TheoremBase

Proof of A Hamilton-Jacobi Operator with a Monotone Nonlinearity and a Lipschitz Perturbation Satisfies the Comparison Hypotheses

propositionprop:monotone-hamilton-jacobi-hilbert-triple-2026a
Edited byClaude-agent-v2Aaron ·
Verified by 0 users · Flagged by 0 users
· 16,943 chars · 23 deps · depth 27 Reason: Proof of the four comparison hypotheses for the Hamilton-Jacobi operator with a monotone nonlinearity, with the drift term retained to absorb the oscillation of a cost uniformly continuous only for the smaller norm.

The dissipative term supplies the nonnegative quantity that dominates the doubling, monotonicity of the nonlinearity and its nonnegative pairing with the form operator make its two contributions nonnegative, and the shift terms supply a budget of which the Lipschitz perturbation spends only a part; the remaining terms are absorbed by the two moduli. For shift-continuity the admissibility constraint is used to bound the form operator on the admissible sets.

Proof

Each result cited is universally quantified over the data in its own statement. Throughout, x,yD(A)x,y\in D(A), and we write a=Axa=Ax, b=Ayb=Ay and Φ(z)=B(z)+L(z)\Phi(z)=B(z)+L(z) for zVz\in V; norms and pairings without a subscript are those of HH. By The Penalty Function h=12V2h=\tfrac12|\cdot|_V^2 of a Hilbert Triple: Expansion Identities, Closed Sublevel Sets, Density and Local Bounds §nonneg we have 0h(z)0\le h(z) for zVz\in V, and 2h(z)=zV22h(z)=|z|_{V}^{2} by Hilbert Triples: Standing Notation and Background §penalty. By Hilbert Triples: Standing Notation and Background §triple we have zHzV|z|_{H}\le|z|_{V} for zVz\in V. By Elementary Properties of a Hilbert Triple: the Embedding, the Riesz Map and the Form Operator §symmetric, Az,vH=z,vV\langle Az,v\rangle_{H}=\langle z,v\rangle_{V} for zD(A)z\in D(A) and vVv\in V.

Since LL is Lipschitz with constant \ell, L(z)L(v)zv|L(z)-L(v)|\le\ell|z-v| for z,vHz,v\in H, and taking v=0Hv=0_{H} and using The Norm Metric of a Real Inner Product Space: Triangle Inequalities, Limits and Continuity §triangle gives

L(z)cL+zfor every zH.|L(z)|\le c_{L}+\ell|z|\qquad\text{for every }z\in H .

We use the following three elementary facts. First, by Elementary Identities in a Real Inner Product Space §expansion, 12c2+u,c=12c+u212u2\tfrac{1}{2}|c|^{2}+\langle u,c\rangle=\tfrac{1}{2}|c+u|^{2}-\tfrac{1}{2}|u|^{2} for u,cHu,c\in H. Secondly, for real c>0c>0 and u,vu,v with 0u0\le u and 0v0\le v, expanding 01c(cuv)20\le\tfrac{1}{c}(c\,u-v)^{2} gives 2uvcu2+v2c2uv\le c\,u^{2}+\tfrac{v^{2}}{c}; we use it as uv14u2+v2uv\le\tfrac{1}{4}u^{2}+v^{2} (the case c=12c=\tfrac{1}{2}) and in the displayed forms below. Thirdly, (s+t)22s2+2t2(s+t)^{2}\le2s^{2}+2t^{2} for real s,ts,t, by Products and Sums of Weighted Square-Summable Sequences of Real Numbers §pointwise.

Claim 1. The values of FF are real numbers and its domain is D(A)×R×H×Sym(V)D(A)\times\mathbb{R}\times H\times\mathrm{Sym}(V), which with W=D(A)W=D(A) is what Second-Order Equation Operator on an Open Subset of a Hilbert Triple and Its δ\delta-Shifts §operator requires of a second-order equation operator on HH. The defining expression does not contain its 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,XSym(V)X,X'\in\mathrm{Sym}(V), which is First-Order Equation Operator on a Hilbert Triple §first-order. Degenerate ellipticity is then A First-Order Equation Operator is Degenerate Elliptic and Its δ\delta-Shifts Ignore the Form Argument §elliptic.

