , U 2 , U 3 U_{1},U_{2},U_{3} U 1 , U 2 , U 3 of one space and real t t t we have ∣ ⟨ U 1 , U 2 ⟩ 2 ∣ ≤ ∥ U 1 ∥ 2 ∥ U 2 ∥ 2 |\langle U_{1},U_{2}\rangle_{2}|\le\lVert U_{1}\rVert_{2}\lVert U_{2}\rVert_{2} ∣ ⟨ U 1 , U 2 ⟩ 2 ∣ ≤ ∥ U 1 ∥ 2 ∥ U 2 ∥ 2 , ∥ U 1 + U 2 ∥ 2 ≤ ∥ U 1 ∥ 2 + ∥ U 2 ∥ 2 \lVert U_{1}+U_{2}\rVert_{2}\le\lVert U_{1}\rVert_{2}+\lVert U_{2}\rVert_{2} ∥ U 1 + U 2 ∥ 2 ≤ ∥ U 1 ∥ 2 + ∥ U 2 ∥ 2 , ∥ t U 1 ∥ 2 = ∣ t ∣ ∥ U 1 ∥ 2 \lVert tU_{1}\rVert_{2}=|t|\,\lVert U_{1}\rVert_{2} ∥ t U 1 ∥ 2 = ∣ t ∣ ∥ U 1 ∥ 2 , ⟨ U 1 , U 2 + t U 3 ⟩ 2 = ⟨ U 1 , U 2 ⟩ 2 + t ⟨ U 1 , U 3 ⟩ 2 \langle U_{1},U_{2}+tU_{3}\rangle_{2}=\langle U_{1},U_{2}\rangle_{2}+t\langle U_{1},U_{3}\rangle_{2} ⟨ U 1 , U 2 + t U 3 ⟩ 2 = ⟨ U 1 , U 2 ⟩ 2 + t ⟨ U 1 , U 3 ⟩ 2 , ∥ U 1 + U 2 ∥ 2 2 = ∥ U 1 ∥ 2 2 + 2 ⟨ U 1 , U 2 ⟩ 2 + ∥ U 2 ∥ 2 2 \lVert U_{1}+U_{2}\rVert_{2}^{2}=\lVert U_{1}\rVert_{2}^{2}+2\langle U_{1},U_{2}\rangle_{2}+\lVert U_{2}\rVert_{2}^{2} ∥ U 1 + U 2 ∥ 2 2 = ∥ U 1 ∥ 2 2 + 2 ⟨ U 1 , U 2 ⟩ 2 + ∥ U 2 ∥ 2 2 , and U 1 = 0 U_{1}=0 U 1 = 0 whenever ∥ U 1 ∥ 2 = 0 \lVert U_{1}\rVert_{2}=0 ∥ U 1 ∥ 2 = 0 . These rules are used below without further mention.
Step A (the gauge). First, h ′ ( s ) ≥ 0 h'(s)\ge0 h ′ ( s ) ≥ 0 for every real s ≥ 0 s\ge0 s ≥ 0 . Indeed, fix a real η > 0 \eta>0 η > 0 , let r η r_{\eta} r η be as in the hypothesis at the point s s s , and put t = s + r η / 2 t=s+r_{\eta}/2 t = s + r η /2 . Since h h h is nondecreasing in the sense of Monotone Real Function §nondecreasing , 0 ≤ h ( t ) − h ( s ) ≤ h ′ ( s ) ( t − s ) + η ( t − s ) 0\le h(t)-h(s)\le h'(s)(t-s)+\eta(t-s) 0 ≤ h ( t ) − h ( s ) ≤ h ′ ( s ) ( t − s ) + η ( t − s ) , and dividing by t − s > 0 t-s>0 t − s > 0 gives h ′ ( s ) ≥ − η h'(s)\ge-\eta h ′ ( s ) ≥ − η . As η > 0 \eta>0 η > 0 was arbitrary, h ′ ( s ) ≥ 0 h'(s)\ge0 h ′ ( s ) ≥ 0 .
Second, ∣ h ( t ) − h ( s ) ∣ ≤ L ∣ t − s ∣ |h(t)-h(s)|\le L|t-s| ∣ h ( t ) − h ( s ) ∣ ≤ L ∣ t − s ∣ for all real s , t ≥ 0 s,t\ge0 s , t ≥ 0 . The hypothesis gives, for every real c ≥ 0 c\ge0 c ≥ 0 and η > 0 \eta>0 η > 0 , a real r η ( c ) > 0 r_{\eta}(c)>0 r η ( c ) > 0 such that
∣ h ( t ) − h ( c ) ∣ ≤ ( ∣ h ′ ( c ) ∣ + η ) ∣ t − c ∣ ≤ ( L + η ) ∣ t − c ∣ for every real t ≥ 0 with ∣ t − c ∣ < r η ( c ) . |h(t)-h(c)|\le(|h'(c)|+\eta)|t-c|\le(L+\eta)|t-c|\qquad\text{for every real }t\ge0\text{ with }|t-c|<r_{\eta}(c). ∣ h ( t ) − h ( c ) ∣ ≤ ( ∣ h ′ ( c ) ∣ + η ) ∣ t − c ∣ ≤ ( L + η ) ∣ t − c ∣ for every real t ≥ 0 with ∣ t − c ∣ < r η ( c ) .
Fix reals 0 ≤ a < b 0\le a<b 0 ≤ a < b and η > 0 \eta>0 η > 0 , and let A A A be the set of real t t t with a ≤ t ≤ b a\le t\le b a ≤ t ≤ b and h ( t ) − h ( a ) ≤ ( L + η ) ( t − a ) h(t)-h(a)\le(L+\eta)(t-a) h ( t ) − h ( a ) ≤ ( L + η ) ( t − a ) . Then a ∈ A a\in A a ∈ A and b b b is an upper bound of A A A , so A A A has a least upper bound c c c with a ≤ c ≤ b a\le c\le b a ≤ c ≤ b , by The Real Numbers: Standing Notation and Background §bounds . We show c ∈ A c\in A c ∈ A . If c = a c=a c = a this is clear. Otherwise, by the least-upper-bound property there is t ∈ A t\in A t ∈ A with c − r η ( c ) < t ≤ c c-r_{\eta}(c)<t\le c c − r η ( c ) < t ≤ c ; the displayed estimate at c c c gives h ( c ) − h ( t ) ≤ ( L + η ) ( c − t ) h(c)-h(t)\le(L+\eta)(c-t) h ( c ) − h ( t ) ≤ ( L + η ) ( c − t ) , and adding h ( t ) − h ( a ) ≤ ( L + η ) ( t − a ) h(t)-h(a)\le(L+\eta)(t-a) h ( t ) − h ( a ) ≤ ( L + η ) ( t − a ) gives c ∈ A c\in A c ∈ A . Next, c = b c=b c = b : otherwise t 1 = min ( b , c + r η ( c ) / 2 ) t_{1}=\min(b,c+r_{\eta}(c)/2) t 1 = min ( b , c + r η ( c ) /2 ) satisfies c < t 1 ≤ b c<t_{1}\le b c < t 1 ≤ b , the displayed estimate at c c c gives h ( t 1 ) − h ( c ) ≤ ( L + η ) ( t 1 − c ) h(t_{1})-h(c)\le(L+\eta)(t_{1}-c) h ( t 1 ) − h ( c ) ≤ ( L + η ) ( t 1 − c ) , and adding h ( c ) − h ( a ) ≤ ( L + η ) ( c − a ) h(c)-h(a)\le(L+\eta)(c-a) h ( c ) − h ( a ) ≤ ( L + η ) ( c − a ) gives t 1 ∈ A t_{1}\in A t 1 ∈ A , contradicting t 1 > c t_{1}>c t 1 > c . Hence b ∈ A b\in A b ∈ A , that is, 0 ≤ h ( b ) − h ( a ) ≤ ( L + η ) ( b − a ) 0\le h(b)-h(a)\le(L+\eta)(b-a) 0 ≤ h ( b ) − h ( a ) ≤ ( L + η ) ( b − a ) , the first inequality because h h h is nondecreasing. As η > 0 \eta>0 η > 0 was arbitrary, 0 ≤ h ( b ) − h ( a ) ≤ L ( b − a ) 0\le h(b)-h(a)\le L(b-a) 0 ≤ h ( b ) − h ( a ) ≤ L ( b − a ) , which gives the claim.
