Each result cited below is universally quantified over the data in its own statement.
Operators are handled in the notation of Complex Hilbert Spaces and Bounded Linear Maps: Standing Notation , and for a tracial W*-probability space and a d d d -tuple V V V of unitaries in it we use the notation V ( l ) V^{(l)} V ( l ) , V ( w ) V^{(w)} V ( w ) , A V \mathcal{A}_{V} A V and λ V \lambda_{V} λ V fixed in the statement of Unitary Laws Are the Laws of Unitary Tuples, and Are Realised Together with a Free Semicircular Family . For letters we use Words in Unitary Letters: Generators, Signs, Inverse Letters, Lengths and Adjoint Words §letters : a letter l l l with ε ( l ) = 1 \varepsilon(l)=1 ε ( l ) = 1 is the letter g ( l ) g(l) g ( l ) , one with ε ( l ) = − 1 \varepsilon(l)=-1 ε ( l ) = − 1 is the letter d + g ( l ) d+g(l) d + g ( l ) .
Fix λ ∈ L d \lambda\in\mathcal{L}_{d} λ ∈ L d . By Unitary Laws Are the Laws of Unitary Tuples, and Are Realised Together with a Free Semicircular Family §realisation there are a tracial W*-probability space ( H , M , Ω ) (H,M,\Omega) ( H , M , Ω ) with trace τ = τ M \tau=\tau_{M} τ = τ M , a d d d -tuple U U U of unitary operators on H H H belonging to M M M with λ U = λ \lambda_{U}=\lambda λ U = λ , and self-adjoint S 1 , … , S d ∈ M S_{1},\dots,S_{d}\in M S 1 , … , S d ∈ M with ∥ S j ∥ o p ≤ 2 \lVert S_{j}\rVert_{\mathrm{op}}\le2 ∥ S j ∥ op ≤ 2 , satisfying Unitary Laws Are the Laws of Unitary Tuples, and Are Realised Together with a Free Semicircular Family §centred and Unitary Laws Are the Laws of Unitary Tuples, and Are Realised Together with a Free Semicircular Family §covariance . We fix these data.
Step 0 (elementary facts). (O1) For T ∈ M T\in M T ∈ M , ∣ τ ( T ) ∣ = ∣ ⟨ Ω , T Ω ⟩ ∣ ≤ ∥ Ω ∥ ∥ T Ω ∥ ≤ ∥ T ∥ o p |\tau(T)|=|\langle\Omega,T\Omega\rangle|\le\lVert\Omega\rVert\,\lVert T\Omega\rVert\le\lVert T\rVert_{\mathrm{op}} ∣ τ ( T ) ∣ = ∣ ⟨ Ω , T Ω ⟩ ∣ ≤ ∥ Ω ∥ ∥ T Ω ∥ ≤ ∥ T ∥ op . Here the equality is Tracial W*-Probability Spaces §trace with Cyclic Tracial Operator Algebras and Their Traces §trace . For the first inequality, Cauchy-Schwarz Inequality in a Complex Inner Product Space gives ∣ ⟨ Ω , T Ω ⟩ ∣ 2 ≤ ⟨ Ω , Ω ⟩ ⟨ T Ω , T Ω ⟩ = ∥ Ω ∥ 2 ∥ T Ω ∥ 2 = ( ∥ Ω ∥ ∥ T Ω ∥ ) 2 |\langle\Omega,T\Omega\rangle|^{2}\le\langle\Omega,\Omega\rangle\langle T\Omega,T\Omega\rangle=\lVert\Omega\rVert^{2}\lVert T\Omega\rVert^{2}=(\lVert\Omega\rVert\,\lVert T\Omega\rVert)^{2} ∣ ⟨ Ω , T Ω ⟩ ∣ 2 ≤ ⟨ Ω , Ω ⟩ ⟨ T Ω , T Ω ⟩ = ∥ Ω ∥ 2 ∥ T Ω ∥ 2 = (∥ Ω ∥ ∥ T Ω ∥ ) 2 , and since both ∣ ⟨ Ω , T Ω ⟩ ∣ |\langle\Omega,T\Omega\rangle| ∣ ⟨ Ω , T Ω ⟩ ∣ and ∥ Ω ∥ ∥ T Ω ∥ \lVert\Omega\rVert\,\lVert T\Omega\rVert ∥ Ω ∥ ∥ T Ω ∥ are nonnegative, claim 2 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field yields ∣ ⟨ Ω , T Ω ⟩ ∣ ≤ ∥ Ω ∥ ∥ T Ω ∥ |\langle\Omega,T\Omega\rangle|\le\lVert\Omega\rVert\,\lVert T\Omega\rVert ∣ ⟨ Ω , T Ω ⟩ ∣ ≤ ∥ Ω ∥ ∥ T Ω ∥ . For the second, ∥ T ∥ o p \lVert T\rVert_{\mathrm{op}} ∥ T ∥ op is a bound for T T T by Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §least-bound , so ∥ T Ω ∥ ≤ ∥ T ∥ o p ∥ Ω ∥ \lVert T\Omega\rVert\le\lVert T\rVert_{\mathrm{op}}\lVert\Omega\rVert ∥ T Ω ∥ ≤ ∥ T ∥ op ∥ Ω ∥ by Bounded Linear Maps between Complex Normed Spaces and the Operator Norm §bounded , and ∥ Ω ∥ = 1 \lVert\Omega\rVert=1 ∥ Ω ∥ = 1 (Cyclic Tracial Operator Algebras and Their Traces §cyclic ). (O2) τ ( A B ) = τ ( B A ) \tau(AB)=\tau(BA) τ ( A B ) = τ ( B A ) for A , B ∈ M A,B\in M A , B ∈ M (Cyclic Tracial Operator Algebras and Their Traces §tracial ), τ \tau τ is linear and τ ( I ) = 1 \tau(I)=1 τ ( I ) = 1 (The Trace, the Conjugation and the Right Action of a Cyclic Tracial Operator Algebra §trace ). (O3) For every letter l l l , ∥ U ( l ) ∥ o p ≤ 1 \lVert U^{(l)}\rVert_{\mathrm{op}}\le1 ∥ U ( l ) ∥ op ≤ 1 , by claim 3 of Properties of Unitary Operators and ∥ U j ∗ ∥ o p = ∥ U j ∥ o p \lVert U_{j}^{*}\rVert_{\mathrm{op}}=\lVert U_{j}\rVert_{\mathrm{op}} ∥ U j ∗ ∥ op = ∥ U j ∥ op (Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint ). (O4) U ( u ) U ( v ) = U ( u v ) U^{(u)}U^{(v)}=U^{(uv)} U ( u ) U ( v ) = U ( uv ) for u , v ∈ W 2 d u,v\in W_{2d} u , v ∈ W 2 d , since U ( ∅ ) = I U^{(\varnothing)}=I U ( ∅ ) = I and the letters of u v uv uv are those of u u u followed by those of v v v (Words over a Finite Alphabet: the Empty Word, Concatenation and Reversal §concatenation ); in particular every U ( u ) U^{(u)} U ( u ) lies in A U \mathcal{A}_{U} A U and every product of operators U ( u ) U^{(u)} U ( u ) lies in A U \mathcal{A}_{U} A U . (O5) A map T ∈ L ( H ) T\in\mathcal{L}(H) T ∈ L ( H ) with T ∗ T = T T ∗ = I T^{*}T=TT^{*}=I T ∗ T = T T ∗ = I is unitary : every v v v equals T ( T ∗ v ) T(T^{*}v) T ( T ∗ v ) , and ⟨ T u , T v ⟩ = ⟨ u , T ∗ T v ⟩ = ⟨ u , v ⟩ \langle Tu,Tv\rangle=\langle u,T^{*}Tv\rangle=\langle u,v\rangle ⟨ T u , T v ⟩ = ⟨ u , T ∗ T v ⟩ = ⟨ u , v ⟩ , because T T T is the adjoint of T ∗ T^{*} T ∗ (Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-calculus , Adjoint of a Linear Map between Complex Inner Product Spaces §adjoint ). Conversely, each U j U_{j} U j is a bounded linear operator on H H H by claim 3 of Properties of Unitary Operators , so U j ∈ L ( H ) U_{j}\in\mathcal{L}(H) U j ∈ L ( H ) by Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §least-bound , and hence U j U_{j} U j has an adjoint U j ∗ ∈ L ( H ) U_{j}^{*}\in\mathcal{L}(H) U j ∗ ∈ L ( H ) by Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint ; it satisfies U j ∗ U j = U j U j ∗ = I U_{j}^{*}U_{j}=U_{j}U_{j}^{*}=I U j ∗ U j = U j U j ∗ = I : from ⟨ u , U j ∗ U j v ⟩ = ⟨ U j u , U j v ⟩ = ⟨ u , v ⟩ \langle u,U_{j}^{*}U_{j}v\rangle=\langle U_{j}u,U_{j}v\rangle=\langle u,v\rangle ⟨ u , U j ∗ U j v ⟩ = ⟨ U j u , U j v ⟩ = ⟨ u , v ⟩ with u = U j ∗ U j v − v u=U_{j}^{*}U_{j}v-v u = U j ∗ U j v − v we get U j ∗ U j v = v U_{j}^{*}U_{j}v=v U j ∗ U j v = v , and writing v = U j u v=U_{j}u v = U j u by surjectivity, U j U j ∗ v = U j ( U j ∗ U j u ) = v U_{j}U_{j}^{*}v=U_{j}(U_{j}^{*}U_{j}u)=v U j U j ∗ v = U j ( U j ∗ U j u ) = v . (O6) For 0 ≤ t ≤ 1 16 0\le t\le\frac1{16} 0 ≤ t ≤ 16 1 let t \sqrt t t be the nonnegative square root ; then t ≤ 1 4 \sqrt t\le\frac14 t ≤ 4 1 by claim 2 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field , hence t = t t ≤ 1 4 t t=\sqrt t\,\sqrt t\le\frac14\sqrt t t = t t ≤ 4 1 t and t t ≤ 1 4 t ≤ 1 16 t t\sqrt t\le\frac14 t\le\frac1{16}\sqrt t t t ≤ 4 1 t ≤ 16 1 t .
Step 1 (Cayley transforms). Let 0 ≤ t ≤ 1 16 0\le t\le\frac1{16} 0 ≤ t ≤ 16 1 and j ∈ [ d ] j\in[d] j ∈ [ d ] . Put X j = i t 2 S j ∈ M X_{j}=\frac{i\sqrt t}{2}S_{j}\in M X j = 2 i t S j ∈ M ; then ∥ X j ∥ o p ≤ t ≤ 1 4 < 1 \lVert X_{j}\rVert_{\mathrm{op}}\le\sqrt t\le\frac14<1 ∥ X j ∥ op ≤ t ≤ 4 1 < 1 by Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §operations . By The Neumann Series, Inverses in Double Commutants, and Invertibility of A - iyI for Self-Adjoint A §neumann , T j = I − X j T_{j}=I-X_{j} T j = I − X j is a bijection of H H H onto H H H with T j − 1 ∈ L ( H ) T_{j}^{-1}\in\mathcal{L}(H) T j − 1 ∈ L ( H ) and ∥ T j − 1 ∥ o p ≤ 1 1 − 1 / 4 = 4 3 \lVert T_{j}^{-1}\rVert_{\mathrm{op}}\le\frac{1}{1-1/4}=\frac43 ∥ T j − 1 ∥ op ≤ 1 − 1/4 1 = 3 4 ; as T j ∈ M T_{j}\in M T j ∈ M and M = M ′ ′ M=M'' M = M ′′ (Tracial W*-Probability Spaces §space ), T j − 1 ∈ M T_{j}^{-1}\in M T j − 1 ∈ M by The Neumann Series, Inverses in Double Commutants, and Invertibility of A - iyI for Self-Adjoint A §commutant . Put T j ′ = I + X j T'_{j}=I+X_{j} T j ′ = I + X j and C j = T j ′ T j − 1 ∈ M C_{j}=T'_{j}T_{j}^{-1}\in M C j = T j ′ T j − 1 ∈ M (Cyclic Tracial Operator Algebras and Their Traces §star-algebra ); we write C j ( t ) C_{j}(t) C j ( t ) when t t t matters.
