Partial Di erential Equations, 2nd Edition, L.C.Evans...
Click here to load reader
-
Upload
nguyenngoc -
Category
Documents
-
view
214 -
download
2
Transcript of Partial Di erential Equations, 2nd Edition, L.C.Evans...
Partial Differential Equations, 2nd Edition, L.C.Evans
Chapter 6 Second-Order Elliptic Equations
Shih-Hsin Chen∗, Yung-Hsiang Huang†
2017/03/27
1. (a) Direct computation. (b) c = ∆√a.
2. Proof. Define
B[u, v] =
∫U
n∑i,j=1
aijuxivxj + cuv dx, u, v ∈ H10 (U).
Clearly,
|B[u, v]| ≤ ‖aij‖∞‖Du‖L2(U)‖Dv‖L2(U) + ‖c‖∞‖u‖L2(U)‖v‖L2(U) ≤ α‖u‖H10 (U)‖v‖H1
0 (U),
for some positive constant α. For c(x) ≥ −µ, where µ is a fixed constant to be assigned later.
Using the uniform ellipticity, we have
θ
∫U
|Du|2 dx ≤∫U
n∑i,j=1
aijuxiuxj
= B[u, u]−∫U
cu2 dx
≤ B[u, u] + µ
∫U
u2 dx
≤ B[u, u] + Cµ
∫U
|Du|2 dx,
where the Poincare inequality is applied in the last inequality. Now we set µ =θ
2Cand see
that if c(x) ≥ θ
2C, then
θ
2
∫U
|Du|2 dx ≤ B[u, u].
∗Department of Math., National Taiwan University. Email: [email protected]†Department of Math., National Taiwan University. Email: [email protected]
1
Therefore,
‖u‖2H1
0 (U) = ‖u‖2L2(U) + ‖Du‖2
L2(U)
≤ (1 + C)‖Du‖2L2(U)
≤ 2
θ(1 + C)B[u, u].
3. Proof. Define
B[u, v] =
∫U
n∑i,j=1
uxixjvxixj dx, u, v ∈ H20 (U).
Use the integration by part (divergence theorem),∫U
|D2u|2 dx =
∫U
n∑i,j=1
uxixjuxixj dx =
∫U
∑i,j=1
uxixiuxjxj dx =
∫U
|∆u|2 dx
for u, v ∈ H20 (U). (Fix i, consider F = uxiDuxi and then use
∫U
divF dx =∫∂U
F · ν dS.)
‖Du‖2L2(U) = −
∫U
u∆u dx
≤ ε‖u‖2L2(U) +
1
4ε‖∆u‖2
L2(U)
≤ εC‖Du‖2L2(U) +
1
4ε‖∆u‖2
L2(U)
the Poincar’e inequality is used in the last inequality. We choose ε =1
2Cso that
‖Du‖2L2(U) ≤ C‖∆u‖2
L2(U).
Therefore,
‖u‖2H2
0 (U) = ‖u‖2L2(U) + ‖Du‖2
L2(U) + ‖D2u‖2L2(U)
≤ (1 + C)‖Du‖2L2(U) + ‖∆u‖2
L2(U)
≤ (1 + C + C2)‖∆u‖2L2(U)
= (1 + C + C2)B[u, u].
Thus,1
1 + C + C2‖u‖2
H20 (U) ≤ B[u, u], u ∈ H2
0 (U).
Remark 1. The validity of the definition of weak solution to Biharmonic equation with zero
Dirichlet and zero Neumann boundary is based on the zero trace theorem for W 2,p0 function,
which is not proved in Evans’ book.
2
4. Proof. ”⇒ ” take v = 1.
”⇐ ” Define
B[u, v] =
∫U
Du ·Dv dx, for u, v ∈ H1(U).