Step B (bounds on couplings). Let ν ∈ Σ d , r \nu\in\Sigma_{d,r} ν ∈ Σ d , r , let γ ∈ C ( κ d ( ν ) ) \gamma\in\mathcal{C}(\kappa_{d}(\nu)) γ ∈ C ( κ d ( ν )) and let ( X , P , X ′ ) (X,P,X') ( X , P , X ′ ) be a realisation of γ \gamma γ . We show
l a w ( X , P ) = κ 2 d ( π ) , l a w ( X ) = κ d ( μ ) , ∥ P ∥ 2 = a π , ∥ X ∥ 2 ≤ r d , ∥ X ′ ∥ 2 ≤ r d , \mathrm{law}(X,P)=\kappa_{2d}(\pi),\qquad\mathrm{law}(X)=\kappa_{d}(\mu),\qquad\lVert P\rVert_{2}=a_{\pi},\qquad\lVert X\rVert_{2}\le r\sqrt{d},\qquad\lVert X'\rVert_{2}\le r\sqrt{d}, law ( X , P ) = κ 2 d ( π ) , law ( X ) = κ d ( μ ) , ∥ P ∥ 2 = a π , ∥ X ∥ 2 ≤ r d , ∥ X ′ ∥ 2 ≤ r d ,
and consequently s ( γ ) ≤ 2 r d s(\gamma)\le2r\sqrt{d} s ( γ ) ≤ 2 r d and p ( γ ) ≤ a π s ( γ ) p(\gamma)\le a_{\pi}\,s(\gamma) p ( γ ) ≤ a π s ( γ ) . Since B ( X , P , X ′ ) = ( X , P ) B(X,P,X')=(X,P) B ( X , P , X ′ ) = ( X , P ) and C ( X , P , X ′ ) = X ′ C(X,P,X')=X' C ( X , P , X ′ ) = X ′ , Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §push-forward and Couplings of a Square-Integrable Plan with a Law: Displacement and Momentum Pairing §couplings give l a w ( X , P ) = B # γ = κ 2 d ( π ) \mathrm{law}(X,P)=B_{\#}\gamma=\kappa_{2d}(\pi) law ( X , P ) = B # γ = κ 2 d ( π ) and l a w ( X ′ ) = C # γ = κ d ( ν ) \mathrm{law}(X')=C_{\#}\gamma=\kappa_{d}(\nu) law ( X ′ ) = C # γ = κ d ( ν ) . By Every Square-Integrable Noncommutative Law is the Law of a Square-Integrable Tuple; Realisation of Couplings and of Almost Optimal Pairs §coupling , l a w ( X ) = p r # 1 κ 2 d ( π ) \mathrm{law}(X)=\mathrm{pr}^{1}_{\#}\kappa_{2d}(\pi) law ( X ) = pr # 1 κ 2 d ( π ) and l a w ( P ) = p r # 2 κ 2 d ( π ) \mathrm{law}(P)=\mathrm{pr}^{2}_{\#}\kappa_{2d}(\pi) law ( P ) = pr # 2 κ 2 d ( π ) . By Calculus of Square-Integrable Noncommutative Laws: Agreement on Bounded Laws, Lipschitz Estimates, Functoriality of Push-Forwards, Moment Formulas, Positivity, the Cost and the Diagonal Coupling §bounded and Affine Substitutions of Noncommutative Laws: Self-Adjointness, Composition, Moment Formulas and Positivity, and the Coordinate Data §coordinate , these equal κ d ( π ∘ ι 1 ) \kappa_{d}(\pi\circ\iota^{1}) κ d ( π ∘ ι 1 ) and κ d ( π ∘ ι 2 ) \kappa_{d}(\pi\circ\iota^{2}) κ d ( π ∘ ι 2 ) (here π ∈ Σ 2 d \pi\in\Sigma_{2d} π ∈ Σ 2 d by Marginal Isometries, Bounded Plans and Displacement Pairings for Noncommutative Laws §plans , so π ∘ ι 1 , π ∘ ι 2 ∈ Σ d \pi\circ\iota^{1},\pi\circ\iota^{2}\in\Sigma_{d} π ∘ ι 1 , π ∘ ι 2 ∈ Σ d by the preamble of Marginal Isometries, Bounded Plans and Displacement Pairings for Noncommutative Laws ), and π ∘ ι 1 = μ \pi\circ\iota^{1}=\mu π ∘ ι 1 = μ because π \pi π is a bounded plan at μ \mu μ in the sense of Marginal Isometries, Bounded Plans and Displacement Pairings for Noncommutative Laws §plans . By Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §moments and Calculus of Square-Integrable Noncommutative Laws: Agreement on Bounded Laws, Lipschitz Estimates, Functoriality of Push-Forwards, Moment Formulas, Positivity, the Cost and the Diagonal Coupling §bounded , ∥ X ∥ 2 2 = M ( μ ) \lVert X\rVert_{2}^{2}=M(\mu) ∥ X ∥ 2 2 = M ( μ ) , ∥ X ′ ∥ 2 2 = M ( ν ) \lVert X'\rVert_{2}^{2}=M(\nu) ∥ X ′ ∥ 2 2 = M ( ν ) and ∥ P ∥ 2 2 = M ( π ∘ ι 2 ) \lVert P\rVert_{2}^{2}=M(\pi\circ\iota^{2}) ∥ P ∥ 2 2 = M ( π ∘ ι 2 ) . Since μ , ν ∈ Σ d , r \mu,\nu\in\Sigma_{d,r} μ , ν ∈ Σ d , r , each of the real numbers μ ( x j x j ) \mu(x_{j}x_{j}) μ ( x j x j ) and ν ( x j x j ) \nu(x_{j}x_{j}) ν ( x j x j ) has modulus at most r 2 r^{2} r 2 by Noncommutative Laws of Finitely Many Self-Adjoint Variables with a Norm Bound §norm-bound (words of length 2 2 2 ), so M ( μ ) ≤ d r 2 M(\mu)\le dr^{2} M ( μ ) ≤ d r 2 and M ( ν ) ≤ d r 2 M(\nu)\le dr^{2} M ( ν ) ≤ d r 2 . Since ι 2 \iota^{2} ι 2 substitutes x d + j x_{d+j} x d + j for x j x_{j} x j by Couplings of Two Noncommutative Laws and Their Quadratic Cost §marginals and is multiplicative by Substitution is the Unique Unital Homomorphism with Prescribed Values on the Variables: Monomials, Products, Adjoints and Composition §homomorphism , M ( π ∘ ι 2 ) = ∑ j = 1 d π ( x d + j x d + j ) = a π 2 M(\pi\circ\iota^{2})=\sum_{j=1}^{d}\pi(x_{d+j}x_{d+j})=a_{\pi}^{2} M ( π ∘ ι 2 ) = ∑ j = 1 d π ( x d + j x d + j ) = a π 2 by Marginal Isometries, Bounded Plans and Displacement Pairings for Noncommutative Laws §plans . Finally s ( γ ) = ∥ X ′ − X ∥ 2 ≤ ∥ X ′ ∥ 2 + ∥ X ∥ 2 ≤ 2 r d s(\gamma)=\lVert X'-X\rVert_{2}\le\lVert X'\rVert_{2}+\lVert X\rVert_{2}\le2r\sqrt{d} s ( γ ) = ∥ X ′ − X ∥ 2 ≤ ∥ X ′ ∥ 2 + ∥ X ∥ 2 ≤ 2 r d and p ( γ ) = ⟨ P , X ′ − X ⟩ 2 ≤ ∥ P ∥ 2 s ( γ ) = a π s ( γ ) p(\gamma)=\langle P,X'-X\rangle_{2}\le\lVert P\rVert_{2}\,s(\gamma)=a_{\pi}\,s(\gamma) p ( γ ) = ⟨ P , X ′ − X ⟩ 2 ≤ ∥ P ∥ 2 s ( γ ) = a π s ( γ ) .
Step C (law invariance). Let k , n ∈ N k,n\in\mathbb{N} k , n ∈ N , let Z Z Z and Z 1 Z_{1} Z 1 be L 2 L^{2} L 2 k k k -tuples of tracial W*-probability spaces (possibly different) with l a w ( Z ) = l a w ( Z 1 ) \mathrm{law}(Z)=\mathrm{law}(Z_{1}) law ( Z ) = law ( Z 1 ) , and let T = ( E , 0 ) T=(E,0) T = ( E , 0 ) be an affine datum from k k k to n n n variables. By Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §push-forward , l a w ( T Z ) = T # l a w ( Z ) = T # l a w ( Z 1 ) = l a w ( T Z 1 ) \mathrm{law}(TZ)=T_{\#}\mathrm{law}(Z)=T_{\#}\mathrm{law}(Z_{1})=\mathrm{law}(TZ_{1}) law ( TZ ) = T # law ( Z ) = T # law ( Z 1 ) = law ( T Z 1 ) . By Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §moments , ∥ T Z ∥ 2 2 \lVert TZ\rVert_{2}^{2} ∥ TZ ∥ 2 2 and ∥ T Z 1 ∥ 2 2 \lVert TZ_{1}\rVert_{2}^{2} ∥ T Z 1 ∥ 2 2 both equal the second moment of this law, so ∥ T Z ∥ 2 = ∥ T Z 1 ∥ 2 \lVert TZ\rVert_{2}=\lVert TZ_{1}\rVert_{2} ∥ TZ ∥ 2 = ∥ T Z 1 ∥ 2 ; and if n = 2 d n=2d n = 2 d and T Z = ( A , A ′ ) TZ=(A,A') TZ = ( A , A ′ ) , T Z 1 = ( A 1 , A 1 ′ ) TZ_{1}=(A_{1},A_{1}') T Z 1 = ( A 1 , A 1 ′ ) with L 2 L^{2} L 2 d d d -tuples A , A ′ , A 1 , A 1 ′ A,A',A_{1},A_{1}' A , A ′ , A 1 , A 1 ′ , then by the same clause and Sums, Real Multiples and the Pairing of Square-Integrable Tuples in a Tracial W*-Probability Space §pairing both ⟨ A , A ′ ⟩ 2 \langle A,A'\rangle_{2} ⟨ A , A ′ ⟩ 2 and ⟨ A 1 , A 1 ′ ⟩ 2 \langle A_{1},A_{1}'\rangle_{2} ⟨ A 1 , A 1 ′ ⟩ 2 equal ∑ i = 1 d m i , d + i ( l a w ( T Z ) ) \sum_{i=1}^{d}\mathrm{m}_{i,d+i}(\mathrm{law}(TZ)) ∑ i = 1 d m i , d + i ( law ( TZ )) , so they are equal. By Square-Integrable Tuples in a Tracial W*-Probability Space: Their Norm, Affine Images, Pairs, Embedded Images and Laws §operations , every tuple formed below from the coordinates of a tuple Z Z Z by selecting or reordering blocks and taking real linear combinations of blocks is T Z TZ TZ for the affine datum T = ( E , 0 ) T=(E,0) T = ( E , 0 ) whose coefficients are read off from the formula; pairs and triples are tuples by the same clause. We use this without writing out the matrices E E E .