(a) C j ∗ C j = C j C j ∗ = I C_{j}^{*}C_{j}=C_{j}C_{j}^{*}=I C j ∗ C j = C j C j ∗ = I . Since S j S_{j} S j is self-adjoint, S j ∗ = S j S_{j}^{*}=S_{j} S j ∗ = S j and X j ∗ = i t 2 ‾ S j = − X j X_{j}^{*}=\overline{\tfrac{i\sqrt t}{2}}S_{j}=-X_{j} X j ∗ = 2 i t S j = − X j (Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-calculus , Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-unique ); so T j ∗ = T j ′ T_{j}^{*}=T'_{j} T j ∗ = T j ′ and T j ′ ∗ = T j T'^{*}_{j}=T_{j} T j ′ ∗ = T j . Also T j T j ′ = I − X j 2 = T j ′ T j T_{j}T'_{j}=I-X_{j}^{2}=T'_{j}T_{j} T j T j ′ = I − X j 2 = T j ′ T j , so T j − 1 T_{j}^{-1} T j − 1 commutes with T j ′ T'_{j} T j ′ by The Neumann Series, Inverses in Double Commutants, and Invertibility of A - iyI for Self-Adjoint A §commutant . From T j T j − 1 = T j − 1 T j = I T_{j}T_{j}^{-1}=T_{j}^{-1}T_{j}=I T j T j − 1 = T j − 1 T j = I and Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-calculus , ( T j − 1 ) ∗ T j ′ = ( T j T j − 1 ) ∗ = I (T_{j}^{-1})^{*}T'_{j}=(T_{j}T_{j}^{-1})^{*}=I ( T j − 1 ) ∗ T j ′ = ( T j T j − 1 ) ∗ = I and T j ′ ( T j − 1 ) ∗ = I T'_{j}(T_{j}^{-1})^{*}=I T j ′ ( T j − 1 ) ∗ = I ; so T j ′ T'_{j} T j ′ is a bijection with inverse T j ′ − 1 = ( T j − 1 ) ∗ T'^{-1}_{j}=(T_{j}^{-1})^{*} T j ′ − 1 = ( T j − 1 ) ∗ , and C j ∗ = T j ′ − 1 T j C_{j}^{*}=T'^{-1}_{j}T_{j} C j ∗ = T j ′ − 1 T j . Then C j ∗ C j = T j ′ − 1 T j T j ′ T j − 1 = T j ′ − 1 T j ′ T j T j − 1 = I C_{j}^{*}C_{j}=T'^{-1}_{j}T_{j}T'_{j}T_{j}^{-1}=T'^{-1}_{j}T'_{j}T_{j}T_{j}^{-1}=I C j ∗ C j = T j ′ − 1 T j T j ′ T j − 1 = T j ′ − 1 T j ′ T j T j − 1 = I . Multiplying T j − 1 T j ′ = T j ′ T j − 1 T_{j}^{-1}T'_{j}=T'_{j}T_{j}^{-1} T j − 1 T j ′ = T j ′ T j − 1 by T j ′ − 1 T'^{-1}_{j} T j ′ − 1 on both sides gives T j ′ − 1 T j − 1 = T j − 1 T j ′ − 1 T'^{-1}_{j}T_{j}^{-1}=T_{j}^{-1}T'^{-1}_{j} T j ′ − 1 T j − 1 = T j − 1 T j ′ − 1 , and hence C j C j ∗ = T j ′ T j − 1 T j ′ − 1 T j = T j ′ T j ′ − 1 T j − 1 T j = I C_{j}C_{j}^{*}=T'_{j}T_{j}^{-1}T'^{-1}_{j}T_{j}=T'_{j}T'^{-1}_{j}T_{j}^{-1}T_{j}=I C j C j ∗ = T j ′ T j − 1 T j ′ − 1 T j = T j ′ T j ′ − 1 T j − 1 T j = I .
(b) The path. Put V j = U j C j ∈ M V_{j}=U_{j}C_{j}\in M V j = U j C j ∈ M and V = V ( t ) = ( V 1 , … , V d ) V=V(t)=(V_{1},\dots,V_{d}) V = V ( t ) = ( V 1 , … , V d ) . By (O5) and (a), V j ∗ V j = C j ∗ U j ∗ U j C j = I V_{j}^{*}V_{j}=C_{j}^{*}U_{j}^{*}U_{j}C_{j}=I V j ∗ V j = C j ∗ U j ∗ U j C j = I and V j V j ∗ = U j C j C j ∗ U j ∗ = I V_{j}V_{j}^{*}=U_{j}C_{j}C_{j}^{*}U_{j}^{*}=I V j V j ∗ = U j C j C j ∗ U j ∗ = I , so V j V_{j} V j is unitary by (O5). Define γ ( t ) = λ V ( t ) \gamma(t)=\lambda_{V(t)} γ ( t ) = λ V ( t ) ; it belongs to L d \mathcal{L}_{d} L d by Unitary Laws Are the Laws of Unitary Tuples, and Are Realised Together with a Free Semicircular Family §law . For t = 0 t=0 t = 0 we have X j = 0 X_{j}=0 X j = 0 , T j = T j ′ = T j − 1 = I T_{j}=T'_{j}=T_{j}^{-1}=I T j = T j ′ = T j − 1 = I , C j = I C_{j}=I C j = I , V = U V=U V = U , so γ ( 0 ) = λ U = λ \gamma(0)=\lambda_{U}=\lambda γ ( 0 ) = λ U = λ .