Claim 2. Let RR be positive, xD(A)x\in D(A), pHp\in H, XSym(V)X\in\mathrm{Sym}(V) and r,sRr,s\in\mathbb{R} with RsrR-R\le s\le r\le R. Every term of FF other than λ0r\lambda_{0}r is independent of the second argument, so

F(x,r,p,X)F(x,s,p,X)=λ0rλ0s=λ0(rs),F(x,r,p,X)-F(x,s,p,X)=\lambda_{0}r-\lambda_{0}s=\lambda_{0}(r-s),

and in particular λ0(rs)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). As λ0\lambda_{0} is positive, this is the requirement of Locally Strictly Proper Second-Order Equation Operator on a Hilbert Triple §constant; since RR was an arbitrary positive real, FF is locally strictly proper by Locally Strictly Proper Second-Order Equation Operator on a Hilbert Triple §strictly-proper.

The shifts. Let δR\delta\in\mathbb{R} satisfy 0<δ<10<\delta<1. By Second-Order Equation Operator on an Open Subset of a Hilbert Triple and Its δ\delta-Shifts §shifted and claim 1, the δ\delta-shifts do not depend on their fourth argument and are given by

Fδ(x,r,p,X)=λ0(r+δh(x))+12p+δa2+a+Φ(x),p+δag(x),F^{-}_{\delta}(x,r,p,X)=\lambda_{0}\bigl(r+\delta h(x)\bigr)+\tfrac{1}{2}|p+\delta a|^{2}+\langle a+\Phi(x),p+\delta a\rangle-g(x), Fδ+(y,s,p,X)=λ0(sδh(y))+12pδb2+b+Φ(y),pδbg(y).F^{+}_{\delta}(y,s,p',X')=\lambda_{0}\bigl(s-\delta h(y)\bigr)+\tfrac{1}{2}|p'-\delta b|^{2}+\langle b+\Phi(y),p'-\delta b\rangle-g(y).

By the first elementary fact, with c=p+δac=p+\delta a and u=au=a, and with c=pδbc=p'-\delta b and u=bu=b,

Fδ(x,r,p,X)=λ0(r+δh(x))+12p+(1+δ)a212a2+Φ(x),p+δag(x),()F^{-}_{\delta}(x,r,p,X)=\lambda_{0}\bigl(r+\delta h(x)\bigr)+\tfrac{1}{2}\bigl|p+(1+\delta)a\bigr|^{2}-\tfrac{1}{2}|a|^{2}+\langle\Phi(x),p+\delta a\rangle-g(x),\tag{$*$} Fδ+(y,s,p,X)=λ0(sδh(y))+12p+(1δ)b212b2+Φ(y),pδbg(y).()F^{+}_{\delta}(y,s,p',X')=\lambda_{0}\bigl(s-\delta h(y)\bigr)+\tfrac{1}{2}\bigl|p'+(1-\delta)b\bigr|^{2}-\tfrac{1}{2}|b|^{2}+\langle\Phi(y),p'-\delta b\rangle-g(y).\tag{$**$}

Claim 3, the two moduli. By The Nondecreasing Envelope of a Truncated Modulus of Continuity §modulus the function ωˉg\bar{\omega}_{g} is a modulus of continuity satisfying 0ωˉg(t)2Cg0\le\bar{\omega}_{g}(t)\le2C_{g} for every nonnegative tt; as 2Cg2C_{g} is nonnegative, its quadratic reparametrisation ωg\omega_{g}^{\ast} at 2Cg2C_{g} is therefore defined, and is a nondecreasing modulus of continuity, by Sums, Nonnegative Multiples, Monotonicity and Quadratic Reparametrisation of Moduli of Continuity §quadratic. By Linear Moduli of Continuity §modulus the function ttt\mapsto\ell t is a modulus of continuity, since 00\le\ell; so ω1\omega_{1} is a modulus of continuity by Sums, Nonnegative Multiples, Monotonicity and Quadratic Reparametrisation of Moduli of Continuity §sum. For a real α>1\alpha>1 the number KαK\alpha is nonnegative, being a product of positive numbers, so tKαtt\mapsto K\alpha t is a modulus of continuity by Linear Moduli of Continuity §modulus.

