Throughout write
y + = Δ − 1 4 χ , y − = − Δ + 1 4 χ , p + = Γ − 2 2 d ~ , p − = − Γ + 2 2 d ~ , y_{+}=\frac{\Delta-1}{4\chi},\qquad y_{-}=-\frac{\Delta+1}{4\chi},\qquad p_{+}=\frac{\Gamma-2}{2\tilde{d}},\qquad p_{-}=-\frac{\Gamma+2}{2\tilde{d}} , y + = 4 χ Δ − 1 , y − = − 4 χ Δ + 1 , p + = 2 d ~ Γ − 2 , p − = − 2 d ~ Γ + 2 ,
and N = n n ⊤ N=\mathsf{n}\mathsf{n}^{\top} N = n n ⊤ . Since χ > 0 \chi>0 χ > 0 and ψ > 0 \psi>0 ψ > 0 we have 1 + χ ψ > 1 1+\chi\psi>1 1 + χ ψ > 1 , so Δ > 1 \Delta>1 Δ > 1 , y + > 0 > y − y_{+}>0>y_{-} y + > 0 > y − and ρ = ( Δ − 1 ) / ( Δ + 1 ) ∈ ( 0 , 1 ) \rho=(\Delta-1)/(\Delta+1)\in(0,1) ρ = ( Δ − 1 ) / ( Δ + 1 ) ∈ ( 0 , 1 ) ; since d ~ > 0 \tilde{d}>0 d ~ > 0 we have Γ > 2 \Gamma>2 Γ > 2 , p + > 0 > p − p_{+}>0>p_{-} p + > 0 > p − and ρ Π ∈ ( 0 , 1 ) \rho_{\Pi}\in(0,1) ρ Π ∈ ( 0 , 1 ) . Note y + = − ρ y − y_{+}=-\rho\,y_{-} y + = − ρ y − and p + = − ρ Π p − p_{+}=-\rho_{\Pi}\,p_{-} p + = − ρ Π p − , and, from Γ 2 − 4 = 2 d ~ \Gamma^{2}-4=2\tilde{d} Γ 2 − 4 = 2 d ~ , that p + = ( Γ − 2 ) / ( ( Γ − 2 ) ( Γ + 2 ) ) = 1 / ( Γ + 2 ) p_{+}=(\Gamma-2)/\bigl((\Gamma-2)(\Gamma+2)\bigr)=1/(\Gamma+2) p + = ( Γ − 2 ) / ( ( Γ − 2 ) ( Γ + 2 ) ) = 1/ ( Γ + 2 ) . Two factorisations will be used: by the quadratic formula the roots of 8 χ y 2 + 4 y − 1 2 ψ 8\chi y^{2}+4y-\tfrac12\psi 8 χ y 2 + 4 y − 2 1 ψ are ( − 1 ± Δ ) / ( 4 χ ) (-1\pm\Delta)/(4\chi) ( − 1 ± Δ ) / ( 4 χ ) , that is y ± y_{\pm} y ± , so
− 4 y − 8 χ y 2 + 1 2 ψ = − 8 χ ( y − y + ) ( y − y − ) ( y ∈ R ) , -4y-8\chi y^{2}+\tfrac12\psi=-8\chi\,(y-y_{+})(y-y_{-})\qquad(y\in\mathbb{R}), − 4 y − 8 χ y 2 + 2 1 ψ = − 8 χ ( y − y + ) ( y − y − ) ( y ∈ R ) ,
and likewise the roots of 2 d ~ p 2 + 4 p − 1 2\tilde{d}\,p^{2}+4p-1 2 d ~ p 2 + 4 p − 1 are ( − 2 ± Γ ) / ( 2 d ~ ) (-2\pm\Gamma)/(2\tilde{d}) ( − 2 ± Γ ) / ( 2 d ~ ) , that is p ± p_{\pm} p ± , so
− 4 p − 2 d ~ p 2 + 1 = − 2 d ~ ( p − p + ) ( p − p − ) ( p ∈ R ) . -4p-2\tilde{d}\,p^{2}+1=-2\tilde{d}\,(p-p_{+})(p-p_{-})\qquad(p\in\mathbb{R}). − 4 p − 2 d ~ p 2 + 1 = − 2 d ~ ( p − p + ) ( p − p − ) ( p ∈ R ) .
For a continuous integrand on a compact interval the Riemann and Lebesgue integrals agree, by Restricted Lebesgue Measure and Integral Toolkit on a Compact Interval , so the integral equations below may be read in either sense.
