Solution to Partial Differential Equations: Methods and Applications (Robert McOwen) Section 4.5
- Let K(x)=−(4π|x|)−1, f∈C(R3) with f(x)=O(|x|−2−ϵ) as |x|→∞, and u(x)=∫R3K(x−y)f(y)dy.
- Show u∈C1(R3) and ∇u(x)=O(|x|−1−ϵ) as |x|→∞.
- If additionally f∈C1(R3), show u∈C2(R3).
- Since f(x)=O(|x|−2−ϵ) as |x|→∞, there exist positive constants C and R such that |f(y)|≤C|y|−2−ϵ for |y|≥R, which implies M=maxy∈R3|f(y)| is finite. For any x∈R3, we set R′=max{R,2|x|} and pick ρ>0 such that Bρ(x)⊆BR′(0). Then u can be represented as
u(x)=∫BR′(0)−Bρ(x)K(x−y)f(y)dy+∫Bρ(x)K(x−y)f(y)dy+∫R3−BR′(0)K(x−y)f(y)dy.
The first integral is proper so it is well-defined. The second integral can be estimated by|∫Bρ(x)K(x−y)f(y)dy|≤M4π∫Bρ(x)dy|x−y|=M4π∫π0∫2π0∫ϵ0ρ2sinϕρdρdθdϕ=M4π⋅2⋅2π⋅ϵ22=Mϵ22.
The third integral can be estimated by|∫R3−BR′(0)K(x−y)f(y)dy|≤C4π∫R3−BR′(0)|y|−2−ϵ|x−y|2dy≤C2π∫R3−BR′(0)|y|−3−ϵdy=C2π∫π0∫2π0∫∞R′ρ−3−ϵ⋅ρ2sinϕdρdθdϕ=C2π⋅2⋅2π⋅ρ−ϵ−ϵ|∞R′=2C(R′)−ϵϵ.
Here we have used the inequality |x−y|≥|y|/2 for y∈R3−BR′(0). Therefore, the Newtonian potential u of f is well-defined for all x∈R3.
For any i=1,…,n, letu(i)(x)=∫R3∂K∂xi(x−y)f(y)dy=14π∫R3xi−yi|x−y|3f(y)dy,
which is well-defined because ∂K∂xi is O(|x−y|−2 as |x−y|→0 and ∂K∂xi(x−y)f(y) is O(|y|−4−ϵ) as |y|→∞ (as for the above estimate of u). To show u(i)=∂u∂xi, we first approximate u by a smooth function as follows. Let η∈C∞(R) satisfy η(t)=0 for t<1, η(t)=1 for t>2 and 0≤η(t)≤1 and 0≤η′(t)≤2 for all t. Then we defineuδ(x)=∫R3Kδ(x−y)f(y)dy:=∫R3K(x−y)η(|x−y|δ)f(y)dy=−14π∫R3η(|x−y|/δ)|x−y|f(y)dyfor x∈R3,
where Kδ(z)=−η(|z|/δ)|z| for δ>0 and z∈R3. Note that Kδ is smooth so uδ is also smooth and satisfies∂uδ∂xi(x)=14π∫R3[xi−yi|x−y|3η(|x−y|δ)+xi−yiδ|x−y|η′(|x−y|δ)]f(y)dy
It is straightforward to check that uδ→u uniformly on any compact subsets of R3, which means u is continuous in R3. Now we prove ∂uδ∂xi→u(i) as δ→0. Since η(t)=1 for t>2, we observe thatu(i)(x)−∂uδ∂xi(x)=14π∫|x−y|<2δ{[1−η(|x−y|δ)]xi−yi|x−y|3−xi−yiδ|x−y|η′(|x−y|δ)}f(y)dy=14π∫|z|<2δ{[1−η(|z|δ)]zi|z|3−ziδ|z|η′(|z|δ)}f(x−z)dz.
It is easy to estimate the following integrals|14π∫|z|<2δziδ|z|η′(|z|δ)f(x−z)dz|≤2M4πδ∫|z|<2δdz=2M4πδ⋅4π(2δ)33=16Mδ23,|14π∫|z|<2δ[1−η(|z|δ)]zi|z|3f(x−z)dz|≤M2π∫|z|<2δdz|z|2=M2π∫π0∫2π0∫δ0ρ2sinϕρ2dρdθdϕ=M2π⋅2⋅2π⋅δ=2Mδ.
Thus, we see that|u(i)(x)−∂uδ∂xi(x)|≤(16δ3+2)Mδfor x∈R3,
which implies u(i) is the uniform limit of the continuous functions ∂uδ∂xi as δ tends to zero. Recall that Theorem 7.12 of "Introduction to Analysis (Fourth Edition)" states that "Let (a,b) be a bounded interval and suppose that fn is a sequence of functions which converges at some x0∈(a,b). If each fn is differentiable on (a,b) and f′n converges uniformly on (a,b) as n→∞, then fn converges uniformly on (a,b) and limn→∞f′n(x)=(limn→∞fn(x))′". Therefore we show that u∈C1(R3) and∂u∂xi(x)=u(i)(x)=14π∫R3xi−yi|x−y|3f(y)dy.
Write x=|x|ω for ω∈S2 and use the change of variable y=|x|z, we find|∇u(x)|≤14π∫R3|f(y)||x−y|2dy=14π∫R3f(|x|z)|x−|x|z|2|x|3dz=|x|4π∫R3f(|x|z)|ω−z|2dz=|x|4π(∫BR/|x|(0)f(|x|z)|ω−z|2dz+∫R3−BR/|x|(0)f(|x|z)|ω−z|2dz).
When |x|≥2R, the first integral can be estimated by∫BR/|x|(0)f(|x|z)|ω−z|2dz≤4M∫BR/|x|(0)dz=4M⋅4π(R/|x|)33=16MπR33|x|3.
On the other hand, the second integral is estimated by∫R3−BR/|x|(0)f(|x|z)|ω−z|2dz≤C∫R3−BR/|x|(0)|x|−2−ϵ|z|−2−ϵ|ω−z|2dz=C|x|−2−ϵ∫R3−BR/|x|(0)|z|−2−ϵ|ω−z|2dz≤˜C|x|−2−ϵ.
Here we have used the fact that improer integral is convergent because |z|−2−ϵ|ω−z|2/1|z|4+ϵ→1 as |z|→∞ and is O(|ω−z|−2) as z→ω. Therefore, we obtain|∇u(x)|≤|x|4π(16MπR33|x|3+˜C|x|−2−ϵ)≤ˆC|x|−1−ϵ.
The original statement is that "Show that u∈C1(Rn) and ∇u(x)=O(|x|−2)", which might be typos so I correct it directly. As my best knowledge, ∇u=O(|x|−1−ϵ) is more reasonable because u=O(|x|−ϵ) as |x|→∞, which is shown in two papers: "On Helmholtz's theorem and the completeness of the Papkovich-Neuber stress functions for infinite domains" (Author: M. E. Gurtin) and "A Constructive Proof of Helmholtz’s Theorem" (Author: Bernardo De La Calle Ysern, José C. Sabina De Lis). If you have a better proof to improve this decaying rate, please tell me! Thank you. - Note that
∂u∂xi=14π∫R3xi−yi|x−y|f(y)dy=14π∫R3yi|y|f(x−y)dy.
Then it is clear that∂2u∂xj∂xi=14π∫R3yi|y|3∂f∂xj(x−y)dy,
which is continuous in R3. Thus, u∈C2(R3).
沒有留言:
張貼留言