(c) Expansion. Since T j ′ = T j + 2 X j T'_{j}=T_{j}+2X_{j} T j ′ = T j + 2 X j , C j = I + 2 X j T j − 1 C_{j}=I+2X_{j}T_{j}^{-1} C j = I + 2 X j T j − 1 ; and T j − 1 = I + X j T j − 1 T_{j}^{-1}=I+X_{j}T_{j}^{-1} T j − 1 = I + X j T j − 1 , because T j − 1 − X j T j − 1 = T j T j − 1 = I T_{j}^{-1}-X_{j}T_{j}^{-1}=T_{j}T_{j}^{-1}=I T j − 1 − X j T j − 1 = T j T j − 1 = I . Substituting twice, X j T j − 1 = X j + X j 2 + X j 3 T j − 1 X_{j}T_{j}^{-1}=X_{j}+X_{j}^{2}+X_{j}^{3}T_{j}^{-1} X j T j − 1 = X j + X j 2 + X j 3 T j − 1 , so, with 2 X j = i t S j 2X_{j}=i\sqrt t\,S_{j} 2 X j = i t S j and 2 X j 2 = − t 2 S j 2 2X_{j}^{2}=-\frac t2S_{j}^{2} 2 X j 2 = − 2 t S j 2 ,
D j : = C j − I = i t S j − t 2 S j 2 + R j , R j = 2 X j 3 T j − 1 , ∥ R j ∥ o p ≤ 2 ( t ) 3 4 3 = 8 3 t t , D_{j}:=C_{j}-I=i\sqrt t\,S_{j}-\tfrac t2S_{j}^{2}+R_{j},\qquad R_{j}=2X_{j}^{3}T_{j}^{-1},\qquad\lVert R_{j}\rVert_{\mathrm{op}}\le2(\sqrt t)^{3}\tfrac43=\tfrac83t\sqrt t, D j := C j − I = i t S j − 2 t S j 2 + R j , R j = 2 X j 3 T j − 1 , ∥ R j ∥ op ≤ 2 ( t ) 3 3 4 = 3 8 t t ,
and, by Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-calculus , D j ∗ = C j ∗ − I = − i t S j − t 2 S j 2 + R j ∗ D_{j}^{*}=C_{j}^{*}-I=-i\sqrt t\,S_{j}-\frac t2S_{j}^{2}+R_{j}^{*} D j ∗ = C j ∗ − I = − i t S j − 2 t S j 2 + R j ∗ with ∥ R j ∗ ∥ o p = ∥ R j ∥ o p \lVert R_{j}^{*}\rVert_{\mathrm{op}}=\lVert R_{j}\rVert_{\mathrm{op}} ∥ R j ∗ ∥ op = ∥ R j ∥ op (Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint ). By Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §operations and (O6), for e ∈ { 1 , − 1 } e\in\{1,-1\} e ∈ { 1 , − 1 } and D j ( 1 ) = D j D^{(1)}_{j}=D_{j} D j ( 1 ) = D j , D j ( − 1 ) = D j ∗ D^{(-1)}_{j}=D_{j}^{*} D j ( − 1 ) = D j ∗ ,
∥ D j ( e ) ∥ o p ≤ 2 t + 2 t + 8 3 t t ≤ ( 2 + 1 2 + 1 6 ) t ≤ 3 t , ∥ D j ( e ) − i e t S j ∥ o p ≤ 2 t + 8 3 t t ≤ 3 t . (B) \lVert D^{(e)}_{j}\rVert_{\mathrm{op}}\le2\sqrt t+2t+\tfrac83t\sqrt t\le\bigl(2+\tfrac12+\tfrac16\bigr)\sqrt t\le3\sqrt t,\qquad\lVert D^{(e)}_{j}-ie\sqrt t\,S_{j}\rVert_{\mathrm{op}}\le2t+\tfrac83t\sqrt t\le3t.\tag{B} ∥ D j ( e ) ∥ op ≤ 2 t + 2 t + 3 8 t t ≤ ( 2 + 2 1 + 6 1 ) t ≤ 3 t , ∥ D j ( e ) − i e t S j ∥ op ≤ 2 t + 3 8 t t ≤ 3 t . ( B )
Step 2 (expansion along a word). Let 0 < t ≤ 1 16 0<t\le\frac1{16} 0 < t ≤ 16 1 and let w ∈ W 2 d w\in W_{2d} w ∈ W 2 d have length k ∈ N k\in\mathbb{N} k ∈ N . For m ∈ [ k ] m\in[k] m ∈ [ k ] put j m = g ( w m ) j_{m}=g(w_{m}) j m = g ( w m ) , ε m = ε ( w m ) \varepsilon_{m}=\varepsilon(w_{m}) ε m = ε ( w m ) , A m = U ( w m ) A_{m}=U^{(w_{m})} A m = U ( w m ) , Y m = i ε m t S j m Y_{m}=i\varepsilon_{m}\sqrt t\,S_{j_{m}} Y m = i ε m t S j m , and
B m = A m D j m if ε m = 1 , B m = D j m ∗ A m if ε m = − 1. B_{m}=A_{m}D_{j_{m}}\ \text{ if }\varepsilon_{m}=1,\qquad B_{m}=D_{j_{m}}^{*}A_{m}\ \text{ if }\varepsilon_{m}=-1 . B m = A m D j m if ε m = 1 , B m = D j m ∗ A m if ε m = − 1.
Then V ( w m ) = A m + B m V^{(w_{m})}=A_{m}+B_{m} V ( w m ) = A m + B m : if ε m = 1 \varepsilon_{m}=1 ε m = 1 , V ( w m ) = U j m C j m = A m ( I + D j m ) V^{(w_{m})}=U_{j_{m}}C_{j_{m}}=A_{m}(I+D_{j_{m}}) V ( w m ) = U j m C j m = A m ( I + D j m ) ; if ε m = − 1 \varepsilon_{m}=-1 ε m = − 1 , V ( w m ) = ( U j m C j m ) ∗ = C j m ∗ U j m ∗ = ( I + D j m ∗ ) A m V^{(w_{m})}=(U_{j_{m}}C_{j_{m}})^{*}=C_{j_{m}}^{*}U_{j_{m}}^{*}=(I+D_{j_{m}}^{*})A_{m} V ( w m ) = ( U j m C j m ) ∗ = C j m ∗ U j m ∗ = ( I + D j m ∗ ) A m (Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-calculus ). Write B m = G m + E m B_{m}=G_{m}+E_{m} B m = G m + E m , where G m = A m Y m G_{m}=A_{m}Y_{m} G m = A m Y m if ε m = 1 \varepsilon_{m}=1 ε m = 1 and G m = Y m A m G_{m}=Y_{m}A_{m} G m = Y m A m if ε m = − 1 \varepsilon_{m}=-1 ε m = − 1 . By (O3), (B) and Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §operations ,
∥ B m ∥ o p ≤ 3 t , ∥ G m ∥ o p ≤ 2 t , ∥ E m ∥ o p ≤ 3 t . \lVert B_{m}\rVert_{\mathrm{op}}\le3\sqrt t,\qquad\lVert G_{m}\rVert_{\mathrm{op}}\le2\sqrt t,\qquad\lVert E_{m}\rVert_{\mathrm{op}}\le3t . ∥ B m ∥ op ≤ 3 t , ∥ G m ∥ op ≤ 2 t , ∥ E m ∥ op ≤ 3 t .
