TheoremBase

Properness is exact; the structure condition follows from the monotonicity identity for the form operator together with two completions of squares; and on the admissible sets the quadratic term in the shifted operator bounds the norm of Ax, which yields an explicit shift modulus.

Proof

Each result cited is universally quantified over the data in its own statement. Throughout, ∣a+b∣H2=∣a∣H2+2⟨a,b⟩H+∣b∣H2|a+b|_{H}^{2}=|a|_{H}^{2}+2\langle a,b\rangle_{H}+|b|_{H}^{2} for a,b∈Ha,b\in H by Elementary Identities in a Real Inner Product Space (v), and ∣⟨a,b⟩H∣≤∣a∣H∣b∣H|\langle a,b\rangle_{H}|\le|a|_{H}|b|_{H} by The Cauchy–Schwarz Inequality in a Real Inner Product Space. We record one identity: for x,y∈D(A)x,y\in D(A),

⟨Ax−Ay, x−y⟩H=∣x−y∣V2,(1)\langle Ax-Ay,\,x-y\rangle_{H}=|x-y|_{V}^{2}, \tag{1}

since x−y∈Vx-y\in V, so that ⟨Ax,x−y⟩H=⟨x,x−y⟩V\langle Ax,x-y\rangle_{H}=\langle x,x-y\rangle_{V} and ⟨Ay,x−y⟩H=⟨y,x−y⟩V\langle Ay,x-y\rangle_{H}=\langle y,x-y\rangle_{V} by Hilbert Triples: Standing Notation and Background (iii), and subtracting gives ⟨x−y,x−y⟩V\langle x-y,x-y\rangle_{V}.

Claim 1. Each of λ0r\lambda_{0}r, 12∣p∣H2\tfrac12|p|_{H}^{2}, ⟨Ax,p⟩H\langle Ax,p\rangle_{H} and g(x)g(x) is a real number for (x,r,p,X)∈D(A)×R×H×Sym(V)(x,r,p,X)\in D(A)\times\mathbb{R}\times H\times\mathrm{Sym}(V), so FF is a function from D(A)×R×H×Sym(V)D(A)\times\mathbb{R}\times H\times\mathrm{Sym}(V) to R\mathbb{R}. Since W=D(A)∩H=D(A)W=D(A)\cap H=D(A), this is exactly the shape required by Second-Order Equation Operator on an Open Subset of a Hilbert Triple and Its δ\delta-Shifts (i). The defining expression does not involve the fourth argument, so F(x,r,p,X)=F(x,r,p,X′)F(x,r,p,X)=F(x,r,p,X') for all X,X′∈Sym(V)X,X'\in\mathrm{Sym}(V), i.e. FF is first order; degenerate ellipticity follows from A First-Order Equation Operator is Degenerate Elliptic and Its δ\delta-Shifts Ignore the Form Argument (i).

Claim 2. Let RR be positive, let x∈D(A)x\in D(A), p∈Hp\in H, X∈Sym(V)X\in\mathrm{Sym}(V) and let r,s∈Rr,s\in\mathbb{R} satisfy −R≤s≤r≤R-R\le s\le r\le R. All terms of FF other than λ0r\lambda_{0}r are unchanged when rr is replaced by ss, so

F(x,r,p,X)−F(x,s,p,X)=λ0(r−s),F(x,r,p,X)-F(x,s,p,X)=\lambda_{0}(r-s),

and in particular λ0(r−s)≤F(x,r,p,X)−F(x,s,p,X)\lambda_{0}(r-s)\le F(x,r,p,X)-F(x,s,p,X). Thus λ0\lambda_{0} is a properness constant for FF at RR, for every positive RR, and FF is locally strictly proper.

Claim 3. The function ω2\omega_{2} is a modulus of continuity, both conditions of Modulus of Continuity holding trivially. So is ω3\omega_{3}: its values 34t\tfrac34t are nonnegative for t≥0t\ge0, and for positive ϵ\epsilon every tt with 0≤t≤ϵ0\le t\le\epsilon satisfies 34t≤ϵ\tfrac34 t\le\epsilon.

