Exam | Partial Differential Equations (CES+SISC) | 16.09.

2022

Aufgabe 3.
Let V be a Hilbert space and consider a bilinear form a : V ⇥ V ! R as well as a linear
form b : V ! R that satisfy the conditions of the Lax-Milgram theorem. Let also VN be an
N -dimensional subspace of V .

(a) Derive a linear system of equations for the Ritz-Galerkin solution uN 2 VN of the
variational formulation in V defined by a and b.

(b) Which properties does the Ritz-Galerkin matrix AN have? Is the linear Ritz-Galerkin
system uniquely solvable?
(c) Consider the following minimization problem:
Z
1
min J(v), J(v) = krvk2L2 (⌦) vdx,
v2V 2 ⌦

where V = H01 (⌦) and ⌦ = (0, ⇡) ⇥ (0, ⇡) ⇢ R2 . Use the M -dimensional subspace VM
spanned by the following basis functions:

i,j (x, y) = sin(ix) sin(jy), 1  i, j  N, M = N 2,

to derive elements of the corresponding Ritz-Galerkin stiffness matrix and right hand
side.

4 + 2 + 4 = 10 Points

a) uNM- o: Yi( with asei

Then Find UNGU Such that

FreUr elv) b
a(naw)= l(v)
=

=>
a(4i,zajni(x) lug) =

=-
Eale:, ei) elei) aj
=

for 1...
i=

aisj=b;for i 1...N ais= ul4i,45) bi l4)

=

(
=

-]

ERNN b (l)ei):= ERN 2 (x:) ERN

==(ii)
i 1,...,
=

= =

1..,
*

is =n....,

Ac b
=

b) symmetric
and positiv definite
En is

N unhous 1,--IN
for
unique solution, Nequation
aii
=

be a
-> the system
even

6/17
 Exam | Partial Differential Equations (CES+SISC) | 16.09.2022
Name: Matriculation-Nr.:

conclude
c) J(v) I alvir)-l(V)
=
we

alvir-Erio=SFr. ürdel
R
-

acars=ftuor dr drEddy

-4i=(sclin) I
icso(ix) siniy)
el)=/vdak f = 1
=

r sinlix

Diedi s undDiali,4uel jR
g and bitte -
(n) sinlis) selln) jeroline e
-
=0
fori H
sind) sinux
=0
forjfl didy
·o
=

süliemlich
Care(i,s) (n,t)
=

-
da de

consideraIi asixsinsys ards

in
-

as Tanin an de

=in [; sin(2ix) J.*[

+
sin (25) +E].
Vi N
with sin (0) 0
=
Sin (2i+) 0
=
= 1 ...

e
-in

ansitionen sizes Aiche) if (i,j) (n,e) =

b14)= 4isdrdy= sin(ixl sinkin) dxdy

-sincial de
Isinly) dy

-alix)! -ein)!
=I alixcoolsy)! cas(0) 1 =

cos(ax) 1
=
with ever
a

1)
i((in) 1)(((jt) with aodd
-

(at) =-1
-

=
cas

I
jadsodd. else O
if 7/17
Exam | Partial Differential Equations (CES+SISC) | 16.09.2022

Aufgabe 4.
For ⌦ ⇢ R2 , we consider a uniform quadrilateral grid that is build from squares of equal
edge length h. On each square we have a bilinear ansatz function

p(x, y) = ↵ + x + y + xy

with parameters ↵, , , 2 R. As numerical degree of freedoms for an unknown function,

we use the function values at the vertices of the squares.

Abbildung 1: Square Q of edge length h with nodes labelled 1, 2, 3, 4.

(a) Consider the square Q with vertices labelled 1, 2, 3, and 4 as displayed in Figure 1.
Show that for given point values u1,2,3,4 , a unique bilinear function can be constructed.

(b) On quadrilateral meshes, we typically consider the bilinear finite element space, i.e.,
the space of continuous piecewise bilinear functions, which is equipped with a basis
of bilinear hat functions.

Let '1 (x, y) and '2 (x, y) denote the bilinear hat functions associated with the nodes 1
and 2 respectively and let N1 (x, y) and N2 (x, y) denote the element form functions on
the square Q associated with the nodes 1 and 2 respectively. Find explicit expressions
for N1 (x, y) and N2 (x, y).

(c) Consider the bilinear form

Z
a(u, v) = ru · rvdx
⌦

Compute the contribution of element Q to the matrix entry A1,2 = a('1 , '2 ).

3 + 2 + 3 = 8 Points

8/17
 Exam | Partial Differential Equations (CES+SISC) | 16.09.2022
Name: Matriculation-Nr.:

N p2(hio) 42 P5 (0,h) u
pnl0,0)
=
=

a) bohing
for 5,8 that un
=

sach
=

a,,
Palhih) =
un

p(0,0) un
= a
=
un
=

Gth uz
=

B
=
=

ai
pin,0)
=

4z
=

25 4
fh uz
= =

ct
=

pl0,h)
=
uz
=

Bh jh ch2 nu
=

pluh)
+

a
+

un
= =
+

Oh
Unt(uz-an)+(ug-un) un
=
+

Iu
+
E =

b) NCx,Y) NaCXiY) in D V(X;) dij

=

↓
=1
N.*(0,0) Uiki) di
=
un = 1 die to

Na(x,y) x y 4 2x y Ezxy
+

u 1
+
- -
=
- +
-

N,(4,0)=n
o
Uz
=

1
=

viez

2xy 2x xy
N,9(44) 4x
-
-

0
= +
+
0 -

c) almi) =

in. ov dx -was -

(E) -man-
(*)
Anz a(y,an)
=

AYz=JIN() Fr(A) dx
·

-41 -

2 +
4)( 24) -
1 2
+
-

+
2x)) x)dx dy -

=(7 +

5y 5y 2,44 +
-

(53x 24x)
+
- dr dy

-IE+y-Ey+EX- *drdy 9/17

=( -

2x 5 +
-

44]IE Ede -

-1 -

7 24
+
-

54) (E- in)

+

de

[ zy 2
=
- +
-

5. E- ]
=( -

1 1
+ -

5) (=+
-

5) 2 3 2
=
- =
-

2 =