In the two instantiations below, the symbols l l l and k k k are the dimension parameters of the cited Riccati results and are set to 1 1 1 ; they are unrelated to the number l = 2 l=2 l = 2 of states of the present model; that model parameter is left as l l l rather than renamed. Claim 1. Existence and uniqueness. Apply The Control Riccati Equation with Cross Term: Global Existence and Uniqueness, Positive Semidefiniteness, and the Lyapunov Bound with l = k = 1 l=k=1 l = k = 1 , horizon T T T , and the constant data A ( t ) = − 2 A(t)=-2 A ( t ) = − 2 , B ( t ) = 2 B(t)=\sqrt{2} B ( t ) = 2 , Q ( t ) = 1 2 ψ Q(t)=\tfrac12\psi Q ( t ) = 2 1 ψ , V ( t ) = 0 V(t)=0 V ( t ) = 0 , R ( t ) = ( 4 χ ) − 1 R(t)=(4\chi)^{-1} R ( t ) = ( 4 χ ) − 1 and F = 0 F=0 F = 0 . All entries are continuous; Q ( t ) = 1 2 ψ ≥ 0 Q(t)=\tfrac12\psi\ge0 Q ( t ) = 2 1 ψ ≥ 0 is symmetric, R ( t ) > 0 R(t)>0 R ( t ) > 0 is symmetric positive definite, F = 0 F=0 F = 0 is symmetric positive semidefinite, and Q ( t ) − V ( t ) R ( t ) − 1 V ( t ) ⊤ = 1 2 ψ ≥ 0 Q(t)-V(t)R(t)^{-1}V(t)^{\top}=\tfrac12\psi\ge0 Q ( t ) − V ( t ) R ( t ) − 1 V ( t ) ⊤ = 2 1 ψ ≥ 0 . The corollary's equation reads, in this one-dimensional case,
Z ( t ) = 0 + ∫ t T ( − 2 Z ( r ) − 2 Z ( r ) − ( 2 Z ( r ) ) ⋅ 4 χ ⋅ ( 2 Z ( r ) ) + 1 2 ψ ) d r = ∫ t T ( − 4 Z ( r ) − 8 χ Z ( r ) 2 + 1 2 ψ ) d r , Z(t)=0+\int_{t}^{T}\Bigl(-2Z(r)-2Z(r)-\bigl(\sqrt{2}\,Z(r)\bigr)\cdot4\chi\cdot\bigl(\sqrt{2}\,Z(r)\bigr)+\tfrac12\psi\Bigr)dr=\int_{t}^{T}\Bigl(-4Z(r)-8\chi Z(r)^{2}+\tfrac12\psi\Bigr)dr , Z ( t ) = 0 + ∫ t T ( − 2 Z ( r ) − 2 Z ( r ) − ( 2 Z ( r ) ) ⋅ 4 χ ⋅ ( 2 Z ( r ) ) + 2 1 ψ ) d r = ∫ t T ( − 4 Z ( r ) − 8 χ Z ( r ) 2 + 2 1 ψ ) d r ,
which is the displayed equation. The corollary provides exactly one continuous solution on [ 0 , T ] [0,T] [ 0 , T ] ; call it z z z .
The closed form. Put g t = exp ( − 4 Δ ( T − t ) ) = exp ( − 4 Δ T ) exp ( 4 Δ t ) g_{t}=\exp\bigl(-4\Delta(T-t)\bigr)=\exp(-4\Delta T)\exp(4\Delta t) g t = exp ( − 4Δ ( T − t ) ) = exp ( − 4Δ T ) exp ( 4Δ t ) and z ^ t = y + ( 1 − g t ) / ( 1 + ρ g t ) \hat{z}_{t}=y_{+}(1-g_{t})/(1+\rho g_{t}) z ^ t = y + ( 1 − g t ) / ( 1 + ρ g t ) . The exponential is positive, so g t > 0 g_{t}>0 g t > 0 ; and g t ≤ 1 g_{t}\le1 g t ≤ 1 with equality exactly at t = T t=T t = T , because − 4 Δ ( T − t ) ≤ 0 -4\Delta(T-t)\le0 − 4Δ ( T − t ) ≤ 0 there and the exponential is increasing. In particular 1 + ρ g t ≥ 1 1+\rho g_{t}\ge1 1 + ρ g t ≥ 1 , so z ^ \hat{z} z ^ is well defined and continuous. By Derivative and Continuity of the Scaled Exponential Function , g g g is differentiable with g t ′ = 4 Δ g t g'_{t}=4\Delta\,g_{t} g t ′ = 4Δ g t . By the reciprocal rule Reciprocal Rule for One-Dimensional Derivatives and the sum and product rules of Derivatives of Constants, the Identity, Sums, Scalar Multiples, Products, Reciprocals, Quotients and Powers ,