Let RR be positive, let x,y∈D(A)x,y\in D(A), r∈Rr\in\mathbb{R}, X,Y∈Sym(H)X,Y\in\mathrm{Sym}(H) and let α,δ\alpha,\delta satisfy 1<α1<\alpha and 0<δ<10<\delta<1; put P=α(x−y)P=\alpha(x-y). By Second-Order Equation Operator on an Open Subset of a Hilbert Triple and Its δ\delta-Shifts (ii) and the expansion of the norm,

Fδ−(x,r,P,X)=λ0r+λ0δh(x)+12∣P∣H2+δ⟨P,Ax⟩H+δ22∣Ax∣H2+⟨Ax,P⟩H+δ∣Ax∣H2−g(x),F^{-}_{\delta}(x,r,P,X)=\lambda_{0}r+\lambda_{0}\delta h(x)+\tfrac12|P|_{H}^{2}+\delta\langle P,Ax\rangle_{H}+\tfrac{\delta^{2}}{2}|Ax|_{H}^{2}+\langle Ax,P\rangle_{H}+\delta|Ax|_{H}^{2}-g(x), Fδ+(y,r,P,Y)=λ0r−λ0δh(y)+12∣P∣H2−δ⟨P,Ay⟩H+δ22∣Ay∣H2+⟨Ay,P⟩H−δ∣Ay∣H2−g(y).F^{+}_{\delta}(y,r,P,Y)=\lambda_{0}r-\lambda_{0}\delta h(y)+\tfrac12|P|_{H}^{2}-\delta\langle P,Ay\rangle_{H}+\tfrac{\delta^{2}}{2}|Ay|_{H}^{2}+\langle Ay,P\rangle_{H}-\delta|Ay|_{H}^{2}-g(y).

Subtracting and using (1) in the form ⟨Ax,P⟩H−⟨Ay,P⟩H=α∣x−y∣V2\langle Ax,P\rangle_{H}-\langle Ay,P\rangle_{H}=\alpha|x-y|_{V}^{2},

Fδ−(x,r,P,X)−Fδ+(y,r,P,Y)=λ0δ(h(x)+h(y))+α∣x−y∣V2+Tx+Ty+g(y)−g(x),F^{-}_{\delta}(x,r,P,X)-F^{+}_{\delta}(y,r,P,Y)=\lambda_{0}\delta\bigl(h(x)+h(y)\bigr)+\alpha|x-y|_{V}^{2}+T_{x}+T_{y}+g(y)-g(x),

where

Tx=δ∣Ax∣H2+δ22∣Ax∣H2+δ⟨P,Ax⟩H,Ty=δ∣Ay∣H2−δ22∣Ay∣H2+δ⟨P,Ay⟩H.T_{x}=\delta|Ax|_{H}^{2}+\tfrac{\delta^{2}}{2}|Ax|_{H}^{2}+\delta\langle P,Ax\rangle_{H},\qquad T_{y}=\delta|Ay|_{H}^{2}-\tfrac{\delta^{2}}{2}|Ay|_{H}^{2}+\delta\langle P,Ay\rangle_{H}.

Since δ22∣Ax∣H2≥0\tfrac{\delta^{2}}{2}|Ax|_{H}^{2}\ge0 and δ⟨P,Ax⟩H≥−δ∣P∣H∣Ax∣H\delta\langle P,Ax\rangle_{H}\ge-\delta|P|_{H}|Ax|_{H}, while δ(∣Ax∣H−∣P∣H2)2≥0\delta\bigl(|Ax|_{H}-\tfrac{|P|_{H}}{2}\bigr)^{2}\ge0 gives δ∣Ax∣H2−δ∣P∣H∣Ax∣H≥−δ4∣P∣H2\delta|Ax|_{H}^{2}-\delta|P|_{H}|Ax|_{H}\ge-\tfrac{\delta}{4}|P|_{H}^{2}, we get Tx≥−δ4∣P∣H2T_{x}\ge-\tfrac{\delta}{4}|P|_{H}^{2}. Since δ≤1\delta\le1 we have δ−δ22≥δ2\delta-\tfrac{\delta^{2}}{2}\ge\tfrac{\delta}{2}, and δ2(∣Ay∣H−∣P∣H)2≥0\tfrac{\delta}{2}\bigl(|Ay|_{H}-|P|_{H}\bigr)^{2}\ge0 gives δ2∣Ay∣H2−δ∣P∣H∣Ay∣H≥−δ2∣P∣H2\tfrac{\delta}{2}|Ay|_{H}^{2}-\delta|P|_{H}|Ay|_{H}\ge-\tfrac{\delta}{2}|P|_{H}^{2}, so Ty≥−δ2∣P∣H2T_{y}\ge-\tfrac{\delta}{2}|P|_{H}^{2}. Moreover h(x)+h(y)≥0h(x)+h(y)\ge0 by The Penalty Function h=12∣⋅∣V2h=\tfrac12|\cdot|_V^2 of a Hilbert Triple: Expansion Identities, Closed Sublevel Sets, Density and Local Bounds (i) together with claim 2 of Elementary Arithmetic in an Ordered Field, α∣x−y∣V2≥0\alpha|x-y|_{V}^{2}\ge0, and g(y)−g(x)≥−∣g(x)−g(y)∣≥−ωg(∣x−y∣H)g(y)-g(x)\ge-|g(x)-g(y)|\ge-\omega_{g}\bigl(|x-y|_{H}\bigr) by hypothesis. Since ∣P∣H2=α2∣x−y∣H2|P|_{H}^{2}=\alpha^{2}|x-y|_{H}^{2}, we conclude

