Technische Universitt Mnchen WiSe 2011/12

Fakultt fr Informatik Solution for Exercise Sheet 1

Dr. Tobias Lasser
Richard Brosig, Jakob Vogel
October 25, 2011
Exercises in Basic Mathematical Tools
Assignment 3 Cauchy-Schwary Inequality (Exemplary Solution)
Reminder Scalar Product
For two vectors x, y R
n
, dene the scalar product
x, y :=
n

i=1
x
i
y
i
with the following properties x, x
/
, y R
n
, R:
a) Linearity
x+x
/
, y =x, y +x
/
, y
x, y =x, y
b) Commutativity
x, y =y, x
c) Positive deniteness
x, x 0
x, x = 0 x = 0 R
n
Proof Cauchy-Schwarz Inequality
x, y R
n
and R, it holds:
0 x+y, x+y =x, x +2x, y +
2
y, y (1)
Consider this equation in the special case y, y = 0, then:
y = 0
and thus
0 x, x x R
n
The Cauchy-Schwarz Inequality holds in this special case!
Therefore, let now w.l.o.g.
1
y, y > 0 and select
=
x, y
y, y
R.
1
without loss of generality; general and well-used remark that a restriction can be made as the other cases have
already been dealt with
Solution 1/ page 2
Insert this into equation (1):
0 x, x 2
x, y
2
y, y
+
x, y
2
y, y
2
y, y [ y, y(> 0!)
0 x, xy, y x, y
2
x, y
2
x, xy, y
This is the Cauchy-Schwarz Inequality in the general case!
Assignment 4 Function Space (Exemplary Solution)
a) As discussed in the lecture, the function space X =R
R
is a vector space when dening the
necessary operations (addition and scalar multiplication) in a point-wise manner as
( f +g)(x) := f (x) +g(x)
(f )(x) := f (x)
for arbitrary vectors (mappings) f , g X and parameters x R.
A mathematical proof of this fact requires checking all the vector space properties for the
function space X and is a rather lengthy procedure (and omitted here).
b) In order to show that a certain subset S X is a subspace, we need to show that all subspace
properties hold for that set. The contrary can, of course, be shown with a single counter-
example.
Reminder Subspace
Let V be a linear space. A nonempty set S V is called subspace of V if
x, y S xy S
x S, R x S
Note that the rst property is an elegant way of requiring both, completeness with respect to
addition and existence of the inverse element. (Subtraction does not work otherwise!) The
requirement of the zero element forming part of the subspace is implicitly contained in the
second property, for = 0.
(i) S := f X[ f (0) = 1+ f (1) X is no subspace.
Dene
f (x) =

1 x = 0
0 else
,
then f S, but 2 f / S as (2 f )(0) = 2 f (0) = 2 ,= 1.
(ii) S := f X[2 f (1) = f (1) X is a subspace.
( f g) S:
( f g)(1) = f (1) g(1) = 2 f (1) 2g(1) = 2( f g)(1)
(f ) S:
(f )(1) = f (1) = 2f (1) = 2(f )(1)
Solution 1/ page 3
(iii) S := f X[ f (1) = 0 X is a subspace.
( f g) S:
( f g)(1) = f (1) g(1) = 0
(f ) S:
(f )(1) = f (1) = 0
(iv) S := f X[ f (x) = f (1x) x R X is a subspace.
( f g) S:
( f g)(x) = f (x) g(x) = f (1x) g(1x) = ( f g)(1x)
(f ) S:
(f )(x) = f (x) = f (1x) = (f )(1x)
Note that these equalities are directly derived from the denitions of the addition and
scalar multiplication operations.
(v) S := f X[ f (x
3
) = f (x)
5
x R X is no subspace.
Dene the constant function
f (x) = 1,
then f S, but 2 f / S as (2 f )(x
3
) = 2 f (x
3
) = 2 ,= (2f )(x)
5
= (2f (x))
5
= 2
5
.
(vi) S := f X[ f is continuous X is a subspace.
Both, addition and scalar multiplication, are continuous functions themselves, and ap-
plying them to further continuous functions does not change this property.
Two mappings f , g S are continuous by denition of the set, and therefore ( f g) S
and (f ) S.
These concepts will be discussed in detail in the lectures calculus part!
Assignment 5 Linear Dependence (Exemplary Solution)
In general, the question (for whether a linear combination exists) yields a system of linear equa-
tions
2
:
v
i
= x
i
+y
i
+z
i
i = 1, 2, 3
For the rst set x = (1, 1, 0)

, y = (1, 0, 1)

, z = (0, 1, 0)

, one gets a particularly simple system:

3 = +
0 = +
2 =
From these equations, it is simple to solve for and :
= 3 = 1
= =1
It is thus possible to express v as linear combination of these x, y, z.
2
also called linear system
Solution 1/ page 4
It is advisable to change (even standardize) the notation to matrix-vector-form. Then, the general
system is:

x
1
y
1
z
1
x
2
y
2
z
2
x
3
y
3
z
3

=: A

=: x
=

v
1
v
2
v
3

=: b
The names A, x (unknowns), and b (right-hand side) are commonly used to refer to the three
components of the system, and there are many general (and specialized) approaches to solving
such problems. Some are going to be discussed in the lecture.
For the rst set, the matrix form is:

1 1 0
1 0 1
0 1 0

3
0
2

1 1 0 3
1 0 1 0
0 1 0 2

The best-known solution algorithm is probably Gaussian Elimination which attempts to bring the
system into triangular or even diagonal shape by adding up multiples of single rows, commuting
rows, etc. Once that simple shape is reached, it is easy to get the solution.
a)

1 1 0 3
1 0 1 0
0 1 0 2

IIII

1 0 0 1
1 0 1 0
0 1 0 2

III

1 0 0 1
0 0 1 1
0 1 0 2

IIIII

1 0 0 1
0 1 0 2
0 0 1 1

The three vectors x (red), y (green) and z (blue) are not linearly dependent themselves (i.e.
in general conguration) and form a basis of R
3
, however, neither an orthogonal nor an
orthonormal one.
b)

1 2 1 3
1 3 0 0
0 1 1 2

III

1 2 1 3
0 1 1 3
0 1 1 2

II+III

1 2 1 3
0 0 0 1
0 1 1 2

In the second row of the system, a problematic situation has occurred. That part is logically
equivalent to the statement 0 =1 which is always wrong. Consequently, the system does
not have a solution.
The three vectors x (red), y (green) and z (blue) turn out to be linearly dependent. They
are no basis and do not generate R
3
, but just a two-dimensional plane within the three-
dimensional space. Such matrices are called singular and said to be rank-decient.
Solution 1/ page 5
The right-hand-side (black) points to a location outside of that plane, a solution is therefore
impossible to exist. The three vectors just miss the information of the error vector (red,
solid).
c)

1 0 1 3
0 0 0 0
0 1 1 2

This time, the matrix is also singular, but the right-hand side vector (black) happens to
point into the plane generated by the selection of vectors x (red), y (green) and z (blue).
It is thus possible to express the right-hand side as linear combination of the three vectors,
however, the solution is not unique. The reason for this phenomenon is that the third vector
contains information already available through the rst two, and is thus redundant.
One can consequently nd an innite set of solutions, for instance (2, 1, 1)

(orange) or
(4, 3, 1)

(violet).