z ^ ′ ( t ) = y + − g t ′ ( 1 + ρ g t ) − ( 1 − g t ) ρ g t ′ ( 1 + ρ g t ) 2 = − y + g t ′ ( 1 + ρ ) ( 1 + ρ g t ) 2 = − 4 Δ ( 1 + ρ ) y + g t ( 1 + ρ g t ) 2 , \hat{z}'(t)=y_{+}\,\frac{-g'_{t}\,(1+\rho g_{t})-(1-g_{t})\,\rho g'_{t}}{(1+\rho g_{t})^{2}}=-\,y_{+}\,\frac{g'_{t}\,(1+\rho)}{(1+\rho g_{t})^{2}}=-\,\frac{4\Delta\,(1+\rho)\,y_{+}\,g_{t}}{(1+\rho g_{t})^{2}} , z ^ ′ ( t ) = y + ( 1 + ρ g t ) 2 − g t ′ ( 1 + ρ g t ) − ( 1 − g t ) ρ g t ′ = − y + ( 1 + ρ g t ) 2 g t ′ ( 1 + ρ ) = − ( 1 + ρ g t ) 2 4Δ ( 1 + ρ ) y + g t ,
the middle step because ( 1 + ρ g ) + ρ ( 1 − g ) = 1 + ρ (1+\rho g)+\rho(1-g)=1+\rho ( 1 + ρ g ) + ρ ( 1 − g ) = 1 + ρ . On the other hand, using y − = − y + / ρ y_{-}=-y_{+}/\rho y − = − y + / ρ ,
z ^ t − y + = y + ( 1 − g t ) − ( 1 + ρ g t ) 1 + ρ g t = − y + ( 1 + ρ ) g t 1 + ρ g t , z ^ t − y − = y + ρ ( 1 − g t ) + 1 + ρ g t ρ ( 1 + ρ g t ) = y + ( 1 + ρ ) ρ ( 1 + ρ g t ) , \hat{z}_{t}-y_{+}=y_{+}\frac{(1-g_{t})-(1+\rho g_{t})}{1+\rho g_{t}}=-\,\frac{y_{+}(1+\rho)g_{t}}{1+\rho g_{t}},\qquad \hat{z}_{t}-y_{-}=y_{+}\frac{\rho(1-g_{t})+1+\rho g_{t}}{\rho\,(1+\rho g_{t})}=\frac{y_{+}(1+\rho)}{\rho\,(1+\rho g_{t})} , z ^ t − y + = y + 1 + ρ g t ( 1 − g t ) − ( 1 + ρ g t ) = − 1 + ρ g t y + ( 1 + ρ ) g t , z ^ t − y − = y + ρ ( 1 + ρ g t ) ρ ( 1 − g t ) + 1 + ρ g t = ρ ( 1 + ρ g t ) y + ( 1 + ρ ) ,
so that
− 8 χ ( z ^ t − y + ) ( z ^ t − y − ) = 8 χ y + 2 ( 1 + ρ ) 2 g t ρ ( 1 + ρ g t ) 2 . -8\chi\,(\hat{z}_{t}-y_{+})(\hat{z}_{t}-y_{-})=\frac{8\chi\,y_{+}^{2}(1+\rho)^{2}g_{t}}{\rho\,(1+\rho g_{t})^{2}} . − 8 χ ( z ^ t − y + ) ( z ^ t − y − ) = ρ ( 1 + ρ g t ) 2 8 χ y + 2 ( 1 + ρ ) 2 g t .
Since 8 χ y + = 2 ( Δ − 1 ) 8\chi y_{+}=2(\Delta-1) 8 χ y + = 2 ( Δ − 1 ) and ( 1 + ρ ) / ρ = 2 Δ / ( Δ − 1 ) (1+\rho)/\rho=2\Delta/(\Delta-1) ( 1 + ρ ) / ρ = 2Δ/ ( Δ − 1 ) , we have 8 χ y + ( 1 + ρ ) / ρ = 4 Δ 8\chi y_{+}(1+\rho)/\rho=4\Delta 8 χ y + ( 1 + ρ ) / ρ = 4Δ , and the last display equals 4 Δ ( 1 + ρ ) y + g t / ( 1 + ρ g t ) 2 = − z ^ ′ ( t ) 4\Delta(1+\rho)y_{+}g_{t}/(1+\rho g_{t})^{2}=-\hat{z}'(t) 4Δ ( 1 + ρ ) y + g t / ( 1 + ρ g t ) 2 = − z ^ ′ ( t ) . By the factorisation recorded above this says
z ^ ′ ( t ) = − ( − 4 z ^ t − 8 χ z ^ t 2 + 1 2 ψ ) ( t ∈ [ 0 , T ] ) . \hat{z}'(t)=-\Bigl(-4\hat{z}_{t}-8\chi\hat{z}_{t}^{2}+\tfrac12\psi\Bigr)\qquad(t\in[0,T]). z ^ ′ ( t ) = − ( − 4 z ^ t − 8 χ z ^ t 2 + 2 1 ψ ) ( t ∈ [ 0 , T ]) .