For J ⊆ [ k ] J\subseteq[k] J ⊆ [ k ] let Π J = F 1 J F 2 J ⋯ F k J \Pi^{J}=F^{J}_{1}F^{J}_{2}\cdots F^{J}_{k} Π J = F 1 J F 2 J ⋯ F k J with F m J = B m F^{J}_{m}=B_{m} F m J = B m for m ∈ J m\in J m ∈ J and F m J = A m F^{J}_{m}=A_{m} F m J = A m for m ∉ J m\notin J m ∈ / J . Multiplying out the product V ( w ) = ( A 1 + B 1 ) ⋯ ( A k + B k ) V^{(w)}=(A_{1}+B_{1})\cdots(A_{k}+B_{k}) V ( w ) = ( A 1 + B 1 ) ⋯ ( A k + B k ) by distributivity of composition over sums (Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §operations ; induction on k k k ) gives V ( w ) = ∑ J ⊆ [ k ] Π J V^{(w)}=\sum_{J\subseteq[k]}\Pi^{J} V ( w ) = ∑ J ⊆ [ k ] Π J , a sum over the 2 k 2^{k} 2 k subsets of [ k ] [k] [ k ] (their number, by induction on k k k ). Since all A m A_{m} A m have norm at most 1 1 1 , ∥ Π J ∥ o p ≤ ∏ m ∈ J ∥ B m ∥ o p \lVert\Pi^{J}\rVert_{\mathrm{op}}\le\prod_{m\in J}\lVert B_{m}\rVert_{\mathrm{op}} ∥ Π J ∥ op ≤ ∏ m ∈ J ∥ B m ∥ op .
Step 3 (the terms). (i) J = ∅ J=\varnothing J = ∅ : Π ∅ = A 1 ⋯ A k = U ( w ) \Pi^{\varnothing}=A_{1}\cdots A_{k}=U^{(w)} Π ∅ = A 1 ⋯ A k = U ( w ) by (O4) and induction on the number of factors, and τ ( Π ∅ ) = λ U ( w ) = λ ( w ) \tau(\Pi^{\varnothing})=\lambda_{U}(w)=\lambda(w) τ ( Π ∅ ) = λ U ( w ) = λ ( w ) .
(ii) J = { m } J=\{m\} J = { m } . Let P = A 1 ⋯ A m P=A_{1}\cdots A_{m} P = A 1 ⋯ A m and Q = A m + 1 ⋯ A k Q=A_{m+1}\cdots A_{k} Q = A m + 1 ⋯ A k if ε m = 1 \varepsilon_{m}=1 ε m = 1 , and P = A 1 ⋯ A m − 1 P=A_{1}\cdots A_{m-1} P = A 1 ⋯ A m − 1 and Q = A m ⋯ A k Q=A_{m}\cdots A_{k} Q = A m ⋯ A k if ε m = − 1 \varepsilon_{m}=-1 ε m = − 1 (empty products being I I I ). By (O4) and induction on the number of factors, P = U ( u ) P=U^{(u)} P = U ( u ) and Q = U ( u ′ ) Q=U^{(u')} Q = U ( u ′ ) for words with u u ′ = w uu'=w u u ′ = w , P Q = U ( w ) PQ=U^{(w)} PQ = U ( w ) , and Q P = U ( u ′ u ) ∈ A U QP=U^{(u'u)}\in\mathcal{A}_{U} QP = U ( u ′ u ) ∈ A U . By (c), Π { m } = P ( Y m − t 2 S j m 2 + R ) Q \Pi^{\{m\}}=P\bigl(Y_{m}-\frac t2S_{j_{m}}^{2}+R\bigr)Q Π { m } = P ( Y m − 2 t S j m 2 + R ) Q with R = R j m R=R_{j_{m}} R = R j m or R = R j m ∗ R=R_{j_{m}}^{*} R = R j m ∗ . By (O2), Unitary Laws Are the Laws of Unitary Tuples, and Are Realised Together with a Free Semicircular Family §centred with Z = Q P Z=QP Z = QP , and Unitary Laws Are the Laws of Unitary Tuples, and Are Realised Together with a Free Semicircular Family §covariance with i = l = j m i=l=j_{m} i = l = j m , Z = I = U ( ∅ ) Z=I=U^{(\varnothing)} Z = I = U ( ∅ ) and Z ′ = Q P Z'=QP Z ′ = QP :
τ ( P Y m Q ) = i ε m t τ ( S j m Q P ) = 0 , τ ( P S j m 2 Q ) = τ ( S j m I S j m Q P ) = τ ( I ) τ ( Q P ) = τ ( P Q ) = λ ( w ) , \tau(PY_{m}Q)=i\varepsilon_{m}\sqrt t\,\tau(S_{j_{m}}QP)=0,\qquad\tau(PS_{j_{m}}^{2}Q)=\tau(S_{j_{m}}IS_{j_{m}}QP)=\tau(I)\tau(QP)=\tau(PQ)=\lambda(w), τ ( P Y m Q ) = i ε m t τ ( S j m QP ) = 0 , τ ( P S j m 2 Q ) = τ ( S j m I S j m QP ) = τ ( I ) τ ( QP ) = τ ( PQ ) = λ ( w ) ,
and ∣ τ ( P R Q ) ∣ ≤ ∥ P R Q ∥ o p ≤ ∥ P ∥ o p ∥ R ∥ o p ∥ Q ∥ o p ≤ ∥ R ∥ o p ≤ 8 3 t t |\tau(PRQ)|\le\lVert PRQ\rVert_{\mathrm{op}}\le\lVert P\rVert_{\mathrm{op}}\lVert R\rVert_{\mathrm{op}}\lVert Q\rVert_{\mathrm{op}}\le\lVert R\rVert_{\mathrm{op}}\le\frac83t\sqrt t ∣ τ ( PRQ ) ∣ ≤ ∥ PRQ ∥ op ≤ ∥ P ∥ op ∥ R ∥ op ∥ Q ∥ op ≤ ∥ R ∥ op ≤ 3 8 t t by (O1) and submultiplicativity (Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §operations ), since P P P and Q Q Q are products of the A n A_{n} A n , each of operator norm at most 1 1 1 by (O3) (an empty product I I I having norm at most 1 1 1 by the same clause), so that ∥ P ∥ o p ≤ 1 \lVert P\rVert_{\mathrm{op}}\le1 ∥ P ∥ op ≤ 1 and ∥ Q ∥ o p ≤ 1 \lVert Q\rVert_{\mathrm{op}}\le1 ∥ Q ∥ op ≤ 1 by induction on the number of factors. Hence τ ( Π { m } ) = − t 2 λ ( w ) + e { m } \tau(\Pi^{\{m\}})=-\frac t2\lambda(w)+e_{\{m\}} τ ( Π { m } ) = − 2 t λ ( w ) + e { m } with ∣ e { m } ∣ ≤ 3 t t |e_{\{m\}}|\le3t\sqrt t ∣ e { m } ∣ ≤ 3 t t .
