TheoremBase

A series whose terms vanish beyond a fixed index converges to the corresponding finite sum, which makes the finite noise square-summable and identifies its trace; the remaining hypotheses of the abstract viscous theorem are then read off from the published data of the non-viscous case together with the failure of finite-dimensionality.

Proof

Each result cited is universally quantified over the data in its own statement, and is applied to the data named here. The field axioms of R\mathbb{R} are used freely for associativity, commutativity and distributivity, and for s⋅1=ss\cdot1=s. By Finite Sum Notation in a Field a finite sum is determined by the map on the initial segment it is formed over, so two such sums agree as soon as the two maps agree; this is used without further comment.

Claim 1. (A series whose terms vanish beyond a fixed index.) Let M∈NM\in\mathbb{N} and let (ak)k∈N(a_{k})_{k\in\mathbb{N}} be a sequence of real numbers with ak=0a_{k}=0 for every k∈Nk\in\mathbb{N} with M<kM<k. Then the series ∑k=1∞ak\sum_{k=1}^{\infty}a_{k} converges, with sum ∑k=1Mak\sum_{k=1}^{M}a_{k}.

Proof of Claim 1. For m∈Nm\in\mathbb{N} let sm=∑k=1maks_{m}=\sum_{k=1}^{m}a_{k} be the partial sums of the series. We first show that sM+j=sMs_{M+j}=s_{M} for every j∈Nj\in\mathbb{N}, by Principle of Induction for the Natural Numbers applied to the set TT of natural numbers jj for which sM+j=sMs_{M+j}=s_{M}.

For j=1j=1 we have M+1=S(M)M+1=S(M) by claim 1 of Arithmetic of Addition on the Natural Numbers, where SS is the successor map of Natural Numbers; by claim 1 of Properties of Finite Sums, sS(M)=sM+aS(M)s_{S(M)}=s_{M}+a_{S(M)}; and M<S(M)M<S(M) by claim 5 of Properties of the Order on the Natural Numbers, so aS(M)=0a_{S(M)}=0 and sM+1=sMs_{M+1}=s_{M}. Thus 1∈T1\in T.

Let j∈Tj\in T. Then M+S(j)=S(M+j)M+S(j)=S(M+j) by Natural Numbers, and sS(M+j)=sM+j+aS(M+j)s_{S(M+j)}=s_{M+j}+a_{S(M+j)} by claim 1 of Properties of Finite Sums. Moreover M<M+jM<M+j by claim 6 of Properties of the Order on the Natural Numbers and M+j<S(M+j)M+j<S(M+j) by claim 5 of that lemma, so M<S(M+j)M<S(M+j) by the transitivity of the strict order in claim 1 of that lemma; hence aS(M+j)=0a_{S(M+j)}=0 and sM+S(j)=sM+j=sMs_{M+S(j)}=s_{M+j}=s_{M}. Thus S(j)∈TS(j)\in T, and T=NT=\mathbb{N}.

Now let ε\varepsilon be a positive real number and let m∈Nm\in\mathbb{N} satisfy M≤mM\le m. By the trichotomy in claim 3 of Properties of the Order on the Natural Numbers exactly one of m<Mm<M, m=Mm=M and M<mM<m holds, and the first is impossible: m<Mm<M would give m≤Mm\le M by claim 1 of that lemma, hence m=Mm=M by the antisymmetry in claim 2 of that lemma together with M≤mM\le m, and then m<mm<m, contradicting the irreflexivity in claim 2. So either m=Mm=M, or M<mM<m and then m=M+jm=M+j for some j∈Nj\in\mathbb{N} by claim 7 of that lemma; in both cases sm=sMs_{m}=s_{M} by the previous paragraph. Hence ∣sm−sM∣=∣0∣=0<ε|s_{m}-s_{M}|=|0|=0<\varepsilon, using claim 1 of Properties of the Absolute Value in an Ordered Field. Taking MM itself as the index required by Limit of a Sequence of Real Numbers, the sequence (sm)m∈N(s_{m})_{m\in\mathbb{N}} converges to sMs_{M}; so by Series of Real Numbers (ii) the series converges with sum sM=∑k=1Maks_{M}=\sum_{k=1}^{M}a_{k}. This proves Claim 1.

Claim 2. (The norms of the terms of the noise.) For j∈[N]j\in[N] one has ∣fj∣V2=cj2(1+4π2∥wj∥2)|f_{j}|_{V}^{2}=c_{j}^{2}\bigl(1+4\pi^{2}\lVert w_{j}\rVert^{2}\bigr), and for j∈Nj\in\mathbb{N} with N<jN<j one has ∣fj∣V2=0|f_{j}|_{V}^{2}=0.