Consider a subspace of H1(U)
A = {u ∈ H1(U)∣∣(u)U = 0}
Let l(u) =∫Uu dx, then l is a continuous linear functional on H1(U) (the boundness of domain
U is used here.) Then A = l−1({0}) is closed in H1(U), so A is a Hilbert space with induced
inner product. Note that
|B[u, v]| ≤ ‖u‖A‖v‖A.
and by Poincare inequality,
‖u‖2L2(U) =
∫U
|u− (u)U |2 dx ≤ C
∫U
|Du|2 dx.
Therefore
‖u‖2A = ‖u‖2
L2(U) + ‖Du‖2L2(U) ≤ (1 + C)‖Du‖2
L2(U) = (1 + C)B[u, u].
By Lax-Milgram, for all f ∈ L2(U) (with∫Uf dx = 0), there exists a unique uf ∈ A such that
B[uf , v] = (f, v)L2(U), ∀v ∈ A.
Let v ∈ H1(U), then v − (v)U ∈ A. Thus, we have
(f, v)L2(U) = (f, (v)U)L2(U) + (f, v − (v)U)L2(U)
= 0 +B[uf , v − (v)U ], (since
∫U
f dx = 0)
= B[uf , v] =
∫U
Duf ·Dv dx, ∀v ∈ H1(U).
Therefore, uf ∈ A ⊂ H1(U) is a weak solution. In general, uf + C is always a weak solution
for any constant C.
5. Motivation. If u ∈ C∞(U) solves the given boundary value problem, then for each v ∈ C∞(U),∫U
fv =
∫U
(−∆u)v =
∫∂U
−v ∂u∂n
+
∫U
∇u∇v =
∫∂U
vu+
∫U
∇u∇v.
By density theorem and trace theorem, we define B on (H1(U))2 by
B[u, v] =
∫∂U
(Tv)(Tu) +
∫U
∇u∇v,
where T is the trace operator with finite operator norm ‖T‖. We call u ∈ H1(U) is a weak
solution of the given Robin boundary value problem if B[u, v] =∫Ufv for all v ∈ H1(U).
3
Proof. We prove uniqueness and existence by Lax-Milgram. Given u, v ∈ H1(U), it’s easy to see
|B[u, v]| ≤ max{1, ‖T‖2}‖u‖H1‖v‖H1 . Next, we prove B is strictly coercive by a contradiction
argument similar to the one for Poincare inequality.
Assume for each k ∈ N, there exists uk ∈ H1 such that ‖uk‖2H1 > kB[uk, uk].
Let vk = uk/‖uk‖H1 , then 1 = ‖vk‖2H1 > kB[vk, vk], so 1
k> B[vk, vk] ≥ ‖Dvk‖2
L2 ,
1k> ‖Tvk‖2
L2(∂U) and hence ‖vk‖2H1 < 1 + 1
k≤ 2. Rellich’s compactness theorem asserts there
exists a subsequence, still denoted by {vk}, converging to a function v in L2 sense. Note
‖Dvk‖L2 → 0, so for every φ ∈ C∞c (U),∫vDφ = lim
k→∞
∫vkDφ = lim
k→∞
∫−(Dvk)φ = 0
Hence v ∈ H1, Dv ≡ 0. Now we know vk → v in H1 sense, so ‖v‖H1 = 1. Moreover, v is
constant in each component of U . But
‖Tv‖L2(∂U) ≤ ‖T (v − vk)‖L2(∂U) + ‖Tvk‖L2(∂U) ≤ ‖T‖‖v − vk‖H1 +
√1
k→ 0, as k →∞,
which means v ≡ 0 on ∂U and then v ≡ 0 on U . This contradicts to ‖v‖H1 = 1. Therefore B
is strictly coercive.
6. (This boundary value problem is called Zaremba problem sometimes)
Definition. Similar to Exercise 5, we define B[u, v] =∫∇u∇v on H := H1 ∩ {u : Tu �Γ1= 0}.
The linearity and continuity of T ensures that H is a closed subspace of H1, so H is a Hilbert
space with induced inner product. u ∈ H is a weak solution of the given Robin boundary
value problem if B[u, v] =∫Ufv for all v ∈ H. (The zero Newmann boundary condition is
involved.)