The right-hand side is continuous in t t t , so by Fundamental Theorem of Calculus, Part II: Evaluating an Integral by an Antiderivative applied on [ t , T ] [t,T] [ t , T ] , z ^ T − z ^ t = ∫ [ t , T ] z ^ ′ ( r ) d r \hat{z}_{T}-\hat{z}_{t}=\int_{[t,T]}\hat{z}'(r)\,dr z ^ T − z ^ t = ∫ [ t , T ] z ^ ′ ( r ) d r ; and z ^ T = 0 \hat{z}_{T}=0 z ^ T = 0 because g T = 1 g_{T}=1 g T = 1 . Rearranging gives exactly the integral equation of the claim, so z ^ = z \hat{z}=z z ^ = z by the uniqueness above.
Bounds and monotonicity. From 0 < g t ≤ 1 0<g_{t}\le1 0 < g t ≤ 1 we get 0 ≤ 1 − g t < 1 ≤ 1 + ρ g t 0\le1-g_{t}<1\le1+\rho g_{t} 0 ≤ 1 − g t < 1 ≤ 1 + ρ g t , hence 0 ≤ z t < y + 0\le z_{t}<y_{+} 0 ≤ z t < y + , with z T = 0 z_{T}=0 z T = 0 . The displayed formula for z ^ ′ \hat{z}' z ^ ′ is negative at every t t t , so z z z is nonincreasing. Finally, from the first displayed factorisation of z ^ t − y + \hat{z}_{t}-y_{+} z ^ t − y + , ρ < 1 \rho<1 ρ < 1 and 1 + ρ g t ≥ 1 1+\rho g_{t}\ge1 1 + ρ g t ≥ 1 ,
0 ≤ y + − z t = y + ( 1 + ρ ) g t 1 + ρ g t ≤ 2 y + g t = Δ − 1 2 χ exp ( − 4 Δ ( T − t ) ) . 0\le y_{+}-z_{t}=\frac{y_{+}(1+\rho)g_{t}}{1+\rho g_{t}}\le2y_{+}g_{t}=\frac{\Delta-1}{2\chi}\exp\bigl(-4\Delta(T-t)\bigr). 0 ≤ y + − z t = 1 + ρ g t y + ( 1 + ρ ) g t ≤ 2 y + g t = 2 χ Δ − 1 exp ( − 4Δ ( T − t ) ) .
Claim 2. Apply The Filter Riccati Equation: Global Existence and Uniqueness, Symmetry, and the Lyapunov Bound on [ 0 , T ] [0,T] [ 0 , T ] with k = 1 k=1 k = 1 and the constant data A ( t ) = − 2 A(t)=-2 A ( t ) = − 2 , C ( t ) = 1 C(t)=1 C ( t ) = 1 , D ( t ) = 2 d ~ D(t)=2\tilde{d} D ( t ) = 2 d ~ and P 0 = 0 P_{0}=0 P 0 = 0 : the entries are continuous, and C ( t ) C(t) C ( t ) , D ( t ) D(t) D ( t ) , P 0 P_{0} P 0 are positive semidefinite because 1 > 0 1>0 1 > 0 , 2 d ~ > 0 2\tilde{d}>0 2 d ~ > 0 and 0 ≥ 0 0\ge0 0 ≥ 0 . Its equation reads
P ( t ) = 0 + ∫ 0 t ( − 2 P ( r ) − 2 P ( r ) − 2 d ~ P ( r ) 2 + 1 ) d r = ∫ 0 t ( − 4 P ( r ) − 2 d ~ P ( r ) 2 + 1 ) d r , P(t)=0+\int_{0}^{t}\bigl(-2P(r)-2P(r)-2\tilde{d}\,P(r)^{2}+1\bigr)dr=\int_{0}^{t}\bigl(-4P(r)-2\tilde{d}P(r)^{2}+1\bigr)dr , P ( t ) = 0 + ∫ 0 t ( − 2 P ( r ) − 2 P ( r ) − 2 d ~ P ( r ) 2 + 1 ) d r = ∫ 0 t ( − 4 P ( r ) − 2 d ~ P ( r ) 2 + 1 ) d r ,
which is the displayed equation, and the theorem provides exactly one continuous solution; call it p p p .