Fδ−(x,r,P,X)−Fδ+(y,r,P,Y) ≥ −ωg(∣x−y∣H)−34δα2∣x−y∣H2=−ωg(∣x−y∣H)−ω2(α∣x−y∣H2)−ω3(δα2∣x−y∣H2).F^{-}_{\delta}(x,r,P,X)-F^{+}_{\delta}(y,r,P,Y)\ \ge\ -\omega_{g}\bigl(|x-y|_{H}\bigr)-\tfrac{3}{4}\delta\alpha^{2}|x-y|_{H}^{2}=-\omega_{g}\bigl(|x-y|_{H}\bigr)-\omega_{2}\bigl(\alpha|x-y|_{H}^{2}\bigr)-\omega_{3}\bigl(\delta\alpha^{2}|x-y|_{H}^{2}\bigr).

No restriction on rr was used, so (ωg,ω2,ω3)(\omega_{g},\omega_{2},\omega_{3}) is a structure triple for FF at every positive RR.

Claim 4. Let δ,R\delta,R satisfy 0<δ<10<\delta<1 and 0<R0<R. Recall from Test Data for a Second-Order Equation Operator on a Hilbert Triple and the Admissible Sets (ii) that an RR-bounded datum (x,r,p,X)(x,r,p,X) satisfies h(x)<Rh(x)<R, ∣r∣<R|r|<R and ∣p∣H<R|p|_{H}<R.

(a) An upper bound for Fδ+F^{+}_{\delta}. For RR-bounded η=(y,s,p′,X′)\eta=(y,s,p',X'), expanding as above,

Fδ+(η)=λ0s−λ0δh(y)+12∣p′∣H2−δ⟨p′,Ay⟩H+δ22∣Ay∣H2+⟨Ay,p′⟩H−δ∣Ay∣H2−g(y).F^{+}_{\delta}(\eta)=\lambda_{0}s-\lambda_{0}\delta h(y)+\tfrac12|p'|_{H}^{2}-\delta\langle p',Ay\rangle_{H}+\tfrac{\delta^{2}}{2}|Ay|_{H}^{2}+\langle Ay,p'\rangle_{H}-\delta|Ay|_{H}^{2}-g(y).

Here λ0s≤λ0R\lambda_{0}s\le\lambda_{0}R, −λ0δh(y)≤0-\lambda_{0}\delta h(y)\le0, 12∣p′∣H2≤12R2\tfrac12|p'|_{H}^{2}\le\tfrac12R^{2}, −δ⟨p′,Ay⟩H≤R∣Ay∣H-\delta\langle p',Ay\rangle_{H}\le R|Ay|_{H}, ⟨Ay,p′⟩H≤R∣Ay∣H\langle Ay,p'\rangle_{H}\le R|Ay|_{H}, δ22∣Ay∣H2−δ∣Ay∣H2≤−δ2∣Ay∣H2\tfrac{\delta^{2}}{2}|Ay|_{H}^{2}-\delta|Ay|_{H}^{2}\le-\tfrac{\delta}{2}|Ay|_{H}^{2} and −g(y)≤Cg-g(y)\le C_{g}, so

Fδ+(η)≤λ0R+12R2+Cg+2R∣Ay∣H−δ2∣Ay∣H2.(2)F^{+}_{\delta}(\eta)\le\lambda_{0}R+\tfrac12R^{2}+C_{g}+2R|Ay|_{H}-\tfrac{\delta}{2}|Ay|_{H}^{2}. \tag{2}

Since δ2(∣Ay∣H−2Rδ)2≥0\tfrac{\delta}{2}\bigl(|Ay|_{H}-\tfrac{2R}{\delta}\bigr)^{2}\ge0 gives 2R∣Ay∣H−δ2∣Ay∣H2≤2R2δ2R|Ay|_{H}-\tfrac{\delta}{2}|Ay|_{H}^{2}\le\tfrac{2R^{2}}{\delta}, we obtain Fδ+(η)≤ΣF^{+}_{\delta}(\eta)\le\Sigma, where Σ=λ0R+12R2+Cg+2R2δ\Sigma=\lambda_{0}R+\tfrac12R^{2}+C_{g}+\tfrac{2R^{2}}{\delta}.