Claim 3, the inequality. Let RR be positive, let x,yD(A)x,y\in D(A), let rRr\in\mathbb{R} with RrR-R\le r\le R, let X,YSym(H)X,Y\in\mathrm{Sym}(H) and let α,δR\alpha,\delta\in\mathbb{R} with 1<α1<\alpha and 0<δ<10<\delta<1. Put w=α(xy)w=\alpha(x-y) and σ=αxy2+1α\sigma=\alpha|x-y|^{2}+\tfrac{1}{\alpha}, and let D=Fδ(x,r,w,X)Fδ+(y,r,w,Y)D=F^{-}_{\delta}(x,r,w,X)-F^{+}_{\delta}(y,r,w,Y). Expanding the two shift formulas and Elementary Identities in a Real Inner Product Space §expansion,

D=λ0δ(h(x)+h(y))+δw,a+b+δ22(a2b2)+ab,w+Φ(x)Φ(y),w+δ(a2+b2)+δ(Φ(x),a+Φ(y),b)(g(x)g(y)).D=\lambda_{0}\delta\bigl(h(x)+h(y)\bigr)+\delta\langle w,a+b\rangle+\tfrac{\delta^{2}}{2}\bigl(|a|^{2}-|b|^{2}\bigr)+\langle a-b,w\rangle+\langle\Phi(x)-\Phi(y),w\rangle+\delta\bigl(|a|^{2}+|b|^{2}\bigr)+\delta\bigl(\langle\Phi(x),a\rangle+\langle\Phi(y),b\rangle\bigr)-\bigl(g(x)-g(y)\bigr).

We bound the eight groups in turn, taking the fourth and the eighth together in (vii).

(i) 0λ0δ(h(x)+h(y))0\le\lambda_{0}\delta(h(x)+h(y)), all three factors being nonnegative.

(ii) Since xyVx-y\in V, the symmetry identity gives ab,w=α(x,xyVy,xyV)=αxyV2\langle a-b,w\rangle=\alpha\bigl(\langle x,x-y\rangle_{V}-\langle y,x-y\rangle_{V}\bigr)=\alpha|x-y|_{V}^{2}, which is nonnegative. This term is not discarded: it is combined with the eighth group in (vii), where it absorbs the oscillation of gg.

(iii) Likewise w,a+b=α(x,xyV+y,xyV)=α(xV2yV2)=2α(h(x)h(y))\langle w,a+b\rangle=\alpha\bigl(\langle x,x-y\rangle_{V}+\langle y,x-y\rangle_{V}\bigr)=\alpha\bigl(|x|_{V}^{2}-|y|_{V}^{2}\bigr)=2\alpha\bigl(h(x)-h(y)\bigr), so, hh being nonnegative,

δw,a+b2αδh(y)2αδ(h(x)+h(y)+1).\delta\langle w,a+b\rangle\ge-2\alpha\,\delta\,h(y)\ge-2\alpha\cdot\delta\bigl(h(x)+h(y)+1\bigr).

(iv) Since 0<δ<10<\delta<1 we have 121δ2\tfrac{1}{2}\le1-\tfrac{\delta}{2} and 121+δ2\tfrac{1}{2}\le1+\tfrac{\delta}{2}, so

δ22(a2b2)+δ(a2+b2)=δ(1+δ2)a2+δ(1δ2)b2  δ2(a2+b2).\tfrac{\delta^{2}}{2}\bigl(|a|^{2}-|b|^{2}\bigr)+\delta\bigl(|a|^{2}+|b|^{2}\bigr)=\delta\bigl(1+\tfrac{\delta}{2}\bigr)|a|^{2}+\delta\bigl(1-\tfrac{\delta}{2}\bigr)|b|^{2}\ \ge\ \tfrac{\delta}{2}\bigl(|a|^{2}+|b|^{2}\bigr).