(iii) J = { p , q } J=\{p,q\} J = { p , q } with p < q p<q p < q . Let Π G J \Pi^{J}_{G} Π G J be Π J \Pi^{J} Π J with B p , B q B_{p},B_{q} B p , B q replaced by G p , G q G_{p},G_{q} G p , G q . By distributivity, Π J − Π G J \Pi^{J}-\Pi^{J}_{G} Π J − Π G J is the sum of the product with B p B_{p} B p and E q E_{q} E q in the places p , q p,q p , q and of the product with E p E_{p} E p and G q G_{q} G q there, so
∥ Π J − Π G J ∥ o p ≤ 3 t ⋅ 3 t + 3 t ⋅ 2 t = 15 t t . \lVert\Pi^{J}-\Pi^{J}_{G}\rVert_{\mathrm{op}}\le3\sqrt t\cdot3t+3t\cdot2\sqrt t=15\,t\sqrt t . ∥ Π J − Π G J ∥ op ≤ 3 t ⋅ 3 t + 3 t ⋅ 2 t = 15 t t .
Let σ p , τ q \sigma_{p},\tau_{q} σ p , τ q be defined by the same formulas as in The Free Unitary Heat Generator on Unitary Laws §positions , which involve only p p p , q q q and the signs ε p , ε q \varepsilon_{p},\varepsilon_{q} ε p , ε q and not the generators g ( w p ) , g ( w q ) g(w_{p}),g(w_{q}) g ( w p ) , g ( w q ) ; thus σ p ∈ { p , p + 1 } \sigma_{p}\in\{p,p+1\} σ p ∈ { p , p + 1 } and τ q ∈ { q − 1 , q } \tau_{q}\in\{q-1,q\} τ q ∈ { q − 1 , q } . Since 1 ≤ p < q ≤ k 1\le p<q\le k 1 ≤ p < q ≤ k , we get 1 ≤ p ≤ σ p ≤ p + 1 ≤ q ≤ τ q + 1 1\le p\le\sigma_{p}\le p+1\le q\le\tau_{q}+1 1 ≤ p ≤ σ p ≤ p + 1 ≤ q ≤ τ q + 1 and 1 ≤ q − 1 ≤ τ q ≤ q ≤ k 1\le q-1\le\tau_{q}\le q\le k 1 ≤ q − 1 ≤ τ q ≤ q ≤ k , so σ p , τ q ∈ [ k ] \sigma_{p},\tau_{q}\in[k] σ p , τ q ∈ [ k ] and σ p ≤ q ≤ τ q + 1 \sigma_{p}\le q\le\tau_{q}+1 σ p ≤ q ≤ τ q + 1 . In Π G J \Pi^{J}_{G} Π G J the factor Y p Y_{p} Y p stands right after A p A_{p} A p if ε p = 1 \varepsilon_{p}=1 ε p = 1 and right before A p A_{p} A p if ε p = − 1 \varepsilon_{p}=-1 ε p = − 1 ; in both cases the factors A n A_{n} A n to its left are those with n < σ p n<\sigma_{p} n < σ p and those to its right begin with n = σ p n=\sigma_{p} n = σ p . Likewise Y q Y_{q} Y q stands right after A q A_{q} A q if ε q = 1 \varepsilon_{q}=1 ε q = 1 (τ q = q \tau_{q}=q τ q = q ) and right before A q A_{q} A q if ε q = − 1 \varepsilon_{q}=-1 ε q = − 1 (τ q = q − 1 \tau_{q}=q-1 τ q = q − 1 ), so the factors A n A_{n} A n to its left end with n = τ q n=\tau_{q} n = τ q and those to its right are those with n > τ q n>\tau_{q} n > τ q . By (O4) and induction on the number of factors, and the definition of w [ a , b ] w_{[a,b]} w [ a , b ] in The Free Unitary Heat Generator on Unitary Laws ,
Π G J = U ( a ) Y p U ( y ) Y q U ( b ) , a = w [ 1 , σ p − 1 ] , y = w [ σ p , τ q ] , b = w [ τ q + 1 , k ] , \Pi^{J}_{G}=U^{(a)}\,Y_{p}\,U^{(y)}\,Y_{q}\,U^{(b)},\qquad a=w_{[1,\sigma_{p}-1]},\quad y=w_{[\sigma_{p},\tau_{q}]},\quad b=w_{[\tau_{q}+1,k]}, Π G J = U ( a ) Y p U ( y ) Y q U ( b ) , a = w [ 1 , σ p − 1 ] , y = w [ σ p , τ q ] , b = w [ τ q + 1 , k ] ,
with a = ∅ a=\varnothing a = ∅ if σ p = 1 \sigma_{p}=1 σ p = 1 , b = ∅ b=\varnothing b = ∅ if τ q = k \tau_{q}=k τ q = k , and y = ∅ y=\varnothing y = ∅ (so U ( y ) = I U^{(y)}=I U ( y ) = I ) if τ q < σ p \tau_{q}<\sigma_{p} τ q < σ p ; when ( p , q ) ∈ P ( w ) (p,q)\in P(w) ( p , q ) ∈ P ( w ) , y = y p q ( w ) y=y_{pq}(w) y = y pq ( w ) and b a = x p q ( w ) ba=x_{pq}(w) ba = x pq ( w ) by The Free Unitary Heat Generator on Unitary Laws §pieces . By (O2), (O4) and Unitary Laws Are the Laws of Unitary Tuples, and Are Realised Together with a Free Semicircular Family §covariance with Z = U ( y ) Z=U^{(y)} Z = U ( y ) and Z ′ = U ( b a ) Z'=U^{(ba)} Z ′ = U ( ba ) ,
τ ( Π G J ) = ( i ε p t ) ( i ε q t ) τ ( S j p U ( y ) S j q U ( b a ) ) = − ε p ε q t δ j p j q λ ( y ) λ ( b a ) . \tau(\Pi^{J}_{G})=(i\varepsilon_{p}\sqrt t)(i\varepsilon_{q}\sqrt t)\,\tau\bigl(S_{j_{p}}U^{(y)}S_{j_{q}}U^{(ba)}\bigr)=-\varepsilon_{p}\varepsilon_{q}\,t\,\delta_{j_{p}j_{q}}\,\lambda(y)\,\lambda(ba). τ ( Π G J ) = ( i ε p t ) ( i ε q t ) τ ( S j p U ( y ) S j q U ( ba ) ) = − ε p ε q t δ j p j q λ ( y ) λ ( ba ) .