Put h t = exp ( − 2 Γ t ) h_{t}=\exp(-2\Gamma t) h t = exp ( − 2Γ t ) and p ^ t = p + ( 1 − h t ) / ( 1 + ρ Π h t ) \hat{p}_{t}=p_{+}(1-h_{t})/(1+\rho_{\Pi}h_{t}) p ^ t = p + ( 1 − h t ) / ( 1 + ρ Π h t ) , so that 0 < h t ≤ 1 0<h_{t}\le1 0 < h t ≤ 1 with equality exactly at t = 0 t=0 t = 0 , and h t ′ = − 2 Γ h t h'_{t}=-2\Gamma h_{t} h t ′ = − 2Γ h t by Derivative and Continuity of the Scaled Exponential Function . Exactly as in claim 1,
p ^ ′ ( t ) = − h t ′ ( 1 + ρ Π ) p + ( 1 + ρ Π h t ) 2 = 2 Γ ( 1 + ρ Π ) p + h t ( 1 + ρ Π h t ) 2 , − 2 d ~ ( p ^ t − p + ) ( p ^ t − p − ) = 2 d ~ p + 2 ( 1 + ρ Π ) 2 h t ρ Π ( 1 + ρ Π h t ) 2 , \hat{p}'(t)=-\,\frac{h'_{t}(1+\rho_{\Pi})p_{+}}{(1+\rho_{\Pi}h_{t})^{2}}=\frac{2\Gamma(1+\rho_{\Pi})p_{+}h_{t}}{(1+\rho_{\Pi}h_{t})^{2}},\qquad -2\tilde{d}(\hat{p}_{t}-p_{+})(\hat{p}_{t}-p_{-})=\frac{2\tilde{d}\,p_{+}^{2}(1+\rho_{\Pi})^{2}h_{t}}{\rho_{\Pi}(1+\rho_{\Pi}h_{t})^{2}} , p ^ ′ ( t ) = − ( 1 + ρ Π h t ) 2 h t ′ ( 1 + ρ Π ) p + = ( 1 + ρ Π h t ) 2 2Γ ( 1 + ρ Π ) p + h t , − 2 d ~ ( p ^ t − p + ) ( p ^ t − p − ) = ρ Π ( 1 + ρ Π h t ) 2 2 d ~ p + 2 ( 1 + ρ Π ) 2 h t ,
using p − = − p + / ρ Π p_{-}=-p_{+}/\rho_{\Pi} p − = − p + / ρ Π . Since 2 d ~ p + = Γ − 2 2\tilde{d}\,p_{+}=\Gamma-2 2 d ~ p + = Γ − 2 and ( 1 + ρ Π ) / ρ Π = 2 Γ / ( Γ − 2 ) (1+\rho_{\Pi})/\rho_{\Pi}=2\Gamma/(\Gamma-2) ( 1 + ρ Π ) / ρ Π = 2Γ/ ( Γ − 2 ) , we get 2 d ~ p + ( 1 + ρ Π ) / ρ Π = 2 Γ 2\tilde{d}p_{+}(1+\rho_{\Pi})/\rho_{\Pi}=2\Gamma 2 d ~ p + ( 1 + ρ Π ) / ρ Π = 2Γ , so the two displayed quantities are equal; by the second factorisation, p ^ ′ ( t ) = − 4 p ^ t − 2 d ~ p ^ t 2 + 1 \hat{p}'(t)=-4\hat{p}_{t}-2\tilde{d}\hat{p}_{t}^{2}+1 p ^ ′ ( t ) = − 4 p ^ t − 2 d ~ p ^ t 2 + 1 , a continuous function of t t t . By Fundamental Theorem of Calculus, Part II: Evaluating an Integral by an Antiderivative on [ 0 , t ] [0,t] [ 0 , t ] and p ^ 0 = 0 \hat{p}_{0}=0 p ^ 0 = 0 we obtain the integral equation, so p ^ = p \hat{p}=p p ^ = p . The bounds 0 ≤ p t < p + = ( Γ + 2 ) − 1 0\le p_{t}<p_{+}=(\Gamma+2)^{-1} 0 ≤ p t < p + = ( Γ + 2 ) − 1 , the monotonicity (p ^ ′ > 0 \hat{p}'>0 p ^ ′ > 0 ) and the estimate 0 ≤ p + − p t = p + ( 1 + ρ Π ) h t / ( 1 + ρ Π h t ) ≤ 2 p + h t 0\le p_{+}-p_{t}=p_{+}(1+\rho_{\Pi})h_{t}/(1+\rho_{\Pi}h_{t})\le2p_{+}h_{t} 0 ≤ p + − p t = p + ( 1 + ρ Π ) h t / ( 1 + ρ Π h t ) ≤ 2 p + h t follow as in claim 1.
Claim 3. We use the rank-one algebra v ⊤ v = 2 v^{\top}v=2 v ⊤ v = 2 , n ⊤ v = 0 \mathsf{n}^{\top}v=0 n ⊤ v = 0 , hence
( v v ⊤ ) ( v v ⊤ ) = 2 v v ⊤ , ( v v ⊤ ) N = N ( v v ⊤ ) = 0. (vv^{\top})(vv^{\top})=2\,vv^{\top},\qquad (vv^{\top})N=N(vv^{\top})=0 . ( v v ⊤ ) ( v v ⊤ ) = 2 v v ⊤ , ( v v ⊤ ) N = N ( v v ⊤ ) = 0.