Uniqueness and Existence. They are established by Lax-Milgram again, the coercivity of B is
established by almost the same argument as Exercise 5.
7. Proof. By definition, we know for each v ∈ H1(Rn),∫∇u∇v + c(u)v =
∫fv.
Take v = −D−hi Dhi u, 0 < |h| � 1. Then v ∈ H1 and has compact support. Hence∫
|DhiDu|2 +Dh
i c(u)Dhi u = −
∫f(D−hi Dh
i u).
By mean value theorem,
Dhi c(u)(x)Dh
i u(x) =c(u(x+ hei))− c(u(x))
hDhi u(x) = c′(· · · )|Dh
i u(x)|2 ≥ 0.
4
Hence ∫|Dh
iDu|2 ≤ |∫f(D−hi Dh
i u)| ≤ 1
2
∫|f |2 +
1
2
∫|D−hi Dh
i u|2
≤ 1
2
∫|f |2 +
1
2
∫|DDh
i u|2
The last inequality follows easily from mean value theorem. Hence∫|Dh
iDu|2 ≤∫|f |2, for all
small enough h. This implies∫|D2u|2 ≤
∫|f |2. (Evans, Page 292)
8. Proof. Differentiate the equation Lu = 0, we get (Daij)uxixj + aij(Du)xixj = 0. Multiply this
equation by 2Du, we have 2Du · (Daij)uxixj = −2aijDu · (Du)xixj . Adding −2aij(Du)xi(Du)xj
to both sides, we find
2Du · (Daij)uxixj − 2aij(Du)xi(Du)xj = −aij(|Du|2)xixj (1)
On the other hand, multiply the origin equation by 2u and add 2aijuxiuxj to both sides, we
have
2aijuxiuxj = 2aijuxiuxj + 2aijuuxixj = aij(u2)xixj (2)
(1)− λ(2) implies
−2λaijuxiuxj + 2Du · (Daij)uxixj − 2aij(Du)xi(Du)xj = −aij(λu2 + |Du|2)xixj
By uniform ellipticity, boundedness of Daij and Young’s inequality, the left hand side is non-
positive for large λ. The second assertion is a consequence of weak maximum principle to
L(|Du|2 + λu2) ≤ 0
9. Proof. By weak maximum principle, minUw = min
∂Uw = w(x0). Let v1 = u + ‖f‖∞w and
v2 = u− ‖f‖∞w, then Lv1 ≥ 0, Lv2 ≤ 0.
By weak maximum principle, minUv1 = min
∂Uv1 = ‖f‖∞min
∂Uw = ‖f‖∞w(x0) = v1(x0) and
maxU
v2 = max∂U
v2 = −‖f‖∞min∂U
w = −‖f‖∞w(x0) = v2(x0). Therefore,
0 ≥ ∂v1
∂ν(x0) =
∂u
∂ν(x0) + ‖f‖∞
∂w
∂ν(x0),
0 ≤ ∂v2
∂ν(x0) =
∂u
∂ν(x0)− ‖f‖∞
∂w
∂ν(x0).
Since u = 0 on ∂U , ∇u//ν. So
|∇u(x0)| = |∂u∂ν
(x0)| ≤ ‖f‖∞|∂w
∂ν(x0)|.
5
10. Proof. We omit (a) since is standard. For (b), if u attains an interior maximum, then the
conclusion follows from strong maximum principle.
If not, then for some x0 ∈ ∂U, u(x0) > u(x) ∀x ∈ U . Then Hopf’s lemma implies ∂u∂ν
(x0) > 0,
which is a contradiction.
Remark 2. A generalization of this problem to mixed boundary conditions is recorded in
Gilbarg-Trudinger, Elliptic PDEs of second order, Problem 3.1.
11. Proof. Define
B[u, v] =
∫U
∑i,j
aijuxivxj dx for u ∈ H1(U), v ∈ H10 (U).