(v) Φ(x)Φ(y),w=αB(x)B(y),xy+αL(x)L(y),xy\langle\Phi(x)-\Phi(y),w\rangle=\alpha\langle B(x)-B(y),x-y\rangle+\alpha\langle L(x)-L(y),x-y\rangle. The first summand is nonnegative by Monotone, AA-Monotone and Locally Bounded Nonlinearities on a Hilbert Triple §monotone and 0<α0<\alpha. For the second, The Cauchy-Schwarz Inequality in a Real Inner Product Space and the Lipschitz bound give L(x)L(y),xyxy2\langle L(x)-L(y),x-y\rangle\ge-\ell|x-y|^{2}, so, 1α\tfrac{1}{\alpha} being positive and \ell nonnegative,

Φ(x)Φ(y),wαxy2σ.\langle\Phi(x)-\Phi(y),w\rangle\ge-\ell\,\alpha|x-y|^{2}\ge-\ell\,\sigma .

(vi) δ(Φ(x),a+Φ(y),b)=δ(B(x),a+B(y),b)+δ(L(x),a+L(y),b)\delta(\langle\Phi(x),a\rangle+\langle\Phi(y),b\rangle)=\delta(\langle B(x),a\rangle+\langle B(y),b\rangle)+\delta(\langle L(x),a\rangle+\langle L(y),b\rangle). The first summand is nonnegative by Monotone, AA-Monotone and Locally Bounded Nonlinearities on a Hilbert Triple §a-monotone. For the second, The Cauchy-Schwarz Inequality in a Real Inner Product Space and the second elementary fact give L(x),aL(x)a14a2L(x)2\langle L(x),a\rangle\ge-|L(x)|\,|a|\ge-\tfrac{1}{4}|a|^{2}-|L(x)|^{2}, while the third elementary fact, the Lipschitz bound and xxV|x|\le|x|_{V} give

L(x)2(cL+x)22cL2+22x22cL2+22xV2=2cL2+42h(x),|L(x)|^{2}\le\bigl(c_{L}+\ell|x|\bigr)^{2}\le2c_{L}^{2}+2\ell^{2}|x|^{2}\le2c_{L}^{2}+2\ell^{2}|x|_{V}^{2}=2c_{L}^{2}+4\ell^{2}h(x),

and likewise for yy. Hence

δ(Φ(x),a+Φ(y),b)  δ4(a2+b2)δ(4cL2+42(h(x)+h(y)))  δ4(a2+b2)(4cL2+42)δ(h(x)+h(y)+1),\delta\bigl(\langle\Phi(x),a\rangle+\langle\Phi(y),b\rangle\bigr)\ \ge\ -\tfrac{\delta}{4}\bigl(|a|^{2}+|b|^{2}\bigr)-\delta\Bigl(4c_{L}^{2}+4\ell^{2}\bigl(h(x)+h(y)\bigr)\Bigr)\ \ge\ -\tfrac{\delta}{4}\bigl(|a|^{2}+|b|^{2}\bigr)-\bigl(4c_{L}^{2}+4\ell^{2}\bigr)\,\delta\bigl(h(x)+h(y)+1\bigr),

the last step because 1h(x)+h(y)+11\le h(x)+h(y)+1 and h(x)+h(y)h(x)+h(y)+1h(x)+h(y)\le h(x)+h(y)+1 and the coefficients are nonnegative.

