Professional Documents
Culture Documents
Elena Beretta
Dipartimento di Matematica
Politecnico di Milano (ITALY)
elena.beretta@polimi.it
1 / 82
Part I
Prerequisits
3 / 82
Banach Spaces
X vector space on R
A norm on X is a function k k : X [0, +) s.t.
kxk 0 and kxk = 0 if and only if x = 0 (positivity)
kxk = ||kxk, x X , R (homogeneity)
kx + y k kxk + ky k, x, y X (triangular inequality)
(X , k k) is a normed space
(X , k k) is a metric space (X , d) w.r.t. d induced by
d(x, y ) = kx y k, x, y X
xn x in X if kxn x k 0 as n +
(strong convergence)
A Banach space is a complete normed space
(any Cauchy sequence is convergent in X )
4 / 82
Examples
(Rn , k q
kn ) and (Cn , k kn ) are Banach spaces with respect to
Pn
2
kxk =
i=1 xi where x = (x1 , . . . , xn ).
C 0 (K ) = {u : Rn R continuous on K }, K is compact
kukC 0 () = kuk = sup |u(x)|,
x
(C 0 (K ), k k ) is a Banach space
1/p
R
Lp () = {u : |u(x)|p dx
< +},
R
1/p
kukp = |u(x)|p dx
1 p < +
5 / 82
Examples
{xk } = (x1 , x2 , . . . ),
1 p < +,
(
p
` =
x = {xk } :
xk R
)
p
|xk | < +
k =1
P
p 1/p
(`p , k k) is a Banach space with kxkp = ( +
k =1 |xk | )
6 / 82
Hilbert Spaces
X vector space on R
An inner product on X is a function (, ) : X X R s.t.
(x, x) 0, x X , (x, x) = 0 iff x = 0 (positivity)
(y , x) = (x, y ), x, y X (symmetry)
(x + y , z) = (x, z) + (y , z), x, y , z X , , R (bilinearity)
7 / 82
Hilbert Spaces
Cauchy-Schwarz inequality
8 / 82
Rn (x, y ) =
Cn
(x, y ) =
Pn
Pi=1
n
xi yi ,
i=1 xi y i
L2 () is Ra Hilbert space
R
(u, v ) = u(x)v (x)dx, (u, v ) = u(x)v (x)dx
`2 is a Hilbert space
X
X
(x, y) =
xk yk
xk y k
k =1
k =1
9 / 82
Orthogonal projections
kPV xk kxk
10 / 82
11 / 82
X
(x, ej )ej , x H
j=1
cj2
j=1
Rn , Cn
e1 = (1, 0, . . . , 0), . . . , e2 = (0, 1, 0 . . . , 0) . . .
L2 (0, 2)
e0 = 1 , e1 =
2
`2
e1 = (1, 0, . . . ),
cos x, e2 =
sin x, e3 =
cos 2x . . .
e2 = (0, 1, . . . ) . . .
13 / 82
x, z H1
Kernel of L (N (L))
N (L) = {x H1 : Lx = 0}
Range of L (R(L))
R(L) = {y H2 : x H1 : Lx = y }
L is bounded if M > 0 : kLxkY MkxkH1 , x H1
L is continuous if xn x in H1 Lxn Lx in H2
14 / 82
kLxkH2
= sup kLxkH2
kxkH1
kxkH =1
1
15 / 82
L : Rn Rm
Lx = Ax
where A is an m n real valued matrix.
V H, closed subset of a Hilbert space H.
L1 : H H with L1 x = PV x and L2 : H H with L1 x = QV x
16 / 82
Theorem
L(H1 , H2 ) is a Banach space
Theorem
L : (H1 , k kH1 ) (H2 , k kH2 ), L linear.
L is continuous L is bounded.
(with proof)
17 / 82
(H , k
kH ) is a Banach space
H h, iH
H hLxiH
: H H R
|Lx|
kxkH
(R is a Banach space)
H hL, xiH
= Lx is a bilinear form
or hL, xi (duality)
18 / 82
19 / 82
Remark
Riesz Theorem we identify H and H endowing H with the
inner product (L1 , L2 ) = (yL1 , yL2 ).
Exercise: Prove that the kernel of a linear bounded operator is a
closed vector space.