(b) A lower bound for Fδ−F^{-}_{\delta}. For RR-bounded ξ=(x,r,p,X)\xi=(x,r,p,X), in the same way,

Fδ−(ξ) ≥ −λ0R−R∣Ax∣H−R∣Ax∣H+δ∣Ax∣H2−Cg=δ∣Ax∣H2−2R∣Ax∣H−λ0R−Cg,(3)F^{-}_{\delta}(\xi)\ \ge\ -\lambda_{0}R-R|Ax|_{H}-R|Ax|_{H}+\delta|Ax|_{H}^{2}-C_{g}=\delta|Ax|_{H}^{2}-2R|Ax|_{H}-\lambda_{0}R-C_{g}, \tag{3}

using λ0δh(x)≥0\lambda_{0}\delta h(x)\ge0, 12∣p∣H2≥0\tfrac12|p|_{H}^{2}\ge0 and δ22∣Ax∣H2≥0\tfrac{\delta^{2}}{2}|Ax|_{H}^{2}\ge0. Since δ(∣Ax∣H−Rδ)2≥0\delta\bigl(|Ax|_{H}-\tfrac{R}{\delta}\bigr)^{2}\ge0 gives δ∣Ax∣H2−2R∣Ax∣H≥−R2δ\delta|Ax|_{H}^{2}-2R|Ax|_{H}\ge-\tfrac{R^{2}}{\delta}, we obtain Fδ−(ξ)≥ΞF^{-}_{\delta}(\xi)\ge\Xi, where Ξ=−R2δ−λ0R−Cg\Xi=-\tfrac{R^{2}}{\delta}-\lambda_{0}R-C_{g}.

(c) A bound on ∣Ax∣H|Ax|_{H} for admissible data. Let ξ=(x,r,p,X)∈Sδ,R−\xi=(x,r,p,X)\in S^{-}_{\delta,R}. By Test Data for a Second-Order Equation Operator on a Hilbert Triple and the Admissible Sets (iii) there is an RR-bounded η\eta with Fδ−(ξ)<R+Fδ+(η)≤R+ΣF^{-}_{\delta}(\xi)<R+F^{+}_{\delta}(\eta)\le R+\Sigma, so by (3)

δ∣Ax∣H2−2R∣Ax∣H≤K,K=R+Σ+λ0R+Cg  (≥0).\delta|Ax|_{H}^{2}-2R|Ax|_{H}\le K,\qquad K=R+\Sigma+\lambda_{0}R+C_{g}\ \ (\ge0).

As 2R∣Ax∣H≤δ2∣Ax∣H2+2R2δ2R|Ax|_{H}\le\tfrac{\delta}{2}|Ax|_{H}^{2}+\tfrac{2R^{2}}{\delta}, this gives δ2∣Ax∣H2≤K+2R2δ\tfrac{\delta}{2}|Ax|_{H}^{2}\le K+\tfrac{2R^{2}}{\delta}, so ∣Ax∣H≤Λ1|Ax|_{H}\le\Lambda_{1}, the nonnegative square root of 2δ(K+2R2δ)\tfrac{2}{\delta}\bigl(K+\tfrac{2R^{2}}{\delta}\bigr).

Let now η=(y,s,p′,X′)∈Sδ,R+\eta=(y,s,p',X')\in S^{+}_{\delta,R}. There is an RR-bounded ξ\xi with Fδ+(η)>Fδ−(ξ)−R≥Ξ−RF^{+}_{\delta}(\eta)>F^{-}_{\delta}(\xi)-R\ge\Xi-R, so by (2)

δ2∣Ay∣H2−2R∣Ay∣H≤K′,K′=λ0R+12R2+Cg−Ξ+R  (≥0),\tfrac{\delta}{2}|Ay|_{H}^{2}-2R|Ay|_{H}\le K',\qquad K'=\lambda_{0}R+\tfrac12R^{2}+C_{g}-\Xi+R\ \ (\ge0),

and since 2R∣Ay∣H≤δ4∣Ay∣H2+4R2δ2R|Ay|_{H}\le\tfrac{\delta}{4}|Ay|_{H}^{2}+\tfrac{4R^{2}}{\delta} we get ∣Ay∣H≤Λ2|Ay|_{H}\le\Lambda_{2}, the nonnegative square root of 4δ(K′+4R2δ)\tfrac{4}{\delta}\bigl(K'+\tfrac{4R^{2}}{\delta}\bigr).