By Exercise 5.17, φ(u) ∈ H1(U). Then, for all v ∈ C∞c (U), v ≥ 0,
B[φ(u), v] =
∫U
∑i,j
aij(φ(u))xivxj dx
=
∫U
∑i,j
aijφ′(u)uxivxj dx, (φ′(u) is bounded since u is bounded)
=
∫U
∑i,j
aijuxi(φ′(u)v)xj −
∑i,j
aijφ′′(u)uxiuxjv dx
≤ 0−∫U
φ′′(u)v|Du|2 dx ≤ 0, by convexity of φ.
(We don’t know whether the product of two H1 functions is weakly differentiable. This is why
we do not take v ∈ H10 .) Now we complete the proof with the standard density argument.
12. Proof. Given u ∈ C2(U) ∩ C(U) with Lu ≤ 0 in U and u ≤ 0 on ∂U . Since U is compact and
v ∈ C(U), v ≥ c > 0. So w := uv∈ C2(U) ∩ C(U). Brutal computation gives us
−aijwxixj =−aijuxixjv + aijvxixju
v2+aijvxiuxj − aijuxivxj
v2− aij 2
vvxj
vxiu− vuxiv2
=(Lu− biuxi − cu)v + (−Lv + bivxi + cv)u
v2+ 0 + aij
2
vvxjwxi , since aij = aji.
=Lu
v− uLv
v2− biwxi + aij
2
vvxjwxi
Therefore,
Mw := −aijwxixj + wxi[bi − aij 2
vvxj]
=Lu
v− uLv
v2≤ 0 on {x ∈ U : u > 0} ⊆ U
If {x ∈ U : u > 0} is not empty, Weak maximum principle to the operator M with bounded
coefficeints (since v ∈ C1(U)) will lead a contradiction that
0 < max{u>0}
w = max∂{u>0}
w =0
v= 0
Hence u ≤ 0 in U .
6
13. (Courant-Fischer’s principle)
Proof. First we need to note the same proof of Theorem 2 on Page 356 leads the following
theorem:
Theorem 3. For the symmetric operator L = −(aijuxi)xj on a bounded smooth domain U where
(aij) is symmetric and satisfies the uniform ellipticity. Denote k-th zero Dirichlet eigenvalue
and eigenfunction by λk and ωk respectively, then
λk = min{B[u, u] | u ∈ H10 , ‖u‖L2 = 1, (u, ωi)L2 = 0, 1 ≤ i ≤ k − 1}
Through this theorem, we know the desired identity is true if we replace equality by ≤. Now
we prove the converse:
Given k − 1 dimensional subspace S. By spectral theory for compact self-adjoint operators,
there exists y = Σk1ciωi ∈ span{ω1 · · ·ωk} with ‖y‖L2 = 1, such that y ∈ S⊥, so
min{B[u, u] | u ∈ S⊥, ‖u‖L2 = 1} ≤ B[y, y] = Σk1|ci|2λi ≤ Σk
1|ci|2λk ≤ λk.
Since S is arbitrary chosen, we are done.
14. Proof. Let Y := {u ∈ C∞(U) | u > 0 in U, u = 0 on ∂U}. Let u1 > 0 be the zero Dirichlet
eigenfunction with respect to the eigenvalue λ1 which exists by Krein-Rutman theorem (Page
361, Theorme 3). By classical Schauder’s theory, we know u1 ∈ Y and therefore supu∈Y
infx∈U
Luu≥ λ1.
Given v ∈ Y , then w = v − u1 ∈ Y . Note that
Lv
v=L(u1 + w)
u1 + w= λ1 +
L(w)− λ1w
u1 + w
Consider the adjoint operator
L∗w = ∂ij(aijw)− ∂i(biw) + cw = aij∂ijw + [∂j(a
ij + aji)− ∂ibi]∂iw + (c+ ∂ijaij − ∂ibi)w,
the Krein-Rutman theorem tells us that, for m > supU |c+∂ijaij−∂ibi|, there exists λ∗1, such that
λ∗1 +m is the simple principle zero Dirichlet eigenvalue of L∗+m with positive eigenfunction u∗1.