Step D (gluing). Let k , m , n ∈ N k,m,n\in\mathbb{N} k , m , n ∈ N , let Z , W Z,W Z , W be L 2 L^{2} L 2 tuples of lengths k , m k,m k , m of one tracial W*-probability space and Z 1 , V Z_{1},V Z 1 , V L 2 L^{2} L 2 tuples of lengths k , n k,n k , n of another, with l a w ( Z ) = l a w ( Z 1 ) \mathrm{law}(Z)=\mathrm{law}(Z_{1}) law ( Z ) = law ( Z 1 ) . With F m , F n F^{m},F^{n} F m , F n as in Gluing Two Square-Integrable Noncommutative Laws along a Common Marginal , F m ( Z , W ) = Z F^{m}(Z,W)=Z F m ( Z , W ) = Z and F n ( Z 1 , V ) = Z 1 F^{n}(Z_{1},V)=Z_{1} F n ( Z 1 , V ) = Z 1 , so F # m l a w ( Z , W ) = l a w ( Z ) = l a w ( Z 1 ) = F # n l a w ( Z 1 , V ) F^{m}_{\#}\mathrm{law}(Z,W)=\mathrm{law}(Z)=\mathrm{law}(Z_{1})=F^{n}_{\#}\mathrm{law}(Z_{1},V) F # m law ( Z , W ) = law ( Z ) = law ( Z 1 ) = F # n law ( Z 1 , V ) by Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §push-forward . By Gluing Two Square-Integrable Noncommutative Laws along a Common Marginal §glue there are a tracial W*-probability space and L 2 L^{2} L 2 tuples Z ^ , W ^ , V ^ \hat{Z},\hat{W},\hat{V} Z ^ , W ^ , V ^ of it, of lengths k , m , n k,m,n k , m , n , with l a w ( Z ^ , W ^ ) = l a w ( Z , W ) \mathrm{law}(\hat{Z},\hat{W})=\mathrm{law}(Z,W) law ( Z ^ , W ^ ) = law ( Z , W ) and l a w ( Z ^ , V ^ ) = l a w ( Z 1 , V ) \mathrm{law}(\hat{Z},\hat{V})=\mathrm{law}(Z_{1},V) law ( Z ^ , V ^ ) = law ( Z 1 , V ) .
Step E (transfer). Let ν , ν ′ ∈ Σ d , r \nu,\nu'\in\Sigma_{d,r} ν , ν ′ ∈ Σ d , r , put w = W 2 ( ν , ν ′ ) w=W_{2}(\nu,\nu') w = W 2 ( ν , ν ′ ) , and let γ ∈ C ( κ d ( ν ) ) \gamma\in\mathcal{C}(\kappa_{d}(\nu)) γ ∈ C ( κ d ( ν )) . We show that there is γ ′ ′ ∈ C ( κ d ( ν ′ ) ) \gamma''\in\mathcal{C}(\kappa_{d}(\nu')) γ ′′ ∈ C ( κ d ( ν ′ )) with ∣ s ( γ ′ ′ ) − s ( γ ) ∣ ≤ w |s(\gamma'')-s(\gamma)|\le w ∣ s ( γ ′′ ) − s ( γ ) ∣ ≤ w and Ψ ( γ ′ ′ ) ≥ Ψ ( γ ) − K w \Psi(\gamma'')\ge\Psi(\gamma)-Kw Ψ ( γ ′′ ) ≥ Ψ ( γ ) − K w . Let ( X , P , X ′ ) (X,P,X') ( X , P , X ′ ) be a realisation of γ \gamma γ . By The Noncommutative Wasserstein Distance: Existence of Optimal Couplings, Symmetry, Separation, a Moment Bound, Weak-Star Lower Semicontinuity, and Displacement Interpolation §attained (with R = r R=r R = r ) there is an optimal coupling ϱ ∈ Π ( ν , ν ′ ) \varrho\in\Pi(\nu,\nu') ϱ ∈ Π ( ν , ν ′ ) , so ϱ ∈ Σ 2 d \varrho\in\Sigma_{2d} ϱ ∈ Σ 2 d by Marginal Isometries, Bounded Plans and Displacement Pairings for Noncommutative Laws §bounded-couplings and I ( ϱ ) = w 2 I(\varrho)=w^{2} I ( ϱ ) = w 2 by The Noncommutative Quadratic Wasserstein Distance and Optimal Couplings §optimal ; by Calculus of Square-Integrable Noncommutative Laws: Agreement on Bounded Laws, Lipschitz Estimates, Functoriality of Push-Forwards, Moment Formulas, Positivity, the Cost and the Diagonal Coupling §bounded , κ 2 d ( ϱ ) ∈ Π 2 ( κ d ( ν ) , κ d ( ν ′ ) ) \kappa_{2d}(\varrho)\in\Pi^{2}(\kappa_{d}(\nu),\kappa_{d}(\nu')) κ 2 d ( ϱ ) ∈ Π 2 ( κ d ( ν ) , κ d ( ν ′ )) and I ( κ 2 d ( ϱ ) ) = w 2 \mathcal{I}(\kappa_{2d}(\varrho))=w^{2} I ( κ 2 d ( ϱ )) = w 2 . By Every Square-Integrable Noncommutative Law is the Law of a Square-Integrable Tuple; Realisation of Couplings and of Almost Optimal Pairs §coupling and Square-Integrable Noncommutative Laws: the Wasserstein Completion of the Laws, Affine Push-Forwards, Moments, Couplings and Cost §couplings there are L 2 L^{2} L 2 d d d -tuples Y , Y ′ ′ Y,Y'' Y , Y ′′ of a tracial W*-probability space with l a w ( Y ) = κ d ( ν ) \mathrm{law}(Y)=\kappa_{d}(\nu) law ( Y ) = κ d ( ν ) , l a w ( Y ′ ′ ) = κ d ( ν ′ ) \mathrm{law}(Y'')=\kappa_{d}(\nu') law ( Y ′′ ) = κ d ( ν ′ ) and ∥ Y − Y ′ ′ ∥ 2 = w \lVert Y-Y''\rVert_{2}=w ∥ Y − Y ′′ ∥ 2 = w . Since l a w ( X ′ ) = κ d ( ν ) \mathrm{law}(X')=\kappa_{d}(\nu) law ( X ′ ) = κ d ( ν ) by Step B, Step D with Z = X ′ Z=X' Z = X ′ , W = ( X , P ) W=(X,P) W = ( X , P ) , Z 1 = Y Z_{1}=Y Z 1 = Y and V = Y ′ ′ V=Y'' V = Y ′′ gives L 2 L^{2} L 2 d d d -tuples X ^ ′ , X ^ , P ^ , Y ^ ′ ′ \hat{X}',\hat{X},\hat{P},\hat{Y}'' X ^ ′ , X ^ , P ^ , Y ^ ′′ of one tracial W*-probability space with l a w ( X ^ ′ , ( X ^ , P ^ ) ) = l a w ( X ′ , ( X , P ) ) \mathrm{law}(\hat{X}',(\hat{X},\hat{P}))=\mathrm{law}(X',(X,P)) law ( X ^ ′ , ( X ^ , P ^ )) = law ( X ′ , ( X , P )) and l a w ( X ^ ′ , Y ^ ′ ′ ) = l a w ( Y , Y ′ ′ ) \mathrm{law}(\hat{X}',\hat{Y}'')=\mathrm{law}(Y,Y'') law ( X ^ ′ , Y ^ ′′ ) = law ( Y , Y ′′ ) . By Step C, l a w ( X ^ , P ^ , X ^ ′ ) = l a w ( X , P , X ′ ) = γ \mathrm{law}(\hat{X},\hat{P},\hat{X}')=\mathrm{law}(X,P,X')=\gamma law ( X ^ , P ^ , X ^ ′ ) = law ( X , P , X ′ ) = γ , l a w ( Y ^ ′ ′ ) = l a w ( Y ′ ′ ) = κ d ( ν ′ ) \mathrm{law}(\hat{Y}'')=\mathrm{law}(Y'')=\kappa_{d}(\nu') law ( Y ^ ′′ ) = law ( Y ′′ ) = κ d ( ν ′ ) and ∥ Y ^ ′ ′ − X ^ ′ ∥ 2 = ∥ Y ′ ′ − Y ∥ 2 = w \lVert\hat{Y}''-\hat{X}'\rVert_{2}=\lVert Y''-Y\rVert_{2}=w ∥ Y ^ ′′ − X ^ ′ ∥ 2 = ∥ Y ′′ − Y ∥ 2 = w . Put γ ′ ′ = l a w ( X ^ , P ^ , Y ^ ′ ′ ) \gamma''=\mathrm{law}(\hat{X},\hat{P},\hat{Y}'') γ ′′ = law ( X ^ , P ^ , Y ^ ′′ ) . By Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §push-forward , B # γ ′ ′ = l a w ( X ^ , P ^ ) = B # γ = κ 2 d ( π ) B_{\#}\gamma''=\mathrm{law}(\hat{X},\hat{P})=B_{\#}\gamma=\kappa_{2d}(\pi) B # γ ′′ = law ( X ^ , P ^ ) = B # γ = κ 2 d ( π ) and C # γ ′ ′ = l a w ( Y ^ ′ ′ ) = κ d ( ν ′ ) C_{\#}\gamma''=\mathrm{law}(\hat{Y}'')=\kappa_{d}(\nu') C # γ ′′ = law ( Y ^ ′′ ) = κ d ( ν ′ ) , so γ ′ ′ ∈ C ( κ d ( ν ′ ) ) \gamma''\in\mathcal{C}(\kappa_{d}(\nu')) γ ′′ ∈ C ( κ d ( ν ′ )) by Couplings of a Square-Integrable Plan with a Law: Displacement and Momentum Pairing §couplings .