So τ ( Π J ) = − t ε p ε q λ ( y p q ( w ) ) λ ( x p q ( w ) ) + e J \tau(\Pi^{J})=-t\,\varepsilon_{p}\varepsilon_{q}\lambda(y_{pq}(w))\lambda(x_{pq}(w))+e_{J} τ ( Π J ) = − t ε p ε q λ ( y pq ( w )) λ ( x pq ( w )) + e J if ( p , q ) ∈ P ( w ) (p,q)\in P(w) ( p , q ) ∈ P ( w ) , and τ ( Π J ) = e J \tau(\Pi^{J})=e_{J} τ ( Π J ) = e J if g ( w p ) ≠ g ( w q ) g(w_{p})\neq g(w_{q}) g ( w p ) = g ( w q ) , where ∣ e J ∣ ≤ 15 t t |e_{J}|\le15t\sqrt t ∣ e J ∣ ≤ 15 t t by (O1).
(iv) J J J with at least three elements: as 3 t ≤ 3 4 ≤ 1 3\sqrt t\le\frac34\le1 3 t ≤ 4 3 ≤ 1 , ∥ Π J ∥ o p ≤ ( 3 t ) 3 = 27 t t \lVert\Pi^{J}\rVert_{\mathrm{op}}\le(3\sqrt t)^{3}=27t\sqrt t ∥ Π J ∥ op ≤ ( 3 t ) 3 = 27 t t , so τ ( Π J ) = e J \tau(\Pi^{J})=e_{J} τ ( Π J ) = e J with ∣ e J ∣ ≤ 27 t t |e_{J}|\le27t\sqrt t ∣ e J ∣ ≤ 27 t t by (O1).
Since γ ( t ) ( w ) = τ ( V ( w ) ) = ∑ J ⊆ [ k ] τ ( Π J ) \gamma(t)(w)=\tau(V^{(w)})=\sum_{J\subseteq[k]}\tau(\Pi^{J}) γ ( t ) ( w ) = τ ( V ( w ) ) = ∑ J ⊆ [ k ] τ ( Π J ) by the linearity of τ \tau τ , and the pairs J = { p , q } J=\{p,q\} J = { p , q } with g ( w p ) = g ( w q ) g(w_{p})=g(w_{q}) g ( w p ) = g ( w q ) correspond to the elements ( p , q ) (p,q) ( p , q ) of P ( w ) P(w) P ( w ) , the main terms of (i)-(iii) add up to λ ( w ) − k 2 t λ ( w ) − t ∑ ( p , q ) ∈ P ( w ) ε ( w p ) ε ( w q ) λ ( y p q ( w ) ) λ ( x p q ( w ) ) = λ ( w ) + t Θ λ ( w ) \lambda(w)-\frac{k}{2}t\lambda(w)-t\sum_{(p,q)\in P(w)}\varepsilon(w_{p})\varepsilon(w_{q})\lambda(y_{pq}(w))\lambda(x_{pq}(w))=\lambda(w)+t\,\Theta\lambda(w) λ ( w ) − 2 k t λ ( w ) − t ∑ ( p , q ) ∈ P ( w ) ε ( w p ) ε ( w q ) λ ( y pq ( w )) λ ( x pq ( w )) = λ ( w ) + t Θ λ ( w ) by The Free Unitary Heat Generator on Unitary Laws §generator . Every nonempty J J J has ∣ e J ∣ ≤ 27 t t |e_{J}|\le27t\sqrt t ∣ e J ∣ ≤ 27 t t and there are fewer than 2 k 2^{k} 2 k of them, so by Sums over Finite Index Sets: Finite Unions, Disjoint Unions, Vanishing Terms, Dependent Pairs, Conjugation and the Modulus §modulus
∣ γ ( t ) ( w ) − λ ( w ) − t Θ λ ( w ) ∣ ≤ 27 ⋅ 2 k t t ( 0 < t ≤ 1 16 , w of length k ) . ( ∗ ) \bigl|\gamma(t)(w)-\lambda(w)-t\,\Theta\lambda(w)\bigr|\le27\cdot2^{k}\,t\sqrt t\qquad(0<t\le\tfrac1{16},\ w\text{ of length }k).\tag{$\ast$} γ ( t ) ( w ) − λ ( w ) − t Θ λ ( w ) ≤ 27 ⋅ 2 k t t ( 0 < t ≤ 16 1 , w of length k ) . ( ∗ )
Step 4 (the inward tangent vector). Put s 0 = 1 16 s_{0}=\frac1{16} s 0 = 16 1 ; by Step 1(b), γ \gamma γ maps { s : 0 ≤ s ≤ s 0 } \{s:0\le s\le s_{0}\} { s : 0 ≤ s ≤ s 0 } into L d \mathcal{L}_{d} L d with γ ( 0 ) = λ \gamma(0)=\lambda γ ( 0 ) = λ . For 0 < s ≤ s 0 0<s\le s_{0} 0 < s ≤ s 0 let v s = 1 s ( ι d ( γ ( s ) ) − ι d ( λ ) ) − Θ ^ λ ∈ E d v_{s}=\frac1s\bigl(\iota_{d}(\gamma(s))-\iota_{d}(\lambda)\bigr)-\widehat{\Theta}\lambda\in E_{d} v s = s 1 ( ι d ( γ ( s )) − ι d ( λ ) ) − Θ λ ∈ E d (The Word Gauge: Weights, the Space of Weighted Square-Summable Word Functions, Its Norm and Inner Product, and the Length-Weighted Gauge §space , The Embedding of Unitary Laws into the Word Gauge Space and the Gauge Distance §embedding , The Free Unitary Heat Generator: Bound, Continuity and Strict Dissipation