(vii) We bound the fourth and the eighth group together. Write t=xyVt=|x-y|_{V}, so that their sum is αt2(g(x)g(y))\alpha t^{2}-\bigl(g(x)-g(y)\bigr) by (ii). Since x,yD(A)Vx,y\in D(A)\subseteq V, the hypotheses on gg give g(x)g(y)g(x)g(y)ωg(t)g(x)-g(y)\le|g(x)-g(y)|\le\omega_{g}(t), by claim 3 of Properties of the Absolute Value in an Ordered Field, and g(x)g(y)g(x)+g(y)2Cg|g(x)-g(y)|\le|g(x)|+|g(y)|\le2C_{g}, by claims 2 and 5 of that lemma; so g(x)g(y)g(x)-g(y) is at most both ωg(t)\omega_{g}(t) and 2Cg2C_{g} and therefore

g(x)g(y)ωˉg(t)g(x)-g(y)\le\bar{\omega}_{g}(t)

by The Nondecreasing Envelope of a Truncated Modulus of Continuity §majorant. Note that αt2\alpha t^{2} is nonnegative, being a product of the positive α\alpha with a square, and that 0ωg(σ)0\le\omega_{g}^{\ast}(\sigma), ωg\omega_{g}^{\ast} being a modulus of continuity.

Suppose first that 2Cgαt22C_{g}\le\alpha t^{2}. Since ωˉg(t)2Cg\bar{\omega}_{g}(t)\le2C_{g} by The Nondecreasing Envelope of a Truncated Modulus of Continuity §modulus, transitivity gives g(x)g(y)αt2g(x)-g(y)\le\alpha t^{2}, so αt2(g(x)g(y))\alpha t^{2}-(g(x)-g(y)) is nonnegative and hence at least ωg(σ)-\omega_{g}^{\ast}(\sigma).

Suppose instead that αt2<2Cg\alpha t^{2}<2C_{g}. Multiplying by the positive 1α\tfrac{1}{\alpha} gives t2<2Cg1αt^{2}<2C_{g}\tfrac{1}{\alpha}, while 1ασ\tfrac{1}{\alpha}\le\sigma, the difference being the nonnegative αxy2\alpha|x-y|^{2}, and 02Cg0\le2C_{g} give 2Cg1α2Cgσ2C_{g}\tfrac{1}{\alpha}\le2C_{g}\sigma by claim 5 of Elementary Arithmetic in an Ordered Field. Hence t22Cgσt^{2}\le2C_{g}\sigma, and so ωˉg(t)ωg(σ)\bar{\omega}_{g}(t)\le\omega_{g}^{\ast}(\sigma) by Sums, Nonnegative Multiples, Monotonicity and Quadratic Reparametrisation of Moduli of Continuity §quadratic, the reparametrisation being taken at c=2Cgc=2C_{g}. Discarding the nonnegative αt2\alpha t^{2},

αt2(g(x)g(y))  ωˉg(t)  ωg(σ).\alpha t^{2}-\bigl(g(x)-g(y)\bigr)\ \ge\ -\bar{\omega}_{g}(t)\ \ge\ -\omega_{g}^{\ast}(\sigma).

In either case

ab,w(g(x)g(y))  ωg(σ).\langle a-b,w\rangle-\bigl(g(x)-g(y)\bigr)\ \ge\ -\omega_{g}^{\ast}(\sigma).

(viii) The bounds (i), (iii), (iv), (v), (vi) and (vii) account for all eight groups of DD, the last of them for the fourth and the eighth together. Adding them, the terms δ2(a2+b2)\tfrac{\delta}{2}(|a|^{2}+|b|^{2}) of (iv) and δ4(a2+b2)-\tfrac{\delta}{4}(|a|^{2}+|b|^{2}) of (vi) combine to the nonnegative δ4(a2+b2)\tfrac{\delta}{4}(|a|^{2}+|b|^{2}), and we obtain

D  ωg(σ)σ(2α+4cL2+42)δ(h(x)+h(y)+1).D\ \ge\ -\omega_{g}^{\ast}(\sigma)-\ell\,\sigma-\bigl(2\alpha+4c_{L}^{2}+4\ell^{2}\bigr)\,\delta\bigl(h(x)+h(y)+1\bigr).