Each Z t Z_{t} Z t is symmetric because v v ⊤ vv^{\top} v v ⊤ and N N N are, and its entries are continuous in t t t because z z z is continuous and t ↦ μ ( T − t ) t\mapsto\mu(T-t) t ↦ μ ( T − t ) is. At t = T t=T t = T both terms vanish, so Z T = 0 = F ^ Z_{T}=0=\hat{F} Z T = 0 = F ^ by claim The Fluctuation LQG Data of the Ising Equilibrium and Its Joint Coercivity (ii) .
By claim The Fluctuation LQG Data of the Ising Equilibrium and Its Joint Coercivity (ii) , E r = − v v ⊤ E_{r}=-vv^{\top} E r = − v v ⊤ is symmetric, B r = − 1 2 v v ⊤ \mathsf{B}_{r}=-\tfrac12vv^{\top} B r = − 2 1 v v ⊤ , V r = 0 V_{r}=0 V r = 0 , R r − 1 = 4 χ I R_{r}^{-1}=4\chi I R r − 1 = 4 χ I and Q r = 1 2 ψ v v ⊤ + μ N Q_{r}=\tfrac12\psi\,vv^{\top}+\mu N Q r = 2 1 ψ v v ⊤ + μ N . Hence
E r ⊤ Z r + Z r E r = − ( v v ⊤ ) ( z r v v ⊤ + μ ( T − r ) N ) − ( z r v v ⊤ + μ ( T − r ) N ) ( v v ⊤ ) = − 4 z r v v ⊤ , E_{r}^{\top}Z_{r}+Z_{r}E_{r}=-\bigl(vv^{\top}\bigr)\bigl(z_{r}vv^{\top}+\mu(T-r)N\bigr)-\bigl(z_{r}vv^{\top}+\mu(T-r)N\bigr)\bigl(vv^{\top}\bigr)=-4z_{r}\,vv^{\top}, E r ⊤ Z r + Z r E r = − ( v v ⊤ ) ( z r v v ⊤ + μ ( T − r ) N ) − ( z r v v ⊤ + μ ( T − r ) N ) ( v v ⊤ ) = − 4 z r v v ⊤ ,
W r = Z r B r + 1 2 V r = − 1 2 ( z r v v ⊤ + μ ( T − r ) N ) ( v v ⊤ ) = − z r v v ⊤ , W_{r}=Z_{r}\mathsf{B}_{r}+\tfrac12V_{r}=-\tfrac12\bigl(z_{r}vv^{\top}+\mu(T-r)N\bigr)\bigl(vv^{\top}\bigr)=-z_{r}\,vv^{\top}, W r = Z r B r + 2 1 V r = − 2 1 ( z r v v ⊤ + μ ( T − r ) N ) ( v v ⊤ ) = − z r v v ⊤ ,
W r R r − 1 W r ⊤ = 4 χ z r 2 ( v v ⊤ ) ( v v ⊤ ) = 8 χ z r 2 v v ⊤ = Ξ r . W_{r}R_{r}^{-1}W_{r}^{\top}=4\chi\,z_{r}^{2}\,(vv^{\top})(vv^{\top})=8\chi\,z_{r}^{2}\,vv^{\top}=\Xi_{r} . W r R r − 1 W r ⊤ = 4 χ z r 2 ( v v ⊤ ) ( v v ⊤ ) = 8 χ z r 2 v v ⊤ = Ξ r .
Therefore
E r ⊤ Z r + Z r E r − W r R r − 1 W r ⊤ + Q r = ( − 4 z r − 8 χ z r 2 + 1 2 ψ ) v v ⊤ + μ N . E_{r}^{\top}Z_{r}+Z_{r}E_{r}-W_{r}R_{r}^{-1}W_{r}^{\top}+Q_{r}=\Bigl(-4z_{r}-8\chi z_{r}^{2}+\tfrac12\psi\Bigr)vv^{\top}+\mu N . E r ⊤ Z r + Z r E r − W r R r − 1 W r ⊤ + Q r = ( − 4 z r − 8 χ z r 2 + 2 1 ψ ) v v ⊤ + μ N .