Exercise: Show that the Representation Theorem holds also in
complex Hilbert spaces.
20 / 82
Bilinear Forms
Definition
Let V be a real vector space. a : V V R is a bilinear form on V if
y V the function x a(x, y ) is linear
x V the function y a(x, y ) is linear
Example
The inner product on a real vector space is a bilinear form
In complex inner product spaces we define sesquilinear forms
substituting the linearity with respect to the second component with
antilinearity
21 / 82
Lax-Milgram Theorem
Problem (P):
Let F H . Find u H such that
a(u, v ) = hF , v i, v H
22 / 82
Theorem
Lax-Milgram Theorem Let H be a real Hilbert space, a be a bilinear
form such that
a is continuous i.e. M > 0:
|a(u, v )| Mkukkv k,
u, v H
v H.
1
kF k
23 / 82
Remark
If M > 0:
|a(u, v )| Mkukkv k,
u, v H
Remark
Lax-Milgram Theorem still holds in complex Hilbert spaces for
sequilinear forms which are continuous and coercive.
24 / 82
Minimization Problem
1
a(v , v ) hF , v i
2
24 / 82
v H
v Vk
Show
kuk ukH 0
as k +.
25 / 82
Proposition
Assume the assumptions of the Lax-Milgram Theorem are satisfied
and let u be the unique solution to Problem (P). If uk is the unique
solution of Problem (Pk ) then
ku uk k
M
inf ku v k
v Vk
Theorem
uk u in V as k +
with proof
26 / 82
Definition
H Hilbert space, a sequence {xn } of elements in H converges strongly
to an element x H
xn x
if
kxn xkH 0
as n +
Definition
Let H be a Hilbert space, xn , x H.
xn * x (weak convergence) if (xn , y ) (x, y ) for any y H.
27 / 82
Proof.
If xn x strongly then
|(xn , y )(x, y )| = |(xn x, y )| kxn xkky k 0, y H
xn * x
xn , x H.
28 / 82
Counterexample
Example
H = `2
y `2
2
y`
yk2 < +
yk 0
k =1
ek orthonormal basis
(ek , y) = yk 0, y `2
ek 9 0 strongly in `2
(ek is not a Cauchy sequence:
ek * 0
kek ej k`2 =
2,
k 6= j )
29 / 82
Proof.
Assume xn * x. Then
(x x, y ) = (x xn , y ) (x xn , y ) 0, y H
Hence
(x x, y ) = 0, y H
which implies
x = x.
30 / 82
Theorem
H Hilbert space, xn H.
If xn * x H then xn is bounded
31 / 82
Theorem
H Hilbert space, xn H.
If xn * x H then kxk lim infn kxn k
Proof.
xn * x
32 / 82
Compactness
33 / 82
Remark
If E is a subset of a finite dimensional normed space (Rn ) then
E sequentially compact Eclosed and bounded
In particular
B1 = {x Rn : kxk 1}
is compact.
Theorem
A normed space is finite dimensional B1 is compact.
34 / 82
2 n, m, n 6= m
so {ej }
1 is not a Cauchy sequence.
Theorem
Any bounded sequence in a Hilbert space H has a weak convergent
subsequence in H.
35 / 82
Distributions
Definition
Given a continuous function : Rn R the support of is the set
supp = {x : (x) 6= 0}
Definition
We indicate by C0 () the set of functions in C () compactly
supported in
36 / 82
Distributions
We now endow C0 () with a suitable notion of convergence (not
inducing any metric)
Definition
{k }
k =1 C0 () , C0 ().
k in C0 () as k
if
D k D ,
uniformly in
and there exists a compact set K which contains the support of all
the k s.
37 / 82
Distributions
Example
Let {ak } R, ak 0 as k , let C0 (). Then
k = ak 0, k
in the above sense.
38 / 82
Distributions
Definition
A distribution f is a linear continuous functional on D() (f D0 ())
i.e. such that
< f , 1 + 2 >= < f , 1 > + < f , 2 >, 1 , 2
D(), , R
< f , k >< f , > as k in D() as k +.