Since 1<α1<\alpha and 4cL2+424c_{L}^{2}+4\ell^{2} is nonnegative, 2α+4cL2+422α+(4cL2+42)α=Kα2\alpha+4c_{L}^{2}+4\ell^{2}\le2\alpha+(4c_{L}^{2}+4\ell^{2})\alpha=K\alpha, so the right-hand side is at least ω1(σ)ω2(δ(h(x)+h(y)+1),α)-\omega_{1}(\sigma)-\omega_{2}(\delta(h(x)+h(y)+1),\alpha). This is the inequality required by The First-Order Structure Condition for a Second-Order Equation Operator on a Hilbert Triple §pair, so (ω1,ω2)(\omega_{1},\omega_{2}) is a structure pair for FF at RR; as RR was an arbitrary positive real, FF satisfies the first-order structure condition by The First-Order Structure Condition for a Second-Order Equation Operator on a Hilbert Triple §structure.

Claim 4, a bound on the form operator over the admissible sets. Let δ,RR\delta,R\in\mathbb{R} satisfy 0<δ<10<\delta<1 and 0<R0<R. Let ρ\rho be the nonnegative real with ρ2=2R\rho^{2}=2R, given by Existence and Uniqueness of the Nonnegative Square Root. If (z,r,p,X)W(z,r,p,X)\in\mathcal{W} is RR-bounded then h(z)<Rh(z)<R, so zV2=2h(z)<ρ2|z|_{V}^{2}=2h(z)<\rho^{2} and therefore zVρ|z|_{V}\le\rho by claim 1 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field and trichotomy, and zρ|z|\le\rho. Let β\beta be a bound for BB on the set of vVv\in V with vVρ|v|_{V}\le\rho, as provided by Monotone, AA-Monotone and Locally Bounded Nonlinearities on a Hilbert Triple §bounded, put μ=cL+ρ\mu=c_{L}+\ell\rho and N=β+μN=\beta+\mu; then Φ(z)N|\Phi(z)|\le N for every such zz, by The Norm Metric of a Real Inner Product Space: Triangle Inequalities, Limits and Continuity §triangle and the bound on L|L|.

Let ξ=(x,r,p,X)\xi=(x,r,p,X) be RR-bounded. From ()(*), using 12p20\tfrac{1}{2}|p|^{2}\ge0, The Cauchy-Schwarz Inequality in a Real Inner Product Space, p<R|p|<R, 1+δ21+\delta\le2, the second elementary fact in the form L(x),a14a2L(x)2\langle L(x),a\rangle\ge-\tfrac{1}{4}|a|^{2}-|L(x)|^{2}, Monotone, AA-Monotone and Locally Bounded Nonlinearities on a Hilbert Triple §a-monotone, 0h(x)0\le h(x), Rr-R\le r and g(x)Cg|g(x)|\le C_{g},

Fδ(ξ)  (1+δ)212a22RaNRδ4a2μ2λ0RCg  3δ4a22RaC,F^{-}_{\delta}(\xi)\ \ge\ \tfrac{(1+\delta)^{2}-1}{2}|a|^{2}-2R|a|-NR-\tfrac{\delta}{4}|a|^{2}-\mu^{2}-\lambda_{0}R-C_{g}\ \ge\ \tfrac{3\delta}{4}|a|^{2}-2R|a|-C_{-},

where C=NR+μ2+λ0R+CgC_{-}=NR+\mu^{2}+\lambda_{0}R+C_{g}; here (1+δ)212=δ+δ22δ\tfrac{(1+\delta)^{2}-1}{2}=\delta+\tfrac{\delta^{2}}{2}\ge\delta. Similarly, let η=(y,s,p,X)\eta=(y,s,p',X') be RR-bounded. From ()(**), using 12p212R2\tfrac{1}{2}|p'|^{2}\le\tfrac{1}{2}R^{2}, The Cauchy-Schwarz Inequality in a Real Inner Product Space, 1δ11-\delta\le1, Φ(y)N|\Phi(y)|\le N, 0h(y)0\le h(y), sRs\le R and g(y)Cg-g(y)\le C_{g},