Put a ^ = X ^ ′ − X ^ \hat{a}=\hat{X}'-\hat{X} a ^ = X ^ ′ − X ^ and v = Y ^ ′ ′ − X ^ ′ v=\hat{Y}''-\hat{X}' v = Y ^ ′′ − X ^ ′ , so that Y ^ ′ ′ − X ^ = a ^ + v \hat{Y}''-\hat{X}=\hat{a}+v Y ^ ′′ − X ^ = a ^ + v and ∥ v ∥ 2 = w \lVert v\rVert_{2}=w ∥ v ∥ 2 = w . Computing from the realisations ( X ^ , P ^ , X ^ ′ ) (\hat{X},\hat{P},\hat{X}') ( X ^ , P ^ , X ^ ′ ) of γ \gamma γ and ( X ^ , P ^ , Y ^ ′ ′ ) (\hat{X},\hat{P},\hat{Y}'') ( X ^ , P ^ , Y ^ ′′ ) of γ ′ ′ \gamma'' γ ′′ , we get s ( γ ) = ∥ a ^ ∥ 2 s(\gamma)=\lVert\hat{a}\rVert_{2} s ( γ ) = ∥ a ^ ∥ 2 , s ( γ ′ ′ ) = ∥ a ^ + v ∥ 2 s(\gamma'')=\lVert\hat{a}+v\rVert_{2} s ( γ ′′ ) = ∥ a ^ + v ∥ 2 , hence ∣ s ( γ ′ ′ ) − s ( γ ) ∣ ≤ w |s(\gamma'')-s(\gamma)|\le w ∣ s ( γ ′′ ) − s ( γ ) ∣ ≤ w , and p ( γ ′ ′ ) − p ( γ ) = ⟨ P ^ , v ⟩ 2 ≥ − ∥ P ^ ∥ 2 w = − a π w p(\gamma'')-p(\gamma)=\langle\hat{P},v\rangle_{2}\ge-\lVert\hat{P}\rVert_{2}\,w=-a_{\pi}w p ( γ ′′ ) − p ( γ ) = ⟨ P ^ , v ⟩ 2 ≥ − ∥ P ^ ∥ 2 w = − a π w by Step B. By Step A, h ( s ( γ ′ ′ ) ) − h ( s ( γ ) ) ≤ L w h(s(\gamma''))-h(s(\gamma))\le Lw h ( s ( γ ′′ )) − h ( s ( γ )) ≤ L w . By Step B, s ( γ ) + s ( γ ′ ′ ) ≤ 4 r d s(\gamma)+s(\gamma'')\le4r\sqrt{d} s ( γ ) + s ( γ ′′ ) ≤ 4 r d , so θ s ( γ ′ ′ ) 2 − θ s ( γ ) 2 = θ ( s ( γ ′ ′ ) − s ( γ ) ) ( s ( γ ′ ′ ) + s ( γ ) ) ≤ 4 θ r d w \theta s(\gamma'')^{2}-\theta s(\gamma)^{2}=\theta(s(\gamma'')-s(\gamma))(s(\gamma'')+s(\gamma))\le4\theta r\sqrt{d}\,w θ s ( γ ′′ ) 2 − θ s ( γ ) 2 = θ ( s ( γ ′′ ) − s ( γ )) ( s ( γ ′′ ) + s ( γ )) ≤ 4 θ r d w . Adding, Ψ ( γ ′ ′ ) ≥ Ψ ( γ ) − ( a π + L + 4 θ r d ) w = Ψ ( γ ) − K w \Psi(\gamma'')\ge\Psi(\gamma)-(a_{\pi}+L+4\theta r\sqrt{d})w=\Psi(\gamma)-Kw Ψ ( γ ′′ ) ≥ Ψ ( γ ) − ( a π + L + 4 θ r d ) w = Ψ ( γ ) − K w .
Proof of clause 1. Fix ν ∈ Σ d , r \nu\in\Sigma_{d,r} ν ∈ Σ d , r . Since π ∈ Σ 2 d \pi\in\Sigma_{2d} π ∈ Σ 2 d by Marginal Isometries, Bounded Plans and Displacement Pairings for Noncommutative Laws §plans , there is a real R π > 0 R_{\pi}>0 R π > 0 with π ∈ Σ 2 d , R π \pi\in\Sigma_{2d,R_{\pi}} π ∈ Σ 2 d , R π by Noncommutative Laws of Finitely Many Self-Adjoint Variables with a Norm Bound §law ; put R ′ = max ( R π , r ) R'=\max(R_{\pi},r) R ′ = max ( R π , r ) .
(a) Every γ ∈ C ( κ d ( ν ) ) \gamma\in\mathcal{C}(\kappa_{d}(\nu)) γ ∈ C ( κ d ( ν )) equals κ 3 d ( λ ) \kappa_{3d}(\lambda) κ 3 d ( λ ) for some λ ∈ Σ 3 d , R ′ \lambda\in\Sigma_{3d,R'} λ ∈ Σ 3 d , R ′ . Indeed, let ( X , P , X ′ ) (X,P,X') ( X , P , X ′ ) be a realisation of γ \gamma γ in ( H , M , Ω ) (H,M,\Omega) ( H , M , Ω ) . By Step B, l a w ( X , P ) = κ 2 d ( π ) \mathrm{law}(X,P)=\kappa_{2d}(\pi) law ( X , P ) = κ 2 d ( π ) and l a w ( X ′ ) = κ d ( ν ) \mathrm{law}(X')=\kappa_{d}(\nu) law ( X ′ ) = κ d ( ν ) , so by Square-Integrable Tuples with a Bounded Law are Vacuum Tuples of Bounded Self-Adjoint Operators §operator and Square-Integrable Tuples with a Bounded Law are Vacuum Tuples of Bounded Self-Adjoint Operators §bound (for 2 d 2d 2 d and d d d variables) there are a self-adjoint 2 d 2d 2 d -tuple u u u and a self-adjoint d d d -tuple u ′ u' u ′ in M M M with u Ω = ( X , P ) u\Omega=(X,P) u Ω = ( X , P ) , u ′ Ω = X ′ u'\Omega=X' u ′ Ω = X ′ , and all operator norms of their entries at most R ′ R' R ′ . The concatenation ( u , u ′ ) (u,u') ( u , u ′ ) is a self-adjoint 3 d 3d 3 d -tuple in M M M in the sense of Self-Adjoint Tuples in a Tracial W*-Probability Space and Their Laws §tuple , and its vacuum tuple is ( X , P , X ′ ) (X,P,X') ( X , P , X ′ ) by Square-Integrable Tuples in a Tracial W*-Probability Space: Their Norm, Affine Images, Pairs, Embedded Images and Laws §operations . By Laws of Self-Adjoint Tuples in a Tracial W*-Probability Space: Moments, Affine Images, Couplings, Embeddings and L^2 Approximation §law , λ = λ ( u , u ′ ) ∈ Σ 3 d , R ′ \lambda=\lambda_{(u,u')}\in\Sigma_{3d,R'} λ = λ ( u , u ′ ) ∈ Σ 3 d , R ′ , and by Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §bounded , γ = l a w ( X , P , X ′ ) = κ 3 d ( λ ) \gamma=\mathrm{law}(X,P,X')=\kappa_{3d}(\lambda) γ = law ( X , P , X ′ ) = κ 3 d ( λ ) .
(b) Let σ B \sigma_{B} σ B , σ C \sigma_{C} σ C and σ U \sigma_{U} σ U be the affine substitutions of the affine data B B B , C C C and U U U of Couplings of a Square-Integrable Plan with a Law: Displacement and Momentum Pairing . For λ ∈ Σ 3 d \lambda\in\Sigma_{3d} λ ∈ Σ 3 d we have λ ∘ σ B ∈ Σ 2 d \lambda\circ\sigma_{B}\in\Sigma_{2d} λ ∘ σ B ∈ Σ 2 d and λ ∘ σ C ∈ Σ d \lambda\circ\sigma_{C}\in\Sigma_{d} λ ∘ σ C ∈ Σ d by Affine Substitutions of Noncommutative Laws: Self-Adjointness, Composition, Moment Formulas and Positivity, and the Coordinate Data §self-adjoint , and κ 3 d ( λ ) ∈ C ( κ d ( ν ) ) \kappa_{3d}(\lambda)\in\mathcal{C}(\kappa_{d}(\nu)) κ 3 d ( λ ) ∈ C ( κ d ( ν )) if and only if λ ∘ σ B = π \lambda\circ\sigma_{B}=\pi λ ∘ σ B = π and λ ∘ σ C = ν \lambda\circ\sigma_{C}=\nu λ ∘ σ C = ν . Indeed B # κ 3 d ( λ ) = κ 2 d ( λ ∘ σ B ) B_{\#}\kappa_{3d}(\lambda)=\kappa_{2d}(\lambda\circ\sigma_{B}) B # κ 3 d ( λ ) = κ 2 d ( λ ∘ σ B ) and C # κ 3 d ( λ ) = κ d ( λ ∘ σ C ) C_{\#}\kappa_{3d}(\lambda)=\kappa_{d}(\lambda\circ\sigma_{C}) C # κ 3 d ( λ ) = κ d ( λ ∘ σ C ) by Calculus of Square-Integrable Noncommutative Laws: Agreement on Bounded Laws, Lipschitz Estimates, Functoriality of Push-Forwards, Moment Formulas, Positivity, the Cost and the Diagonal Coupling §bounded , and the canonical maps κ 2 d , κ d \kappa_{2d},\kappa_{d} κ 2 d , κ d of the completions of Square-Integrable Noncommutative Laws: the Wasserstein Completion of the Laws, Affine Push-Forwards, Moments, Couplings and Cost §laws are injective by The Metric Completion is a Complete Metric Space with a Dense Isometric Copy of the Space, and Maps Preserving Cauchy Sequences Extend to It §isometry ; now apply Couplings of a Square-Integrable Plan with a Law: Displacement and Momentum Pairing §couplings .
(c) For λ ∈ Σ 3 d \lambda\in\Sigma_{3d} λ ∈ Σ 3 d and i , j ∈ [ 2 d ] i,j\in[2d] i , j ∈ [ 2 d ] put c i j ( λ ) = λ ( σ U ( x i x j ) ) c_{ij}(\lambda)=\lambda(\sigma_{U}(x_{i}x_{j})) c ij ( λ ) = λ ( σ U ( x i x j )) . Since λ ∘ σ U \lambda\circ\sigma_{U} λ ∘ σ U is a tracial state by Affine Substitutions of Noncommutative Laws: Self-Adjointness, Composition, Moment Formulas and Positivity, and the Coordinate Data §self-adjoint , c i j ( λ ) = m i j ( λ ∘ σ U ) c_{ij}(\lambda)=\mathrm{m}_{ij}(\lambda\circ\sigma_{U}) c ij ( λ ) = m ij ( λ ∘ σ U ) is real by Affine Substitutions of Noncommutative Laws: Self-Adjointness, Composition, Moment Formulas and Positivity, and the Coordinate Data §moments , and m i j ( U # κ 3 d ( λ ) ) = m i j ( κ 2 d ( λ ∘ σ U ) ) = c i j ( λ ) \mathrm{m}_{ij}(U_{\#}\kappa_{3d}(\lambda))=\mathrm{m}_{ij}(\kappa_{2d}(\lambda\circ\sigma_{U}))=c_{ij}(\lambda) m ij ( U # κ 3 d ( λ )) = m ij ( κ 2 d ( λ ∘ σ U )) = c ij ( λ ) by Calculus of Square-Integrable Noncommutative Laws: Agreement on Bounded Laws, Lipschitz Estimates, Functoriality of Push-Forwards, Moment Formulas, Positivity, the Cost and the Diagonal Coupling §bounded . Hence, by the formulas of Couplings of a Square-Integrable Plan with a Law: Displacement and Momentum Pairing §displacement-pairing ,
s ( κ 3 d ( λ ) ) 2 = ∑ i = 1 d c d + i , d + i ( λ ) , p ( κ 3 d ( λ ) ) = ∑ i = 1 d c i , d + i ( λ ) . s(\kappa_{3d}(\lambda))^{2}=\sum_{i=1}^{d}c_{d+i,d+i}(\lambda),\qquad p(\kappa_{3d}(\lambda))=\sum_{i=1}^{d}c_{i,d+i}(\lambda). s ( κ 3 d ( λ ) ) 2 = i = 1 ∑ d c d + i , d + i ( λ ) , p ( κ 3 d ( λ )) = i = 1 ∑ d c i , d + i ( λ ) .