(d) The shift modulus. Put Λ=Λ1+Λ2\Lambda=\Lambda_{1}+\Lambda_{2}, c=R+2Λc=R+2\Lambda and let ω\omega have value ct+12t2ct+\tfrac12t^{2} at a nonnegative tt. Then ω\omega is a modulus of continuity: its values are nonnegative, and for positive ϵ\epsilon every tt with 0≤t≤τ0\le t\le\tau, where τ\tau is the smaller of 11 and ϵc+1\tfrac{\epsilon}{c+1}, satisfies 12t2≤12t\tfrac12t^{2}\le\tfrac12 t and hence ω(t)≤(c+1)t≤ϵ\omega(t)\le(c+1)t\le\epsilon. Note also that ω\omega is nondecreasing on the nonnegative reals, since c≥0c\ge0.

Let q∈Hq\in H and Y∈Sym(H)Y\in\mathrm{Sym}(H). For ξ=(x,r,p,X)∈Sδ,R−\xi=(x,r,p,X)\in S^{-}_{\delta,R} we have Fδ−(x,r,p+q,X+Y)=Fδ−(x,r,p+q,X)F^{-}_{\delta}(x,r,p+q,X+Y)=F^{-}_{\delta}(x,r,p+q,X) by A First-Order Equation Operator is Degenerate Elliptic and Its δ\delta-Shifts Ignore the Form Argument (ii) and claim 1, and, expanding the norm,

Fδ−(x,r,p+q,X)−Fδ−(x,r,p,X)=⟨p+δAx,q⟩H+12∣q∣H2+⟨Ax,q⟩H ≤ (R+Λ+Λ)∣q∣H+12∣q∣H2=ω(∣q∣H),F^{-}_{\delta}(x,r,p+q,X)-F^{-}_{\delta}(x,r,p,X)=\langle p+\delta Ax,q\rangle_{H}+\tfrac12|q|_{H}^{2}+\langle Ax,q\rangle_{H}\ \le\ \bigl(R+\Lambda+\Lambda\bigr)|q|_{H}+\tfrac12|q|_{H}^{2}=\omega\bigl(|q|_{H}\bigr),

using ∣p∣H<R|p|_{H}<R, δ<1\delta<1 and ∣Ax∣H≤Λ1≤Λ|Ax|_{H}\le\Lambda_{1}\le\Lambda. Since ω\omega is nondecreasing, ω(∣q∣H)≤ω(∣q∣H+∥Y∥)\omega(|q|_{H})\le\omega\bigl(|q|_{H}+\lVert Y\rVert\bigr), which is the first condition of The Shift-Continuity Condition on Admissible Test Data (i).

For η=(y,s,p′,X′)∈Sδ,R+\eta=(y,s,p',X')\in S^{+}_{\delta,R} we have Fδ+(y,s,p′+q,X′+Y)=Fδ+(y,s,p′+q,X′)F^{+}_{\delta}(y,s,p'+q,X'+Y)=F^{+}_{\delta}(y,s,p'+q,X'), again by A First-Order Equation Operator is Degenerate Elliptic and Its δ\delta-Shifts Ignore the Form Argument (ii) and claim 1, and the same computation gives

Fδ+(y,s,p′+q,X′+Y)−Fδ+(y,s,p′,X′)=⟨p′−δAy,q⟩H+12∣q∣H2+⟨Ay,q⟩H ≥ −(R+Λ+Λ)∣q∣H ≥ −ω(∣q∣H+∥Y∥),F^{+}_{\delta}(y,s,p'+q,X'+Y)-F^{+}_{\delta}(y,s,p',X')=\langle p'-\delta Ay,q\rangle_{H}+\tfrac12|q|_{H}^{2}+\langle Ay,q\rangle_{H}\ \ge\ -\bigl(R+\Lambda+\Lambda\bigr)|q|_{H}\ \ge\ -\omega\bigl(|q|_{H}+\lVert Y\rVert\bigr),

using 12∣q∣H2≥0\tfrac12|q|_{H}^{2}\ge0 and ∣Ay∣H≤Λ2≤Λ|Ay|_{H}\le\Lambda_{2}\le\Lambda. Hence ω\omega is a shift modulus for FF at (δ,R)(\delta,R), and as δ\delta and RR were arbitrary, FF satisfies the shift-continuity condition.

Claim 5. By claims 1 to 4 the operator FF satisfies all hypotheses on the operator in A Comparison Principle on a Hilbert Triple under the First-Order Structure Condition, and uu, vv, CC satisfy the remaining ones, so A Comparison Principle on a Hilbert Triple under the First-Order Structure Condition (ii) gives u(x)≤v(x)u(x)\le v(x) for every x∈Vx\in V.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…