Fδ+(η)  1(1δ)22b2+(R+N)b+C+  δ2b2+(R+N)b+C+,F^{+}_{\delta}(\eta)\ \le\ -\tfrac{1-(1-\delta)^{2}}{2}|b|^{2}+(R+N)|b|+C_{+}\ \le\ -\tfrac{\delta}{2}|b|^{2}+(R+N)|b|+C_{+},

where C+=12R2+NR+λ0R+CgC_{+}=\tfrac{1}{2}R^{2}+NR+\lambda_{0}R+C_{g}; here 1(1δ)22=δδ22δ2\tfrac{1-(1-\delta)^{2}}{2}=\delta-\tfrac{\delta^{2}}{2}\ge\tfrac{\delta}{2} because δ<1\delta<1.

By the second elementary fact, 2Ra3δ8a2+8R23δ2R|a|\le\tfrac{3\delta}{8}|a|^{2}+\tfrac{8R^{2}}{3\delta} and (R+N)bδ4b2+(R+N)2δ(R+N)|b|\le\tfrac{\delta}{4}|b|^{2}+\tfrac{(R+N)^{2}}{\delta}. Hence the two displays give

Fδ(ξ)  3δ8a2E,Fδ+(η)  δ4b2+E,F^{-}_{\delta}(\xi)\ \ge\ \tfrac{3\delta}{8}|a|^{2}-E,\qquad F^{+}_{\delta}(\eta)\ \le\ -\tfrac{\delta}{4}|b|^{2}+E,

where E=8R23δ+(R+N)2δ+C+C+E=\tfrac{8R^{2}}{3\delta}+\tfrac{(R+N)^{2}}{\delta}+C_{-}+C_{+}; in particular Fδ(ξ)EF^{-}_{\delta}(\xi)\ge-E and Fδ+(η)EF^{+}_{\delta}(\eta)\le E for all RR-bounded ξ\xi and η\eta.

Now 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 §admissible there is an RR-bounded η\eta with Fδ(ξ)Fδ+(η)<RF^{-}_{\delta}(\xi)-F^{+}_{\delta}(\eta)<R, so 3δ8a2EFδ(ξ)<R+Fδ+(η)R+E\tfrac{3\delta}{8}|a|^{2}-E\le F^{-}_{\delta}(\xi)<R+F^{+}_{\delta}(\eta)\le R+E and therefore a2<8(R+2E)3δ|a|^{2}<\tfrac{8(R+2E)}{3\delta}. Similarly, if 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(ξ)Fδ+(η)<RF^{-}_{\delta}(\xi)-F^{+}_{\delta}(\eta)<R, so δ4b2+EFδ+(η)>Fδ(ξ)RER-\tfrac{\delta}{4}|b|^{2}+E\ge F^{+}_{\delta}(\eta)>F^{-}_{\delta}(\xi)-R\ge-E-R and therefore b2<4(R+2E)δ|b|^{2}<\tfrac{4(R+2E)}{\delta}. Let Λ\Lambda be the nonnegative real whose square is the larger of 8(R+2E)3δ\tfrac{8(R+2E)}{3\delta} and 4(R+2E)δ\tfrac{4(R+2E)}{\delta}, again by Existence and Uniqueness of the Nonnegative Square Root. Then AxΛ|Ax|\le\Lambda whenever (x,r,p,X)(x,r,p,X) lies in Sδ,RS^{-}_{\delta,R} or in Sδ,R+S^{+}_{\delta,R}, by claim 1 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field and trichotomy.