Now choose, for every n ∈ N n\in\mathbb{N} n ∈ N , some γ n ∈ C ( κ d ( ν ) ) \gamma_{n}\in\mathcal{C}(\kappa_{d}(\nu)) γ n ∈ C ( κ d ( ν )) with Ψ ( γ n ) > Φ ( κ d ( ν ) ) − 2 − n \Psi(\gamma_{n})>\Phi(\kappa_{d}(\nu))-2^{-n} Ψ ( γ n ) > Φ ( κ d ( ν )) − 2 − n (Notation), and by (a) some λ n ∈ Σ 3 d , R ′ \lambda_{n}\in\Sigma_{3d,R'} λ n ∈ Σ 3 d , R ′ with γ n = κ 3 d ( λ n ) \gamma_{n}=\kappa_{3d}(\lambda_{n}) γ n = κ 3 d ( λ n ) . By Sequential Weak-Star Compactness of the Noncommutative Laws with a Given Norm Bound §compact there are a subsequence ( λ n k ) k ∈ N (\lambda_{n_{k}})_{k\in\mathbb{N}} ( λ n k ) k ∈ N in the sense of Subsequence of a Sequence in a Set and λ ∈ Σ 3 d , R ′ \lambda\in\Sigma_{3d,R'} λ ∈ Σ 3 d , R ′ with λ n k → λ \lambda_{n_{k}}\to\lambda λ n k → λ weak-star; by Weak-Star Convergence of Noncommutative Laws §weak-star , for every q ∈ P 3 d q\in\mathcal{P}_{3d} q ∈ P 3 d the real and imaginary parts of λ n k ( q ) \lambda_{n_{k}}(q) λ n k ( q ) converge to those of λ ( q ) \lambda(q) λ ( q ) . For q ′ ∈ P 2 d q'\in\mathcal{P}_{2d} q ′ ∈ P 2 d , (b) gives λ n k ( σ B ( q ′ ) ) = π ( q ′ ) \lambda_{n_{k}}(\sigma_{B}(q'))=\pi(q') λ n k ( σ B ( q ′ )) = π ( q ′ ) for every k k k , so λ ( σ B ( q ′ ) ) = π ( q ′ ) \lambda(\sigma_{B}(q'))=\pi(q') λ ( σ B ( q ′ )) = π ( q ′ ) ; thus λ ∘ σ B = π \lambda\circ\sigma_{B}=\pi λ ∘ σ B = π , and likewise λ ∘ σ C = ν \lambda\circ\sigma_{C}=\nu λ ∘ σ C = ν . By (b), γ ∗ = κ 3 d ( λ ) ∈ C ( κ d ( ν ) ) \gamma^{*}=\kappa_{3d}(\lambda)\in\mathcal{C}(\kappa_{d}(\nu)) γ ∗ = κ 3 d ( λ ) ∈ C ( κ d ( ν )) . The real numbers c i j ( λ n k ) c_{ij}(\lambda_{n_{k}}) c ij ( λ n k ) converge to c i j ( λ ) c_{ij}(\lambda) c ij ( λ ) , so by (c) s ( γ n k ) 2 → s ( γ ∗ ) 2 s(\gamma_{n_{k}})^{2}\to s(\gamma^{*})^{2} s ( γ n k ) 2 → s ( γ ∗ ) 2 and p ( γ n k ) → p ( γ ∗ ) p(\gamma_{n_{k}})\to p(\gamma^{*}) p ( γ n k ) → p ( γ ∗ ) . Since ∣ x − y ∣ 2 ≤ ∣ x − y ∣ ( x + y ) = ∣ x 2 − y 2 ∣ |x-y|^{2}\le|x-y|(x+y)=|x^{2}-y^{2}| ∣ x − y ∣ 2 ≤ ∣ x − y ∣ ( x + y ) = ∣ x 2 − y 2 ∣ for reals x , y ≥ 0 x,y\ge0 x , y ≥ 0 , also s ( γ n k ) → s ( γ ∗ ) s(\gamma_{n_{k}})\to s(\gamma^{*}) s ( γ n k ) → s ( γ ∗ ) , and by Step A ∣ h ( s ( γ n k ) ) − h ( s ( γ ∗ ) ) ∣ ≤ L ∣ s ( γ n k ) − s ( γ ∗ ) ∣ → 0 |h(s(\gamma_{n_{k}}))-h(s(\gamma^{*}))|\le L|s(\gamma_{n_{k}})-s(\gamma^{*})|\to0 ∣ h ( s ( γ n k )) − h ( s ( γ ∗ )) ∣ ≤ L ∣ s ( γ n k ) − s ( γ ∗ ) ∣ → 0 . Hence Ψ ( γ n k ) → Ψ ( γ ∗ ) \Psi(\gamma_{n_{k}})\to\Psi(\gamma^{*}) Ψ ( γ n k ) → Ψ ( γ ∗ ) by the limit rules of The Real Numbers: Standing Notation and Background §sequences . Since n k ≥ k n_{k}\ge k n k ≥ k , we have Φ ( κ d ( ν ) ) − 2 − k < Ψ ( γ n k ) ≤ Φ ( κ d ( ν ) ) \Phi(\kappa_{d}(\nu))-2^{-k}<\Psi(\gamma_{n_{k}})\le\Phi(\kappa_{d}(\nu)) Φ ( κ d ( ν )) − 2 − k < Ψ ( γ n k ) ≤ Φ ( κ d ( ν )) for every k k k , so Ψ ( γ ∗ ) = Φ ( κ d ( ν ) ) \Psi(\gamma^{*})=\Phi(\kappa_{d}(\nu)) Ψ ( γ ∗ ) = Φ ( κ d ( ν )) and γ ∗ \gamma^{*} γ ∗ is a maximising coupling for ν \nu ν .
Proof of clause 2. We show ∣ Φ ( κ d ( ν ) ) − Φ ( κ d ( ν ′ ) ) ∣ ≤ K W 2 ( ν , ν ′ ) |\Phi(\kappa_{d}(\nu))-\Phi(\kappa_{d}(\nu'))|\le K\,W_{2}(\nu,\nu') ∣Φ ( κ d ( ν )) − Φ ( κ d ( ν ′ )) ∣ ≤ K W 2 ( ν , ν ′ ) for all ν , ν ′ ∈ Σ d , r \nu,\nu'\in\Sigma_{d,r} ν , ν ′ ∈ Σ d , r . Let γ \gamma γ be a maximising coupling for ν \nu ν , which exists by clause 1. Step E gives γ ′ ′ ∈ C ( κ d ( ν ′ ) ) \gamma''\in\mathcal{C}(\kappa_{d}(\nu')) γ ′′ ∈ C ( κ d ( ν ′ )) with Φ ( κ d ( ν ′ ) ) ≥ Ψ ( γ ′ ′ ) ≥ Ψ ( γ ) − K W 2 ( ν , ν ′ ) = Φ ( κ d ( ν ) ) − K W 2 ( ν , ν ′ ) \Phi(\kappa_{d}(\nu'))\ge\Psi(\gamma'')\ge\Psi(\gamma)-K\,W_{2}(\nu,\nu')=\Phi(\kappa_{d}(\nu))-K\,W_{2}(\nu,\nu') Φ ( κ d ( ν ′ )) ≥ Ψ ( γ ′′ ) ≥ Ψ ( γ ) − K W 2 ( ν , ν ′ ) = Φ ( κ d ( ν )) − K W 2 ( ν , ν ′ ) . Exchanging ν \nu ν and ν ′ \nu' ν ′ and using W 2 ( ν ′ , ν ) = W 2 ( ν , ν ′ ) W_{2}(\nu',\nu)=W_{2}(\nu,\nu') W 2 ( ν ′ , ν ) = W 2 ( ν , ν ′ ) from The Noncommutative Wasserstein Distance: Existence of Optimal Couplings, Symmetry, Separation, a Moment Bound, Weak-Star Lower Semicontinuity, and Displacement Interpolation §symmetry gives the claim. Consequently, for ν ∈ Σ d , r \nu\in\Sigma_{d,r} ν ∈ Σ d , r and real ε > 0 \varepsilon>0 ε > 0 , the choice δ = ε / K \delta=\varepsilon/K δ = ε / K gives ∣ Φ ( κ d ( ν ) ) − Φ ( κ d ( ν ′ ) ) ∣ < ε |\Phi(\kappa_{d}(\nu))-\Phi(\kappa_{d}(\nu'))|<\varepsilon ∣Φ ( κ d ( ν )) − Φ ( κ d ( ν ′ )) ∣ < ε whenever ν ′ ∈ Σ d , r \nu'\in\Sigma_{d,r} ν ′ ∈ Σ d , r and W 2 ( ν , ν ′ ) < δ W_{2}(\nu,\nu')<\delta W 2 ( ν , ν ′ ) < δ , which is continuity in the sense of Continuous Map Between Metric Spaces for the metric space of The Noncommutative Laws with a Norm Bound Form a Complete Bounded Metric Space with Interpolation Points and the metric of Plan Jets and Hamiltonians on Square-Integrable Noncommutative Laws: Standing Notation §metrics .