Integrating entrywise over [ t , T ] [t,T] [ t , T ] and using claim 1 for the first coefficient and ∫ [ t , T ] μ d r = μ ( T − t ) \int_{[t,T]}\mu\,dr=\mu(T-t) ∫ [ t , T ] μ d r = μ ( T − t ) for the second gives
∫ [ t , T ] ( E r ⊤ Z r + Z r E r − W r R r − 1 W r ⊤ + Q r ) d r = z t v v ⊤ + μ ( T − t ) N = Z t = F ^ + Z t − F ^ , \int_{[t,T]}\bigl(E_{r}^{\top}Z_{r}+Z_{r}E_{r}-W_{r}R_{r}^{-1}W_{r}^{\top}+Q_{r}\bigr)dr=z_{t}\,vv^{\top}+\mu(T-t)N=Z_{t}=\hat{F}+Z_{t}-\hat{F} , ∫ [ t , T ] ( E r ⊤ Z r + Z r E r − W r R r − 1 W r ⊤ + Q r ) d r = z t v v ⊤ + μ ( T − t ) N = Z t = F ^ + Z t − F ^ ,
which, since F ^ = 0 \hat{F}=0 F ^ = 0 , is the asserted integral equation entry by entry. The integrand is continuous in r r r , so subtracting the equations at t t t and at 0 0 0 gives the equivalent forward form Z t γ δ = Z 0 γ δ + ∫ [ 0 , t ] z ˙ γ δ ( r ) d r Z^{\gamma\delta}_{t}=Z^{\gamma\delta}_{0}+\int_{[0,t]}\dot{z}^{\gamma\delta}(r)\,dr Z t γ δ = Z 0 γ δ + ∫ [ 0 , t ] z ˙ γ δ ( r ) d r with the stated continuous densities. Symmetry, continuity of the entries, the density identity and the terminal condition Z T = F ^ Z_{T}=\hat{F} Z T = F ^ are precisely what hypothesis (H2) asks for.
Claim 4. The matrices E t \mathcal{E}_{t} E t , Θ t ⋆ \Theta^{\star}_{t} Θ t ⋆ and D ~ t \tilde{D}_{t} D ~ t are constant in t t t by claims The Fluctuation LQG Data of the Ising Equilibrium and Its Joint Coercivity (iii) and The Fluctuation LQG Data of the Ising Equilibrium and Its Joint Coercivity (iv) , hence have continuous entries. For z ∈ R 2 z\in\mathbb{R}^{2} z ∈ R 2 we have z ⋅ Θ t ⋆ z = ( v ⋅ z ) 2 ≥ 0 z\cdot\Theta^{\star}_{t}z=(v\cdot z)^{2}\ge0 z ⋅ Θ t ⋆ z = ( v ⋅ z ) 2 ≥ 0 , so Θ t ⋆ \Theta^{\star}_{t} Θ t ⋆ is symmetric positive semidefinite; and, writing c = q 2 + q 0 > 0 c=\tfrac{q}{2}+q_{0}>0 c = 2 q + q 0 > 0 , D ~ t = c − 1 E ~ t ⊤ E ~ t \tilde{D}_{t}=c^{-1}\tilde{\mathcal{E}}_{t}^{\top}\tilde{\mathcal{E}}_{t} D ~ t = c − 1 E ~ t ⊤ E ~ t is symmetric with z ⋅ D ~ t z = c − 1 ∣ E ~ t z ∣ 2 ≥ 0 z\cdot\tilde{D}_{t}z=c^{-1}|\tilde{\mathcal{E}}_{t}z|^{2}\ge0 z ⋅ D ~ t z = c − 1 ∣ E ~ t z ∣ 2 ≥ 0 , hence positive semidefinite; the zero matrix is positive semidefinite. So The Filter Riccati Equation: Global Existence and Uniqueness, Symmetry, and the Lyapunov Bound applies and furnishes a unique Π \Pi Π .