Since λ∗1(u∗1, u1)L2 = (L∗u∗1, u1)L2 = (u∗1, Lu1)L2 = λ1(u∗1, u1)L2 , the fact that u∗1, u1 > 0 implies
λ1 = λ∗1. Consider
(L(w)− λ1w, u∗1) = λ∗1(w, u∗1)− λ1(w, u∗1) = 0
Due to u∗1 > 0, there are only two possibilities:
(1) If L(w)− λ1w ≡ 0. Since λ1 is simple, w = u1. Then Lvv
= L(2u1)2u1
= λ1.
7
(2) If L(w) − λ1w changes sign in U . Then there exist x ∈ U such that (L(w) − λ1w)(x) < 0
and hence L(v)(x)v(x)
= λ1 + (L(w)−λ1w)(x)(u1+w)(x)
< λ1. Since v is arbitrary, supu∈Y
infx∈U
Luu≤ λ1.
Hence supu∈Y
infx∈U
Luu
= λ1.
15. Proof. (We use Reynolds transport theorem without mentions. )
λ(τ) = λ(τ)
∫U(τ)
|ω(x, τ)|2 =
∫U(τ)
−(∆xω)(x, τ)ω(x, τ)
=
∫U(τ)
|∇xω(x, τ)|2.
Differentiate the original equation and ‖w‖L2(Uτ ) = 1 with respect to τ , we get
−∆xωτ = λωτ + λω
0 =
∫∂U(τ)
ω2v · ν +
∫U(τ)
(ω2)τ = 0 +
∫U(τ)
(ω2)τ ,
Hence,
λ(τ) =
∫∂U(τ)
|∇xω|2v · ν +
∫U(τ)
(|∇xω|2)τ
=
∫∂U(τ)
|∂ω∂ν|2v · ν +
∫U(τ)
2∇xω∇xωτ , Since u = 0 on ∂U.
=
∫∂U(τ)
|∂ω∂ν|2v · ν +
∫U(τ)
2ω(−∆xωτ )
=
∫∂U(τ)
|∂ω∂ν|2v · ν +
∫U(τ)
2ω(λωτ + λω)
=
∫∂U(τ)
|∂ω∂ν|2v · ν + λ · 0 + 2λ
16. Proof. Direct computations tell us e−iσtv := eiσ(ω·x−t) satisfies the wave equation utt−σ2∆u = 0
and r(∂v∂r− iσv) = iσ(x · ω − |x|)v which did not tend to 0 as |x| → 0. On the other hand,
r(∂Φ∂r−iσΦ) = −Φ which tends to 0 as |x| → ∞. The proof to show Φ is a fundamental solution
is almost the same as usual Laplacian case. Try to mimic the proof Evans gives on page 24.
17. I think we need to assume w = O(R−1). and the second expression Evans gives in this problem
is wrong, the integrand on the left-hand side should be |wr|2 + σ2|w|2 − 2σIm(wwr).
Proof. Given x0 ∈ R3, let BR denote the ball centered at x0 with radius R and Φx0(x) :=
Φ(x− x0), where Φ is defined in Exercise 16 (b). Hence
w(x0) =
∫BR
[(∆ + σ2)Φx0
]w −
[(∆ + σ2)w
]Φx0 =
∫∂BR
∂Φx0
∂rw − ∂w
∂rΦx0 (3)
=
∫∂BR
[∂Φx0
∂r− iσΦx0
]w −
[∂w∂r− iσw
]Φx0 (4)
8
the magnitude of the integrand is o(R−1) ∗O(R−1), so its integral tends to 0 as R→∞.
Remark 4. Further results for Radiation condition on general dimension N ≥ 2 can be found
in G. Eskin, Lectures on Linear PDEs, Section 19. (Need some knowledges on stationary phase
formula.)
9