Proof of clause 3. Put c = h ′ ( s ) / s + 2 θ c=h'(s)/s+2\theta c = h ′ ( s ) / s + 2 θ if s > 0 s>0 s > 0 and c = 0 c=0 c = 0 if s = 0 s=0 s = 0 , so that S = P − c a S=P-ca S = P − c a in both cases (for s = 0 s=0 s = 0 , S = P S=P S = P ), and S S S is an L 2 L^{2} L 2 d d d -tuple by Sums, Real Multiples and the Pairing of Square-Integrable Tuples in a Tracial W*-Probability Space §operations . Since ( X , P , X ′ ) (X,P,X') ( X , P , X ′ ) realises γ \gamma γ , s = s ( γ ) s=s(\gamma) s = s ( γ ) and p ( γ ) = ⟨ P , a ⟩ 2 p(\gamma)=\langle P,a\rangle_{2} p ( γ ) = ⟨ P , a ⟩ 2 . By Step B, l a w ( X ′ ) = κ d ( ν ) \mathrm{law}(X')=\kappa_{d}(\nu) law ( X ′ ) = κ d ( ν ) , and by Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §push-forward , p r # 1 l a w ( X ′ , S ) = l a w ( X ′ ) \mathrm{pr}^{1}_{\#}\mathrm{law}(X',S)=\mathrm{law}(X') pr # 1 law ( X ′ , S ) = law ( X ′ ) , so l a w ( X ′ , S ) \mathrm{law}(X',S) law ( X ′ , S ) is a plan at κ d ( ν ) \kappa_{d}(\nu) κ d ( ν ) in the sense of Plans at a Square-Integrable Noncommutative Law §plan . By Step A, h ′ ( s ) ≥ 0 h'(s)\ge0 h ′ ( s ) ≥ 0 , so c ≥ 0 c\ge0 c ≥ 0 ; if s > 0 s>0 s > 0 , ∥ S − P ∥ 2 = c s = h ′ ( s ) + 2 θ s \lVert S-P\rVert_{2}=c\,s=h'(s)+2\theta s ∥ S − P ∥ 2 = c s = h ′ ( s ) + 2 θ s , and if s = 0 s=0 s = 0 , ∥ S − P ∥ 2 = 0 = h ′ ( 0 ) + 2 θ ⋅ 0 \lVert S-P\rVert_{2}=0=h'(0)+2\theta\cdot0 ∥ S − P ∥ 2 = 0 = h ′ ( 0 ) + 2 θ ⋅ 0 because h ′ ( 0 ) = 0 h'(0)=0 h ′ ( 0 ) = 0 . This proves the norm bound.
Fix a real η > 0 \eta>0 η > 0 . The radius is chosen as follows, in this order. If s > 0 s>0 s > 0 , let ρ > 0 \rho>0 ρ > 0 be the radius r η / 2 r_{\eta/2} r η /2 of the hypothesis on h h h at the point s s s with η / 2 \eta/2 η /2 in place of η \eta η , and put r ∗ = min ( ρ , η / ( 2 ( L / ( 2 s ) + θ ) ) ) r_{*}=\min\bigl(\rho,\ \eta/(2(L/(2s)+\theta))\bigr) r ∗ = min ( ρ , η / ( 2 ( L / ( 2 s ) + θ )) ) . If s = 0 s=0 s = 0 , let ρ > 0 \rho>0 ρ > 0 be the radius r η / 2 r_{\eta/2} r η /2 of the hypothesis on h h h at the point 0 0 0 , and put r ∗ = min ( ρ , η / ( 2 θ ) ) r_{*}=\min(\rho,\ \eta/(2\theta)) r ∗ = min ( ρ , η / ( 2 θ )) . We verify the inequality of Plan Superdifferentials, Plan Subdifferentials and Plan Jets with Slack on Square-Integrable Noncommutative Laws §sub with slack 0 0 0 and radius r ∗ r_{*} r ∗ . Let ( H 1 , M 1 , Ω 1 ) (H_{1},M_{1},\Omega_{1}) ( H 1 , M 1 , Ω 1 ) be a tracial W*-probability space and Y , S ~ , Y ′ ′ Y,\tilde{S},Y'' Y , S ~ , Y ′′ L 2 L^{2} L 2 d d d -tuples of it with l a w ( Y , S ~ ) = l a w ( X ′ , S ) \mathrm{law}(Y,\tilde{S})=\mathrm{law}(X',S) law ( Y , S ~ ) = law ( X ′ , S ) and ∥ Y ′ ′ − Y ∥ 2 < r ∗ \lVert Y''-Y\rVert_{2}<r_{*} ∥ Y ′′ − Y ∥ 2 < r ∗ .
Step D with Z = ( X ′ , S ) Z=(X',S) Z = ( X ′ , S ) , W = ( X , P ) W=(X,P) W = ( X , P ) , Z 1 = ( Y , S ~ ) Z_{1}=(Y,\tilde{S}) Z 1 = ( Y , S ~ ) and V = Y ′ ′ V=Y'' V = Y ′′ gives L 2 L^{2} L 2 d d d -tuples X ^ ′ , S ^ , X ^ , P ^ , Y ^ ′ ′ \hat{X}',\hat{S},\hat{X},\hat{P},\hat{Y}'' X ^ ′ , S ^ , X ^ , P ^ , Y ^ ′′ of one tracial W*-probability space with l a w ( ( X ^ ′ , S ^ ) , ( X ^ , P ^ ) ) = l a w ( ( X ′ , S ) , ( X , P ) ) \mathrm{law}((\hat{X}',\hat{S}),(\hat{X},\hat{P}))=\mathrm{law}((X',S),(X,P)) law (( X ^ ′ , S ^ ) , ( X ^ , P ^ )) = law (( X ′ , S ) , ( X , P )) and l a w ( ( X ^ ′ , S ^ ) , Y ^ ′ ′ ) = l a w ( ( Y , S ~ ) , Y ′ ′ ) \mathrm{law}((\hat{X}',\hat{S}),\hat{Y}'')=\mathrm{law}((Y,\tilde{S}),Y'') law (( X ^ ′ , S ^ ) , Y ^ ′′ ) = law (( Y , S ~ ) , Y ′′ ) . By Step C: l a w ( X ^ , P ^ , X ^ ′ ) = l a w ( X , P , X ′ ) = γ \mathrm{law}(\hat{X},\hat{P},\hat{X}')=\mathrm{law}(X,P,X')=\gamma law ( X ^ , P ^ , X ^ ′ ) = law ( X , P , X ′ ) = γ ; ∥ S ^ − P ^ + c ( X ^ ′ − X ^ ) ∥ 2 = ∥ S − P + c a ∥ 2 = 0 \lVert\hat{S}-\hat{P}+c(\hat{X}'-\hat{X})\rVert_{2}=\lVert S-P+ca\rVert_{2}=0 ∥ S ^ − P ^ + c ( X ^ ′ − X ^ ) ∥ 2 = ∥ S − P + c a ∥ 2 = 0 , so S ^ = P ^ − c a ^ \hat{S}=\hat{P}-c\hat{a} S ^ = P ^ − c a ^ with a ^ = X ^ ′ − X ^ \hat{a}=\hat{X}'-\hat{X} a ^ = X ^ ′ − X ^ ; and, with v = Y ^ ′ ′ − X ^ ′ v=\hat{Y}''-\hat{X}' v = Y ^ ′′ − X ^ ′ , l a w ( Y ^ ′ ′ ) = l a w ( Y ′ ′ ) \mathrm{law}(\hat{Y}'')=\mathrm{law}(Y'') law ( Y ^ ′′ ) = law ( Y ′′ ) , ∥ v ∥ 2 = ∥ Y ′ ′ − Y ∥ 2 \lVert v\rVert_{2}=\lVert Y''-Y\rVert_{2} ∥ v ∥ 2 = ∥ Y ′′ − Y ∥ 2 and ⟨ S ^ , v ⟩ 2 = ⟨ S ~ , Y ′ ′ − Y ⟩ 2 \langle\hat{S},v\rangle_{2}=\langle\tilde{S},Y''-Y\rangle_{2} ⟨ S ^ , v ⟩ 2 = ⟨ S ~ , Y ′′ − Y ⟩ 2 . Computing from the realisation ( X ^ , P ^ , X ^ ′ ) (\hat{X},\hat{P},\hat{X}') ( X ^ , P ^ , X ^ ′ ) of γ \gamma γ , s = ∥ a ^ ∥ 2 s=\lVert\hat{a}\rVert_{2} s = ∥ a ^ ∥ 2 and p ( γ ) = ⟨ P ^ , a ^ ⟩ 2 p(\gamma)=\langle\hat{P},\hat{a}\rangle_{2} p ( γ ) = ⟨ P ^ , a ^ ⟩ 2 . Put γ ′ ′ = l a w ( X ^ , P ^ , Y ^ ′ ′ ) \gamma''=\mathrm{law}(\hat{X},\hat{P},\hat{Y}'') γ ′′ = law ( X ^ , P ^ , Y ^ ′′ ) . As in Step E, B # γ ′ ′ = l a w ( X ^ , P ^ ) = B # γ = κ 2 d ( π ) B_{\#}\gamma''=\mathrm{law}(\hat{X},\hat{P})=B_{\#}\gamma=\kappa_{2d}(\pi) B # γ ′′ = law ( X ^ , P ^ ) = B # γ = κ 2 d ( π ) and C # γ ′ ′ = l a w ( Y ^ ′ ′ ) C_{\#}\gamma''=\mathrm{law}(\hat{Y}'') C # γ ′′ = law ( Y ^ ′′ ) by Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §push-forward , so γ ′ ′ ∈ C ( l a w ( Y ^ ′ ′ ) ) \gamma''\in\mathcal{C}(\mathrm{law}(\hat{Y}'')) γ ′′ ∈ C ( law ( Y ^ ′′ )) by Couplings of a Square-Integrable Plan with a Law: Displacement and Momentum Pairing §couplings , and s ( γ ′ ′ ) = t s(\gamma'')=t s ( γ ′′ ) = t with t = ∥ a ^ + v ∥ 2 t=\lVert\hat{a}+v\rVert_{2} t = ∥ a ^ + v ∥ 2 and p ( γ ′ ′ ) = ⟨ P ^ , a ^ + v ⟩ 2 p(\gamma'')=\langle\hat{P},\hat{a}+v\rangle_{2} p ( γ ′′ ) = ⟨ P ^ , a ^ + v ⟩ 2 . We have ∣ t − s ∣ ≤ ∥ v ∥ 2 < r ∗ ≤ ρ |t-s|\le\lVert v\rVert_{2}<r_{*}\le\rho ∣ t − s ∣ ≤ ∥ v ∥ 2 < r ∗ ≤ ρ and t 2 − s 2 = 2 ⟨ a ^ , v ⟩ 2 + ∥ v ∥ 2 2 t^{2}-s^{2}=2\langle\hat{a},v\rangle_{2}+\lVert v\rVert_{2}^{2} t 2 − s 2 = 2 ⟨ a ^ , v ⟩ 2 + ∥ v ∥ 2 2 .