39 / 82
Distributions
Examples
f C 0 () defines a distribution
Z
< f , >=
since
Z
|
f | kf kK sup ||
f L2 () defines a distribution
Z
< f , >=
since
Z
|
f | kf kL2 () kkL2 ()
40 / 82
Distributions
Examples
f = 0 is a distribution
< f , >= (0)
since
|(0)| sup ||
f L1loc () is a distribution
Z
< f , >=
since
Z
|
f | kf kL1 (K ) sup ||
1
|x|p
D0 (Rn )
41 / 82
Distributions
Theorem
The map I : L2 () D0 () where
Z
< I(f ), >=
f dx
is injective
with proof
42 / 82
Distributional derivatives
Let f C 1 (R) with and D(R). Then
Z
Z
Z
+
0
0
f dx = f |
f dx = f 0 dx
R
Hence
< f 0 , >= < f , 0 >
Definition
Let f D0 () where Rn . The derivative xi f is the distribution
defined by
< xi f , >= < f , xi >
Every distribution possesses (distributional) derivatives of any order
< D f , >= (1)|| < f , D >
43 / 82
Examples
f = 0
< xi 0 , >= xi (0)
f = H (Heaviside function)
Z
< H, >= < H, >=
R
H 0 dx =
0 dx = (0)
Hence
< H0 , >= (0) =< , >
Exercise: Compute the derivative of f (x) = |x| D(R)
44 / 82
Z
< xi v , >=
v xi dx, D0 ()
Theorem
H 1 () is a Hilbert space with inner product
Z
Z
(u, v )1 =
uv dx +
u v dx
and
sZ
kuk1 =
|v |2 dx
Z
+
|v |2 dx
45 / 82
Theorem
Poincar inequality. Let Rn . There exists a positive constant CP
such that
kuk0 CP kuk0
We can choose
(u, v )1 = (u, v )0
as inner product
46 / 82
Definition
We denote by H 1 () the dual of H01 () with
kF k1 = sup{|Fv | : v H01 (), kv k1 1}
Theorem
H 1 () is the set of distributions of the form
F = f0 + div f
where f0 L2 () and f L2 ()
47 / 82
Part II
Elliptic and parabolic PDEs
48 / 82
Elliptic equations
Laplace equation
u = 0
Stationary diffusion equation, conductivity equation, equilibrium of an
elastic membrane
Poisson equation
u = f
f external force or source e.g. density of electric charges, load
49 / 82
Harmonic functions
Definition
A solution of
u = 0
is called harmonic.
Examples
z = ei , <z m and =z m , m N are harmonic functions
x, y , xy , x 2 y 2 , x 3 3xy 2 , . . . are harmonic functions
u (x) = ex , Cn : = 0, x Rn
50 / 82
Theorem
If u is harmonic in Rn , n 2. Then, for any ball BR (x)
Z
1
u(x) =
u(y )dy
|BR (x)| BR (x)
1
u(x) =
|BR (x)|
Z
u(y )dy
BR (x)
Theorem
If u C 0 () has the mean value property then u is C () and is
harmonic in .
51 / 82
Maximum principle
Theorem
If u C 0 () has the mean value property then, if u attains a maximum
or a minimum at x0 then u is constant. In particular if is
bounded, u C 0 () and u is not constant then, x
min u < u(x) < max u
52 / 82
53 / 82
u = f in Rn , n 2
u = g on .
Neumann Problem
u = f in Rn , n 2
u
n = h on .
Robin Problem
u
u = f in Rn , n 2
+ u = h on .
54 / 82
Well-posedness
Definition
A problem is well posed if
there exists a solution (existence)
the solution is unique (uniqueness)
the solution depends continuously upon the data (stability)
55 / 82
Uniqueness
Theorem
Rn bounded and smooth. There exists a unique solution
u C 2 () C 1 () of the Dirichlet and of the Robin problem. In the
case of the Neumann problem the solution is unique up to an additive
constant.
with proof
56 / 82
Existence
57 / 82
Theorem
Poisson formula The unique solution to the Dirichlet problem
u = 0 in BR (P) R2
u = g on BR (P).
is given by
R 2 |x P|2
u(x) =
2R
Z
BR (P)
g(y )
dy
|x y |2
with proof
58 / 82
59 / 82
2 u 2 u
=0
+
r u
r 2
Then
u(r ) = C1
1
+ C2
r
60 / 82
61 / 82
Theorem
Representation Formula
Rn smooth and bounded, u C 2 (). Then, x
Z
Z
Z
u
(x y )
u(x) =
(x y )u dy
(x y )
dy +
u
dy
62 / 82
Definition
Fix y
G(, y ) = y in
G(, y ) = 0 on .