in the Length-Weighted Gauge §bound ); its value at w ∈ W d ∘ w\in W^{\circ}_{d} w ∈ W d ∘ is v s ( w ) = 1 s ( γ ( s ) ( w ) − λ ( w ) ) − Θ λ ( w ) v_{s}(w)=\frac1s(\gamma(s)(w)-\lambda(w))-\Theta\lambda(w) v s ( w ) = s 1 ( γ ( s ) ( w ) − λ ( w )) − Θ λ ( w ) , the operations being pointwise. At w = ∅ w=\varnothing w = ∅ , v s ( ∅ ) = 0 v_{s}(\varnothing)=0 v s ( ∅ ) = 0 , since γ ( s ) ( ∅ ) = λ ( ∅ ) = 1 \gamma(s)(\varnothing)=\lambda(\varnothing)=1 γ ( s ) ( ∅ ) = λ ( ∅ ) = 1 (Laws of d-Tuples of Unitaries §normalised ) and Θ λ ( ∅ ) = 0 \Theta\lambda(\varnothing)=0 Θ λ ( ∅ ) = 0 . For w ∈ W d , k ∘ w\in W^{\circ}_{d,k} w ∈ W d , k ∘ , (∗ \ast ∗ ) gives ∣ v s ( w ) ∣ ≤ 27 ⋅ 2 k s |v_{s}(w)|\le27\cdot2^{k}\sqrt s ∣ v s ( w ) ∣ ≤ 27 ⋅ 2 k s , so c w ∣ v s ( w ) ∣ 2 ≤ 729 s ( 4 θ d ) k c_{w}|v_{s}(w)|^{2}\le729\,s\,(4\theta_{d})^{k} c w ∣ v s ( w ) ∣ 2 ≤ 729 s ( 4 θ d ) k by Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field and claim 3 of Properties of Natural Number Powers in a Field . By Counting Cyclically Reduced Words: Finiteness and the Bound (2d)^k §count and claim 3 of Properties of Natural Number Powers in a Field , the k k k -th block sum satisfies
0 ≤ ∑ w ∈ W d , k ∘ c w ∣ v s ( w ) ∣ 2 ≤ 729 s ( 8 d θ d ) k = 729 s ( 1 768 ) k , 0\le\sum_{w\in W^{\circ}_{d,k}}c_{w}|v_{s}(w)|^{2}\le729\,s\,(8d\,\theta_{d})^{k}=729\,s\,\bigl(\tfrac1{768}\bigr)^{k}, 0 ≤ w ∈ W d , k ∘ ∑ c w ∣ v s ( w ) ∣ 2 ≤ 729 s ( 8 d θ d ) k = 729 s ( 768 1 ) k ,
as θ d = 1 6144 d \theta_{d}=\frac1{6144d} θ d = 6144 d 1 (The Word Gauge: Weights, the Space of Weighted Square-Summable Word Functions, Its Norm and Inner Product, and the Length-Weighted Gauge §weights ). By Series of Nonnegative Real Numbers, Comparison, and the Geometric Series §geometric and Elementary Properties of Series of Real Numbers §linearity , ∑ k = 1 ∞ 729 s ( 1 768 ) k = 729 s ⋅ 1 / 768 1 − 1 / 768 = 729 767 s ≤ s \sum_{k=1}^{\infty}729\,s\,(\frac1{768})^{k}=729\,s\cdot\frac{1/768}{1-1/768}=\frac{729}{767}s\le s ∑ k = 1 ∞ 729 s ( 768 1 ) k = 729 s ⋅ 1 − 1/768 1/768 = 767 729 s ≤ s , so by Series of Nonnegative Real Numbers, Comparison, and the Geometric Series §comparison and The Word Gauge: Weights, the Space of Weighted Square-Summable Word Functions, Its Norm and Inner Product, and the Length-Weighted Gauge §norm , ∥ v s ∥ d 2 = 0 + ∑ k = 1 ∞ ∑ w ∈ W d , k ∘ c w ∣ v s ( w ) ∣ 2 ≤ s \lVert v_{s}\rVert_{d}^{2}=0+\sum_{k=1}^{\infty}\sum_{w\in W^{\circ}_{d,k}}c_{w}|v_{s}(w)|^{2}\le s ∥ v s ∥ d 2 = 0 + ∑ k = 1 ∞ ∑ w ∈ W d , k ∘ c w ∣ v s ( w ) ∣ 2 ≤ s .
Now let ε > 0 \varepsilon>0 ε > 0 be given, and choose δ = ε 2 > 0 \delta=\varepsilon^{2}>0 δ = ε 2 > 0 . For every real s s s with 0 < s ≤ s 0 0<s\le s_{0} 0 < s ≤ s 0 and s < δ s<\delta s < δ we get ∥ v s ∥ d 2 ≤ s < ε 2 \lVert v_{s}\rVert_{d}^{2}\le s<\varepsilon^{2} ∥ v s ∥ d 2 ≤ s < ε 2 , hence ∥ v s ∥ d ≤ ε \lVert v_{s}\rVert_{d}\le\varepsilon ∥ v s ∥ d ≤ ε by claim 2 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field . By Inward and Two-Sided Tangent Vectors to the Space of Unitary Laws §inward , Θ ^ λ ∈ T λ + \widehat{\Theta}\lambda\in T^{+}_{\lambda} Θ λ ∈ T λ + . ■ \blacksquare ■