Case s > 0 s>0 s > 0 . Since 0 ≤ ( t − s ) 2 0\le(t-s)^{2} 0 ≤ ( t − s ) 2 gives 2 s ( t − s ) ≤ t 2 − s 2 2s(t-s)\le t^{2}-s^{2} 2 s ( t − s ) ≤ t 2 − s 2 , we get t − s ≤ ⟨ a ^ , v ⟩ 2 / s + ∥ v ∥ 2 2 / ( 2 s ) t-s\le\langle\hat{a},v\rangle_{2}/s+\lVert v\rVert_{2}^{2}/(2s) t − s ≤ ⟨ a ^ , v ⟩ 2 / s + ∥ v ∥ 2 2 / ( 2 s ) . By the choice of ρ \rho ρ , h ( t ) − h ( s ) ≤ h ′ ( s ) ( t − s ) + η 2 ∣ t − s ∣ h(t)-h(s)\le h'(s)(t-s)+\tfrac{\eta}{2}|t-s| h ( t ) − h ( s ) ≤ h ′ ( s ) ( t − s ) + 2 η ∣ t − s ∣ , and, since 0 ≤ h ′ ( s ) ≤ L 0\le h'(s)\le L 0 ≤ h ′ ( s ) ≤ L and ∣ t − s ∣ ≤ ∥ v ∥ 2 |t-s|\le\lVert v\rVert_{2} ∣ t − s ∣ ≤ ∥ v ∥ 2 ,
h ( t ) − h ( s ) ≤ h ′ ( s ) s ⟨ a ^ , v ⟩ 2 + L 2 s ∥ v ∥ 2 2 + η 2 ∥ v ∥ 2 . h(t)-h(s)\le\frac{h'(s)}{s}\langle\hat{a},v\rangle_{2}+\frac{L}{2s}\lVert v\rVert_{2}^{2}+\frac{\eta}{2}\lVert v\rVert_{2}. h ( t ) − h ( s ) ≤ s h ′ ( s ) ⟨ a ^ , v ⟩ 2 + 2 s L ∥ v ∥ 2 2 + 2 η ∥ v ∥ 2 .
Therefore
Ψ ( γ ′ ′ ) − Ψ ( γ ) = ⟨ P ^ , v ⟩ 2 − ( h ( t ) − h ( s ) ) − θ ( 2 ⟨ a ^ , v ⟩ 2 + ∥ v ∥ 2 2 ) ≥ ⟨ P ^ − c a ^ , v ⟩ 2 − ( η 2 + ( L 2 s + θ ) ∥ v ∥ 2 ) ∥ v ∥ 2 , \Psi(\gamma'')-\Psi(\gamma)=\langle\hat{P},v\rangle_{2}-\bigl(h(t)-h(s)\bigr)-\theta\bigl(2\langle\hat{a},v\rangle_{2}+\lVert v\rVert_{2}^{2}\bigr)\ge\langle\hat{P}-c\hat{a},v\rangle_{2}-\Bigl(\frac{\eta}{2}+\Bigl(\frac{L}{2s}+\theta\Bigr)\lVert v\rVert_{2}\Bigr)\lVert v\rVert_{2}, Ψ ( γ ′′ ) − Ψ ( γ ) = ⟨ P ^ , v ⟩ 2 − ( h ( t ) − h ( s ) ) − θ ( 2 ⟨ a ^ , v ⟩ 2 + ∥ v ∥ 2 2 ) ≥ ⟨ P ^ − c a ^ , v ⟩ 2 − ( 2 η + ( 2 s L + θ ) ∥ v ∥ 2 ) ∥ v ∥ 2 ,
and ( L / ( 2 s ) + θ ) ∥ v ∥ 2 < η / 2 (L/(2s)+\theta)\lVert v\rVert_{2}<\eta/2 ( L / ( 2 s ) + θ ) ∥ v ∥ 2 < η /2 by the choice of r ∗ r_{*} r ∗ . Hence Ψ ( γ ′ ′ ) ≥ Ψ ( γ ) + ⟨ S ^ , v ⟩ 2 − η ∥ v ∥ 2 \Psi(\gamma'')\ge\Psi(\gamma)+\langle\hat{S},v\rangle_{2}-\eta\lVert v\rVert_{2} Ψ ( γ ′′ ) ≥ Ψ ( γ ) + ⟨ S ^ , v ⟩ 2 − η ∥ v ∥ 2 .
Case s = 0 s=0 s = 0 . Then a ^ = 0 \hat{a}=0 a ^ = 0 , S ^ = P ^ \hat{S}=\hat{P} S ^ = P ^ , t = ∥ v ∥ 2 t=\lVert v\rVert_{2} t = ∥ v ∥ 2 , p ( γ ) = 0 p(\gamma)=0 p ( γ ) = 0 and Ψ ( γ ) = − h ( 0 ) = 0 \Psi(\gamma)=-h(0)=0 Ψ ( γ ) = − h ( 0 ) = 0 . By the choice of ρ \rho ρ and h ( 0 ) = h ′ ( 0 ) = 0 h(0)=h'(0)=0 h ( 0 ) = h ′ ( 0 ) = 0 , h ( t ) ≤ η 2 t h(t)\le\tfrac{\eta}{2}t h ( t ) ≤ 2 η t ; and θ t 2 ≤ η 2 t \theta t^{2}\le\tfrac{\eta}{2}t θ t 2 ≤ 2 η t since t < η / ( 2 θ ) t<\eta/(2\theta) t < η / ( 2 θ ) . Hence Ψ ( γ ′ ′ ) = ⟨ P ^ , v ⟩ 2 − h ( t ) − θ t 2 ≥ Ψ ( γ ) + ⟨ S ^ , v ⟩ 2 − η ∥ v ∥ 2 \Psi(\gamma'')=\langle\hat{P},v\rangle_{2}-h(t)-\theta t^{2}\ge\Psi(\gamma)+\langle\hat{S},v\rangle_{2}-\eta\lVert v\rVert_{2} Ψ ( γ ′′ ) = ⟨ P ^ , v ⟩ 2 − h ( t ) − θ t 2 ≥ Ψ ( γ ) + ⟨ S ^ , v ⟩ 2 − η ∥ v ∥ 2 .
In both cases, since γ ′ ′ ∈ C ( l a w ( Y ^ ′ ′ ) ) \gamma''\in\mathcal{C}(\mathrm{law}(\hat{Y}'')) γ ′′ ∈ C ( law ( Y ^ ′′ )) , l a w ( Y ^ ′ ′ ) = l a w ( Y ′ ′ ) \mathrm{law}(\hat{Y}'')=\mathrm{law}(Y'') law ( Y ^ ′′ ) = law ( Y ′′ ) and Ψ ( γ ) = Φ ( κ d ( ν ) ) \Psi(\gamma)=\Phi(\kappa_{d}(\nu)) Ψ ( γ ) = Φ ( κ d ( ν )) , the Notation and the identities from Step C give
Φ ( l a w ( Y ′ ′ ) ) ≥ Ψ ( γ ′ ′ ) ≥ Φ ( κ d ( ν ) ) + ⟨ S ~ , Y ′ ′ − Y ⟩ 2 − ( 0 + η ) ∥ Y ′ ′ − Y ∥ 2 . \Phi(\mathrm{law}(Y''))\ge\Psi(\gamma'')\ge\Phi(\kappa_{d}(\nu))+\langle\tilde{S},Y''-Y\rangle_{2}-(0+\eta)\lVert Y''-Y\rVert_{2}. Φ ( law ( Y ′′ )) ≥ Ψ ( γ ′′ ) ≥ Φ ( κ d ( ν )) + ⟨ S ~ , Y ′′ − Y ⟩ 2 − ( 0 + η ) ∥ Y ′′ − Y ∥ 2 .
The left side is Φ M 1 ( Y ′ ′ ) \Phi_{M_{1}}(Y'') Φ M 1 ( Y ′′ ) by Plan Jets and Hamiltonians on Square-Integrable Noncommutative Laws: Standing Notation §lifts and Lifts of Functions on Square-Integrable Noncommutative Laws to Square-Integrable Tuples §lift . So l a w ( X ′ , S ) \mathrm{law}(X',S) law ( X ′ , S ) is a plan subdifferential of Φ \Phi Φ at κ d ( ν ) \kappa_{d}(\nu) κ d ( ν ) with slack 0 0 0 in the sense of Plan Superdifferentials, Plan Subdifferentials and Plan Jets with Slack on Square-Integrable Noncommutative Laws §sub , that is, l a w ( X ′ , S ) ∈ J − Φ ( κ d ( ν ) ) \mathrm{law}(X',S)\in J^{-}\Phi(\kappa_{d}(\nu)) law ( X ′ , S ) ∈ J − Φ ( κ d ( ν )) by Plan Superdifferentials, Plan Subdifferentials and Plan Jets with Slack on Square-Integrable Noncommutative Laws §subjet .