Put Π ^ t = p t v v ⊤ \hat{\Pi}_{t}=p_{t}vv^{\top} Π ^ t = p t v v ⊤ , which has continuous entries. Using E t = − v v ⊤ \mathcal{E}_{t}=-vv^{\top} E t = − v v ⊤ , the rank-one algebra of claim 3 and the identity v v ⊤ D ~ t = d ~ v v ⊤ vv^{\top}\tilde{D}_{t}=\tilde{d}\,vv^{\top} v v ⊤ D ~ t = d ~ v v ⊤ of claim The Fluctuation LQG Data of the Ising Equilibrium and Its Joint Coercivity (iv) ,
E r Π ^ r + Π ^ r E r ⊤ = − 4 p r v v ⊤ , Π ^ r D ~ r Π ^ r = p r 2 d ~ ( v v ⊤ ) ( v v ⊤ ) = 2 d ~ p r 2 v v ⊤ , Θ r ⋆ = v v ⊤ , \mathcal{E}_{r}\hat{\Pi}_{r}+\hat{\Pi}_{r}\mathcal{E}_{r}^{\top}=-4p_{r}\,vv^{\top},\qquad \hat{\Pi}_{r}\tilde{D}_{r}\hat{\Pi}_{r}=p_{r}^{2}\,\tilde{d}\,(vv^{\top})(vv^{\top})=2\tilde{d}\,p_{r}^{2}\,vv^{\top},\qquad \Theta^{\star}_{r}=vv^{\top}, E r Π ^ r + Π ^ r E r ⊤ = − 4 p r v v ⊤ , Π ^ r D ~ r Π ^ r = p r 2 d ~ ( v v ⊤ ) ( v v ⊤ ) = 2 d ~ p r 2 v v ⊤ , Θ r ⋆ = v v ⊤ ,
so the integrand of the Kalman equation is ( − 4 p r − 2 d ~ p r 2 + 1 ) v v ⊤ \bigl(-4p_{r}-2\tilde{d}p_{r}^{2}+1\bigr)vv^{\top} ( − 4 p r − 2 d ~ p r 2 + 1 ) v v ⊤ . Integrating entrywise over [ 0 , t ] [0,t] [ 0 , t ] and using claim 2 gives ∫ 0 t ( ⋯ ) d r = p t v v ⊤ = Π ^ t \int_{0}^{t}(\cdots)dr=p_{t}vv^{\top}=\hat{\Pi}_{t} ∫ 0 t ( ⋯ ) d r = p t v v ⊤ = Π ^ t , and P 0 = 0 P_{0}=0 P 0 = 0 ; so Π ^ \hat{\Pi} Π ^ solves the equation. By the uniqueness in The Filter Riccati Equation: Global Existence and Uniqueness, Symmetry, and the Lyapunov Bound , Π t = p t v v ⊤ \Pi_{t}=p_{t}vv^{\top} Π t = p t v v ⊤ .
Claim 5. For real 2 × 2 2\times2 2 × 2 matrices of the form λ v v ⊤ \lambda vv^{\top} λ v v ⊤ and κ v v ⊤ \kappa vv^{\top} κ v v ⊤ the entrywise pairing is
∑ γ , δ λ v γ v δ κ v γ v δ = λ κ ( ∑ γ ( v γ ) 2 ) 2 = 4 λ κ , \sum_{\gamma,\delta}\lambda v^{\gamma}v^{\delta}\,\kappa v^{\gamma}v^{\delta}=\lambda\kappa\Bigl(\sum_{\gamma}(v^{\gamma})^{2}\Bigr)^{2}=4\lambda\kappa , γ , δ ∑ λ v γ v δ κ v γ v δ = λκ ( γ ∑ ( v γ ) 2 ) 2 = 4 λκ ,
while
∑ γ , δ n γ n δ v γ v δ = ( ∑ γ n γ v γ ) 2 = 0. \sum_{\gamma,\delta}\mathsf{n}^{\gamma}\mathsf{n}^{\delta}\,v^{\gamma}v^{\delta}=\Bigl(\sum_{\gamma}\mathsf{n}^{\gamma}v^{\gamma}\Bigr)^{2}=0 . γ , δ ∑ n γ n δ v γ v δ = ( γ ∑ n γ v γ ) 2 = 0.
With Z t = z t v v ⊤ + μ ( T − t ) N Z_{t}=z_{t}vv^{\top}+\mu(T-t)N Z t = z t v v ⊤ + μ ( T − t ) N and Θ t ⋆ = v v ⊤ \Theta^{\star}_{t}=vv^{\top} Θ t ⋆ = v v ⊤ the first pairing is 4 z t ⋅ 1 + μ ( T − t ) ⋅ 0 = 4 z t 4z_{t}\cdot1+\mu(T-t)\cdot0=4z_{t} 4 z t ⋅ 1 + μ ( T − t ) ⋅ 0 = 4 z t ; with Ξ t = 8 χ z t 2 v v ⊤ \Xi_{t}=8\chi z_{t}^{2}vv^{\top} Ξ t = 8 χ z t 2 v v ⊤ and Π t = p t v v ⊤ \Pi_{t}=p_{t}vv^{\top} Π t = p t v v ⊤ the second is 4 ⋅ 8 χ z t 2 ⋅ p t = 32 χ z t 2 p t 4\cdot8\chi z_{t}^{2}\cdot p_{t}=32\chi z_{t}^{2}p_{t} 4 ⋅ 8 χ z t 2 ⋅ p t = 32 χ z t 2 p t . Neither expression contains μ \mu μ . Both z z z and p p p are continuous by claims 1 and 2, so t ↦ 4 z t + 32 χ z t 2 p t t\mapsto4z_{t}+32\chi z_{t}^{2}p_{t} t ↦ 4 z t + 32 χ z t 2 p t is continuous by Continuity of Sums and Products of Real-Valued Functions on a Metric Space .