ANSATZ:
G(x, y ) = (x y ) (x, y )
where
(, y ) = 0 in
(, y ) = (x y ) on .
63 / 82
G < 0 (negativity)
G(x, y ) = G(y , x) (symmetry)
|G(x, y )| + as x y (singularity)
For particular geometries of the Greens function can be computed
explicitly.
64 / 82
Representation formula
Theorem
Consider
u = f in
u = g on .
Then,
Z
u(x) =
Z
G(x, y )f (y ) dy +
If u is harmonic
Z
u(x) =
G(x, y )
dy
G(x, y )
dy
65 / 82
Representation formula
Theorem
Consider
Then,
u = f in
u
= h on .
Z
N(x, y )f (y ) dy
u(x) =
hN(x, y ) dy
66 / 82
n + = h on ,
k diffusion coefficient, absorption coefficient, optical density, h
illumination function.
67 / 82
Definition
Consider Rn , A(x) = (aij (x))ni,j=1 , c(x) = (ci (x))ni=1 , a0 (x) and f (x)
scalar functions. An equation of the form
div (Au) + c u + a0 u = f
in
is elliptic in if
A(x) > 0, x , Rn , 6= 0
68 / 82
69 / 82
Variational formulation
Problem (DP)
Find u H01 () such that
Z
Z
au v + a0 uv =
fv , v H01 ()
Let
Z
au v + a0 uv
B(u, v ) :=
fv
Well-posedness
Theorem
Let f L2 () and 1
0 a 0 , 0 a0 (x) 0 in . Then, Problem
(DP) has a unique solution u H01 (). Moreover
kuk0 Ckf k0
71 / 82
Remarks
72 / 82
73 / 82
Z
au v + a0 uv =
Z
fv +
gv , v C 1 ()
u
a v=
gv v C 1 ()
which gives
a
u
=g
74 / 82
Let
gv , v H 1 ()
Z
au v + a0 uv
B(u, v ) :=
Z
fv +
gv
Well-posedness
Theorem
Let f L2 (), g L2 (), 0 < 1
0 a(x) 0 and
0 < 01 a0 (x) 0 in . Then, Problem (NP) has a unique solution
u H01 (). Moreover
kuk1
1
1
min{1
0 , 0 }
(kf k0 + CkgkL2 () )
76 / 82
Well-posedness
Remark
The condition a0 01 > 0 is necessary for existence and
uniqueness!
If a0 = 0 then uniqueness fails and a normalization condition is needed
to restore uniqueness.
Existence of solutions requires that the compatibility condition
Z
Z
f+
g=0
is satisfied.
77 / 82
Eigenvalues
Definition
Let Rn bounded. A nontrivial weak solution u H01 () to the
problem
u = u in
u = 0 on .
is called Dirichlet eigenfunction and the corresponding
Dirichlet eigenvalue
Theorem
There exists in L2 () an orthonormal basis {uk } consisting of Dirichlet
eigenfunctions. The corresponding eigenvalues are an increasing
sequence
0 < 1 < 2 k . . .
with k +. The sequence {uk /k } is an orthonormal basis in
H01 ().
78 / 82
Eigenvalues
Remark
If u L2 () then u =
ck uk and kuk2 =
ck2 . So
ck2 k <
79 / 82
Eigenvalues
By
kuk20 =
ck2 k <
|u| : u
1 = min
=1
80 / 82
Eigenvalues
Let F (u) =
2
|u|
and G(u) =
Theorem
Let u H01 () be a local extremum of the functional F subject to the
condition G(u) = 0, then u is an eigenfunction and = F (u) is the
corresponding eigenvalue.
Proof. Follows by using Lagrange multiplier Theorem
81 / 82
Eigenvalues
Theorem
There exists a global minimum u H01 () of the functional F subject to
the condition G(u) = 0.
Main steps of the proof.
82 / 82