Proof of clause 4. First, Ψ ( γ ) ≤ − θ 2 s ( γ ) 2 \Psi(\gamma)\le-\tfrac{\theta}{2}s(\gamma)^{2} Ψ ( γ ) ≤ − 2 θ s ( γ ) 2 for every γ ∈ C ( κ d ( μ ) ) \gamma\in\mathcal{C}(\kappa_{d}(\mu)) γ ∈ C ( κ d ( μ )) . Write s = s ( γ ) s=s(\gamma) s = s ( γ ) . If s < r 0 s<r_{0} s < r 0 , the hypothesis gives p ( γ ) ≤ 1 2 h ( s ) p(\gamma)\le\tfrac12h(s) p ( γ ) ≤ 2 1 h ( s ) , so Ψ ( γ ) ≤ − 1 2 h ( s ) − θ s 2 ≤ − θ 2 s 2 \Psi(\gamma)\le-\tfrac12h(s)-\theta s^{2}\le-\tfrac{\theta}{2}s^{2} Ψ ( γ ) ≤ − 2 1 h ( s ) − θ s 2 ≤ − 2 θ s 2 because h ≥ 0 h\ge0 h ≥ 0 . If s ≥ r 0 s\ge r_{0} s ≥ r 0 , Step B (with ν = μ ∈ Σ d , r \nu=\mu\in\Sigma_{d,r} ν = μ ∈ Σ d , r ) gives p ( γ ) ≤ a π s ≤ θ r 0 2 s ≤ θ 2 s 2 p(\gamma)\le a_{\pi}s\le\tfrac{\theta r_{0}}{2}s\le\tfrac{\theta}{2}s^{2} p ( γ ) ≤ a π s ≤ 2 θ r 0 s ≤ 2 θ s 2 , so Ψ ( γ ) ≤ θ 2 s 2 − h ( s ) − θ s 2 ≤ − θ 2 s 2 \Psi(\gamma)\le\tfrac{\theta}{2}s^{2}-h(s)-\theta s^{2}\le-\tfrac{\theta}{2}s^{2} Ψ ( γ ) ≤ 2 θ s 2 − h ( s ) − θ s 2 ≤ − 2 θ s 2 . Hence Φ ( κ d ( μ ) ) ≤ 0 \Phi(\kappa_{d}(\mu))\le0 Φ ( κ d ( μ )) ≤ 0 .
Second, by Every Square-Integrable Noncommutative Law is the Law of a Square-Integrable Tuple; Realisation of Couplings and of Almost Optimal Pairs §coupling there are L 2 L^{2} L 2 d d d -tuples X , P X,P X , P of a tracial W*-probability space with l a w ( X , P ) = κ 2 d ( π ) \mathrm{law}(X,P)=\kappa_{2d}(\pi) law ( X , P ) = κ 2 d ( π ) , and then l a w ( X ) = p r # 1 κ 2 d ( π ) = κ d ( μ ) \mathrm{law}(X)=\mathrm{pr}^{1}_{\#}\kappa_{2d}(\pi)=\kappa_{d}(\mu) law ( X ) = pr # 1 κ 2 d ( π ) = κ d ( μ ) as computed in Step B. Put γ 0 = l a w ( X , P , X ) \gamma_{0}=\mathrm{law}(X,P,X) γ 0 = law ( X , P , X ) . By Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §push-forward , B # γ 0 = l a w ( X , P ) = κ 2 d ( π ) B_{\#}\gamma_{0}=\mathrm{law}(X,P)=\kappa_{2d}(\pi) B # γ 0 = law ( X , P ) = κ 2 d ( π ) and C # γ 0 = l a w ( X ) = κ d ( μ ) C_{\#}\gamma_{0}=\mathrm{law}(X)=\kappa_{d}(\mu) C # γ 0 = law ( X ) = κ d ( μ ) , so γ 0 ∈ C ( κ d ( μ ) ) \gamma_{0}\in\mathcal{C}(\kappa_{d}(\mu)) γ 0 ∈ C ( κ d ( μ )) by Couplings of a Square-Integrable Plan with a Law: Displacement and Momentum Pairing §couplings . From this realisation, s ( γ 0 ) = ∥ X − X ∥ 2 = 0 s(\gamma_{0})=\lVert X-X\rVert_{2}=0 s ( γ 0 ) = ∥ X − X ∥ 2 = 0 and p ( γ 0 ) = ⟨ P , X − X ⟩ 2 = 0 p(\gamma_{0})=\langle P,X-X\rangle_{2}=0 p ( γ 0 ) = ⟨ P , X − X ⟩ 2 = 0 , so Ψ ( γ 0 ) = − h ( 0 ) = 0 \Psi(\gamma_{0})=-h(0)=0 Ψ ( γ 0 ) = − h ( 0 ) = 0 and Φ ( κ d ( μ ) ) ≥ 0 \Phi(\kappa_{d}(\mu))\ge0 Φ ( κ d ( μ )) ≥ 0 . Therefore Φ ( κ d ( μ ) ) = 0 \Phi(\kappa_{d}(\mu))=0 Φ ( κ d ( μ )) = 0 .
Proof of clause 5. Put w n = W 2 ( ν n , μ ) w_{n}=W_{2}(\nu_{n},\mu) w n = W 2 ( ν n , μ ) and s n = s ( γ n ) s_{n}=s(\gamma_{n}) s n = s ( γ n ) . The hypotheses of clause 4 hold, so Φ ( κ d ( μ ) ) = 0 \Phi(\kappa_{d}(\mu))=0 Φ ( κ d ( μ )) = 0 , and the estimate proved for clause 2 gives ∣ Φ ( κ d ( ν n ) ) ∣ ≤ K w n |\Phi(\kappa_{d}(\nu_{n}))|\le Kw_{n} ∣Φ ( κ d ( ν n )) ∣ ≤ K w n for every n n n ; hence ( Φ ( κ d ( ν n ) ) ) n (\Phi(\kappa_{d}(\nu_{n})))_{n} ( Φ ( κ d ( ν n )) ) n converges to 0 0 0 . For every n n n , Step E (with ν = ν n \nu=\nu_{n} ν = ν n , ν ′ = μ \nu'=\mu ν ′ = μ , γ = γ n \gamma=\gamma_{n} γ = γ n ) gives γ n ′ ′ ∈ C ( κ d ( μ ) ) \gamma''_{n}\in\mathcal{C}(\kappa_{d}(\mu)) γ n ′′ ∈ C ( κ d ( μ )) with ∣ s ( γ n ′ ′ ) − s n ∣ ≤ w n |s(\gamma''_{n})-s_{n}|\le w_{n} ∣ s ( γ n ′′ ) − s n ∣ ≤ w n and Ψ ( γ n ′ ′ ) ≥ Ψ ( γ n ) − K w n = Φ ( κ d ( ν n ) ) − K w n ≥ − 2 K w n \Psi(\gamma''_{n})\ge\Psi(\gamma_{n})-Kw_{n}=\Phi(\kappa_{d}(\nu_{n}))-Kw_{n}\ge-2Kw_{n} Ψ ( γ n ′′ ) ≥ Ψ ( γ n ) − K w n = Φ ( κ d ( ν n )) − K w n ≥ − 2 K w n . By the first part of the proof of clause 4, Ψ ( γ n ′ ′ ) ≤ − θ 2 s ( γ n ′ ′ ) 2 \Psi(\gamma''_{n})\le-\tfrac{\theta}{2}s(\gamma''_{n})^{2} Ψ ( γ n ′′ ) ≤ − 2 θ s ( γ n ′′ ) 2 , so s ( γ n ′ ′ ) 2 ≤ 4 K w n / θ s(\gamma''_{n})^{2}\le4Kw_{n}/\theta s ( γ n ′′ ) 2 ≤ 4 K w n / θ . Given a real ε > 0 \varepsilon>0 ε > 0 , choose N ∈ N N\in\mathbb{N} N ∈ N such that w n < min ( ε / 2 , θ ε 2 / ( 16 K ) ) w_{n}<\min\bigl(\varepsilon/2,\ \theta\varepsilon^{2}/(16K)\bigr) w n < min ( ε /2 , θ ε 2 / ( 16 K ) ) for all n ≥ N n\ge N n ≥ N , by the convergence of ( w n ) n (w_{n})_{n} ( w n ) n to 0 0 0 in the sense of The Real Numbers: Standing Notation and Background §sequences ; then for n ≥ N n\ge N n ≥ N , s ( γ n ′ ′ ) 2 < ε 2 / 4 s(\gamma''_{n})^{2}<\varepsilon^{2}/4 s ( γ n ′′ ) 2 < ε 2 /4 , so s ( γ n ′ ′ ) < ε / 2 s(\gamma''_{n})<\varepsilon/2 s ( γ n ′′ ) < ε /2 , and 0 ≤ s n ≤ s ( γ n ′ ′ ) + w n < ε 0\le s_{n}\le s(\gamma''_{n})+w_{n}<\varepsilon 0 ≤ s n ≤ s ( γ n ′′ ) + w n < ε . Hence ( s n ) n (s_{n})_{n} ( s n ) n converges to 0 0 0 .
For the last assertion, let ( X n , P n , X n ′ ) (X_{n},P_{n},X'_{n}) ( X n , P n , X n ′ ) and S n S_{n} S n be as stated. Since γ n \gamma_{n} γ n is a maximising coupling for ν n \nu_{n} ν n , clause 3 gives ∥ S n − P n ∥ 2 ≤ h ′ ( s n ) + 2 θ s n \lVert S_{n}-P_{n}\rVert_{2}\le h'(s_{n})+2\theta s_{n} ∥ S n − P n ∥ 2 ≤ h ′ ( s n ) + 2 θ s n , and h ′ ( s n ) ≥ 0 h'(s_{n})\ge0 h ′ ( s n ) ≥ 0 by Step A. Given a real ε > 0 \varepsilon>0 ε > 0 , let r h > 0 r_{h}>0 r h > 0 be the radius of the hypothesis on h ′ h' h ′ for ε / 2 \varepsilon/2 ε /2 in place of η \eta η , and then choose N ∈ N N\in\mathbb{N} N ∈ N with s n < min ( r h , ε / ( 4 θ ) ) s_{n}<\min\bigl(r_{h},\ \varepsilon/(4\theta)\bigr) s n < min ( r h , ε / ( 4 θ ) ) for n ≥ N n\ge N n ≥ N . Then 0 ≤ ∥ S n − P n ∥ 2 < ε / 2 + ε / 2 = ε 0\le\lVert S_{n}-P_{n}\rVert_{2}<\varepsilon/2+\varepsilon/2=\varepsilon 0 ≤ ∥ S n − P n ∥ 2 < ε /2 + ε /2 = ε for n ≥ N n\ge N n ≥ N , so ( ∥ S n − P n ∥ 2 ) n (\lVert S_{n}-P_{n}\rVert_{2})_{n} (∥ S n − P n ∥ 2 ) n converges to 0 0 0 .