Proof of Claim 2. The norm of H1(Tn)H^{1}(\mathbb{T}^{n}) satisfies ∥X∥H12=⟨X,X⟩H1\lVert X\rVert_{H^{1}}^{2}=\langle X,X\rangle_{H^{1}} by that clause, and is absolutely homogeneous by Elementary Identities in a Real Inner Product Space (iv), so for j∈[N]j\in[N]

∣fj∣V2=∥cjEwj∥H12=(∣cj∣ ∥Ewj∥H1)2=∣cj∣2 ⟨Ewj,Ewj⟩H1=cj2(1+4π2∥wj∥2),|f_{j}|_{V}^{2}=\lVert c_{j}E_{w_{j}}\rVert_{H^{1}}^{2}=\bigl(|c_{j}|\,\lVert E_{w_{j}}\rVert_{H^{1}}\bigr)^{2}=|c_{j}|^{2}\,\langle E_{w_{j}},E_{w_{j}}\rangle_{H^{1}}=c_{j}^{2}\bigl(1+4\pi^{2}\lVert w_{j}\rVert^{2}\bigr),

using claim 1 of Nonnegativity of Squares in an Ordered Field for ∣cj∣2=cj2|c_{j}|^{2}=c_{j}^{2} and The Trigonometric System is Continuously Differentiable and Orthogonal in the Sobolev Space of the Torus (iii), taken with k=m=wjk=m=w_{j}, for the last factor. For jj with N<jN<j we have fj=0Vf_{j}=0_{V}; since 0V=0⋅0V0_{V}=0\cdot0_{V} and the inner product is linear in its first argument by Real Inner Product Spaces (i), ⟨0V,0V⟩H1=0⋅⟨0V,0V⟩H1=0\langle 0_{V},0_{V}\rangle_{H^{1}}=0\cdot\langle 0_{V},0_{V}\rangle_{H^{1}}=0 by claim 1 of Zero Products and Elementary Identities in a Field, so ∣fj∣V2=0|f_{j}|_{V}^{2}=0. This proves Claim 2.

Proof of claim 1 of the statement. Apply Claim 1 with M=NM=N to the sequence ak=∣fk∣V2a_{k}=|f_{k}|_{V}^{2}, whose terms vanish for N<kN<k by Claim 2. It gives that ∑k=1∞∣fk∣V2\sum_{k=1}^{\infty}|f_{k}|_{V}^{2} converges with sum ∑k=1N∣fk∣V2\sum_{k=1}^{N}|f_{k}|_{V}^{2}; so ff is square-summable in VV by Square-Summable Sequences in the Form Space of a Hilbert Triple and the Trace of a Form along Them (i), and σ(f)\sigma(f), being that sum, equals ∑j=1Ncj2(1+4π2∥wj∥2)\sum_{j=1}^{N}c_{j}^{2}(1+4\pi^{2}\lVert w_{j}\rVert^{2}) by Claim 2.

Now let Y∈Sym(H1(Tn))Y\in\mathrm{Sym}(H^{1}(\mathbb{T}^{n})) and put ak=Y(fk,fk)a_{k}=Y(f_{k},f_{k}). By Bounded Symmetric Bilinear Forms on a Real Inner Product Space: Norm, Order, Identity Form and Restriction (i), which gives additivity and homogeneity in the first argument together with symmetry, a scalar may be taken out of either argument; so aj=Y(cjEwj,cjEwj)=cj2 Y(Ewj,Ewj)a_{j}=Y(c_{j}E_{w_{j}},c_{j}E_{w_{j}})=c_{j}^{2}\,Y(E_{w_{j}},E_{w_{j}}) for j∈[N]j\in[N], and aj=Y(0V,0V)=0⋅0⋅Y(0V,0V)=0a_{j}=Y(0_{V},0_{V})=0\cdot 0\cdot Y(0_{V},0_{V})=0 for N<jN<j, using 0V=0⋅0V0_{V}=0\cdot0_{V} and claim 1 of Zero Products and Elementary Identities in a Field. Claim 1, applied with M=NM=N to this sequence, gives that ∑k=1∞Y(fk,fk)\sum_{k=1}^{\infty}Y(f_{k},f_{k}) converges with sum ∑j=1Ncj2Y(Ewj,Ewj)\sum_{j=1}^{N}c_{j}^{2}Y(E_{w_{j}},E_{w_{j}}). By Square-Summable Sequences in the Form Space of a Hilbert Triple and the Trace of a Form along Them (ii) the real number TrfY\mathrm{Tr}_{f}Y is the sum of that series, so it equals ∑j=1Ncj2Y(Ewj,Ewj)\sum_{j=1}^{N}c_{j}^{2}Y(E_{w_{j}},E_{w_{j}}).