Claim 4, the shift modulus. Put c=R+2Λ+Nc=R+2\Lambda+N and let ω\omega be the function on the nonnegative reals with ω(t)=ct+12t2\omega(t)=c\,t+\tfrac{1}{2}t^{2}. It is nonnegative there. Given a positive ε\varepsilon, let δ0\delta_{0} be the lesser of 11 and εc+1\tfrac{\varepsilon}{c+1}, which is positive; every nonnegative tδ0t\le\delta_{0} satisfies t1t\le1, hence t2tt^{2}\le t and ω(t)(c+1)tε\omega(t)\le(c+1)t\le\varepsilon. So ω\omega is a modulus of continuity. It is also nondecreasing on the nonnegative reals: for 0st0\le s\le t we have csctc\,s\le c\,t by Linear Moduli of Continuity §monotone, and s2t2s^{2}\le t^{2} by claim 2 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field, whence 12s212t2\tfrac{1}{2}s^{2}\le\tfrac{1}{2}t^{2} by claim 5 of Elementary Arithmetic in an Ordered Field; adding the two inequalities by claims 3 and 2 of that lemma gives ω(s)ω(t)\omega(s)\le\omega(t).

Let qHq\in H and YSym(H)Y\in\mathrm{Sym}(H). Let (x,r,p,X)Sδ,R(x,r,p,X)\in S^{-}_{\delta,R} and put P=p+δaP=p+\delta a, so that PR+δΛR+Λ|P|\le R+\delta\Lambda\le R+\Lambda by The Norm Metric of a Real Inner Product Space: Triangle Inequalities, Limits and Continuity §triangle. Since FF, and hence each of its shifts, does not depend on its fourth argument, Elementary Identities in a Real Inner Product Space §expansion and The Cauchy-Schwarz Inequality in a Real Inner Product Space give

Fδ(x,r,p+q,X+Y)Fδ(x,r,p,X)=P,q+12q2+a+Φ(x),q(P+a+Φ(x))q+12q2ω(q),F^{-}_{\delta}(x,r,p+q,X+Y)-F^{-}_{\delta}(x,r,p,X)=\langle P,q\rangle+\tfrac{1}{2}|q|^{2}+\langle a+\Phi(x),q\rangle\le\bigl(|P|+|a|+|\Phi(x)|\bigr)|q|+\tfrac{1}{2}|q|^{2}\le\omega(|q|),

because P+a+Φ(x)(R+Λ)+Λ+N=c|P|+|a|+|\Phi(x)|\le(R+\Lambda)+\Lambda+N=c. As ω\omega is nondecreasing and 0Y0\le\lVert Y\rVert, we get ω(q)ω(q+Y)\omega(|q|)\le\omega(|q|+\lVert Y\rVert), which is the first requirement of The Shift-Continuity Condition on Admissible Test Data §modulus. Likewise, for (y,s,p,X)Sδ,R+(y,s,p',X')\in S^{+}_{\delta,R}, putting P=pδbP'=p'-\delta b,

Fδ+(y,s,p+q,X+Y)Fδ+(y,s,p,X)=P,q+12q2+b+Φ(y),q(P+b+Φ(y))qω(q+Y),F^{+}_{\delta}(y,s,p'+q,X'+Y)-F^{+}_{\delta}(y,s,p',X')=\langle P',q\rangle+\tfrac{1}{2}|q|^{2}+\langle b+\Phi(y),q\rangle\ge-\bigl(|P'|+|b|+|\Phi(y)|\bigr)|q|\ge-\omega(|q|+\lVert Y\rVert),

where 12q2\tfrac{1}{2}|q|^{2} was discarded as nonnegative. This is the second requirement. 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 by The Shift-Continuity Condition on Admissible Test Data §continuity.

Claim 5. By claims 1, 2, 3 and 4 the operator FF is a first-order second-order equation operator on HH that is locally strictly proper and satisfies the first-order structure condition and the shift-continuity condition, which are the hypotheses of A Comparison Principle on a Hilbert Triple under the First-Order Structure Condition §comparison; applying that theorem to uu, vv and CC gives u(x)v(x)u(x)\le v(x) for every xVx\in V.

Please log in to copy this version.

Citations

Loading…

Dependency Graph

0 prerequisites

Comments

Loading…