Proof of claim 2 of the statement. That L2(Tn)L^{2}(\mathbb{T}^{n}) is not finite-dimensional as a vector space over R\mathbb{R} is The Lebesgue Space of Square-Integrable Functions on the Torus is Not Finite-Dimensional, applied with the present nn.

The data bb, κ\kappa, λ0\lambda_{0}, CgC_{g}, ωg\omega_{g}, gg, BB and LL here satisfy exactly the hypotheses imposed on them in Well-Posedness of the Allen-Cahn Hamilton-Jacobi Equation on the Torus, and the Hilbert triple, the settings and the readings of ⟨ ⋅ ,⋅ ⟩L2\langle\,\cdot\,,\cdot\,\rangle_{L^{2}}, ∥ ⋅ ∥H1\lVert\,\cdot\,\rVert_{H^{1}}, D(A)D(A) and Sym(H1(Tn))\mathrm{Sym}(H^{1}(\mathbb{T}^{n})) are the ones adopted there. That BB is a monotone nonlinearity for (H,V,A)(H,V,A) is The Cube Map is a Monotone Nonlinearity on the Sobolev Hilbert Triple of the Torus (v), applied to the present bb and the present Hilbert triple; and Well-Posedness of the Allen-Cahn Hamilton-Jacobi Equation on the Torus (i) applies and gives that LL is Lipschitz with the nonnegative constant ∣κ∣|\kappa| from (L2(Tn),dL2)(L^{2}(\mathbb{T}^{n}),d_{L^{2}}) to itself.

We check the remaining hypotheses of Well-Posedness for a Viscous Hamilton-Jacobi Equation with a Monotone Nonlinearity on a Hilbert Triple, read for the Hilbert triple (H,V,A)(H,V,A) fixed above with ℓ=∣κ∣\ell=|\kappa|. The standing hypothesis of Hilbert Triples: Standing Notation and Background holds for this triple, as recorded in the statement. The reals λ0\lambda_{0} and CgC_{g} satisfy 0<λ00<\lambda_{0} and 0≤Cg0\le C_{g}, ωg\omega_{g} is a modulus of continuity, and g:V→Rg:V\to\mathbb{R} is bounded by CgC_{g} and has modulus ωg\omega_{g} with respect to ∣⋅∣V=∥ ⋅ ∥H1|\cdot|_{V}=\lVert\,\cdot\,\rVert_{H^{1}}, all by hypothesis. The reals θ\theta and ν\nu satisfy 0≤θ≤10\le\theta\le1 and 0≤ν0\le\nu by hypothesis, and ff is square-summable in VV by claim 1 of the statement, so the trace Trf\mathrm{Tr}_{f} is defined by Square-Summable Sequences in the Form Space of a Hilbert Triple and the Trace of a Form along Them (ii). Every hypothesis of that theorem is therefore met, and in particular the displayed function FF is the function denoted FF there, hence is defined on D(A)×R×L2(Tn)×Sym(H1(Tn))D(A)\times\mathbb{R}\times L^{2}(\mathbb{T}^{n})\times\mathrm{Sym}(H^{1}(\mathbb{T}^{n})).

Proof of claim 3 of the statement. By claim 2 of the statement the hypotheses of Well-Posedness for a Viscous Hamilton-Jacobi Equation with a Monotone Nonlinearity on a Hilbert Triple hold for the present data. Its Well-Posedness for a Viscous Hamilton-Jacobi Equation with a Monotone Nonlinearity on a Hilbert Triple (i) provides a function u:L2(Tn)→Ru:L^{2}(\mathbb{T}^{n})\to\mathbb{R} that is a viscosity solution of FF on L2(Tn)L^{2}(\mathbb{T}^{n}), satisfies ∣u(X)∣≤C|u(X)|\le C for every XX, where CC is the quotient of CgC_{g} by λ0\lambda_{0} and is nonnegative, and is uniformly continuous on L2(Tn)L^{2}(\mathbb{T}^{n}) with respect to dL2d_{L^{2}} and the metric of Real Hilbert Spaces: Standing Notation and Background (i). Its Well-Posedness for a Viscous Hamilton-Jacobi Equation with a Monotone Nonlinearity on a Hilbert Triple (iii), applied to this uu, gives that any viscosity solution u′u' of FF on L2(Tn)L^{2}(\mathbb{T}^{n}) that is continuous on L2(Tn)L^{2}(\mathbb{T}^{n}) and bounded by some C′′∈RC''\in\mathbb{R} satisfies u′(X)=u(X)u'(X)=u(X) for every X∈L2(Tn)X\in L^{2}(\mathbb{T}^{n}). These are the two assertions of the claim.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…