0% found this document useful (0 votes)
42 views3 pages

Norms 2010

This document defines norms and normed spaces, which provide a way to measure the size of quantities in stability theory. It discusses vector and function norms, induced matrix norms, and the Lp gain of linear operators. Key results shown include that all norms are equivalent, and the Lp gain of a convolution-type operator H can be given in terms of the Laplace transform of its impulse response.

Uploaded by

Ciprian Biris
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
42 views3 pages

Norms 2010

This document defines norms and normed spaces, which provide a way to measure the size of quantities in stability theory. It discusses vector and function norms, induced matrix norms, and the Lp gain of linear operators. Key results shown include that all norms are equivalent, and the Lp gain of a convolution-type operator H can be given in terms of the Laplace transform of its impulse response.

Uploaded by

Ciprian Biris
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

29.

October 2010, TOA

NORMS AND
NORMED SPACES

This note states some basic results which are useful in stability theory for
providing a measure of the size of various quantities. For more details see
Desoer and Vidyasagar [1] and Vidyasagar [2].

Normed Spaces

A normed space is a vector space V with a given norm. A norm on a vector


space V is a rule which, given any x ∈V , specifies a real number x , such that

(a) x >0 if x ≠ 0, and 0 = 0;


(b) ax = a ⋅ x for any x ∈V and any scalar a ;
(c) x + y ≤ x + y for any x , y ∈V (the triangle inequality) ;

A normed vector space is called real or complex according to weather the


underlying vector space V is real or complex. Subsequent only norms defined
on ℜ n are considered.

Examples of norms which satisfy the axioms of the norm are

n
x 1 =$ ∑ xi (1)
i =1
1/ P
⎛ n P⎞
x =$ ⎜ ∑ xi ⎟ , where 1 ≤ P < ∞ (2)
P
⎝ i=1 ⎠
x ∞
=$ max xi (3)
i

x 2
is the usual Euclidean norm of x .

An important result is that all norms are equivalent in the sense that there exist
positive numbers κ 1 and κ 2 such that

κ1 x a ≤ x b ≤κ 2 x a, ∀x ∈ℜ n (4)

This means that any norm can be used to show boundedness or convergence.
Hence, often the symbol ⋅ is used without specifying exactly which norm is
meant.

Page 1 of 3
29. October 2010, TOA

Function Norms

In continuous-time systems, the inputs and outputs are functions of time t, and t
is usually restricted to t ≥ 0 . Consider functions mapping ℜ + to ℜ . They are
assumed to be piecewise continuous.
For any fixed p ∈[1, ∞) , f : ℜ + → ℜ belongs to LP if ∫0 f (t ) dt < ∞ . The
∞ p

associated norm is
1/ p
= ⎛⎜⎝ ∫ f (t ) dt ⎞⎟⎠
∞ p
f p
(5)
0

and f : ℜ + → ℜ belongs to L∞ if supt ≥0 f (t ) < ∞ , and the norm is

f ∞
= sup f (t ) (6)
t ≥0

Remark 1: Considering functions from ℜ + into ℜ n , all above definitions hold,


except that f (t ) is interpreted as the chosen norm of the vector f (t ) ∈ℜ n .

Remark 2: Associated with the normed space Lp is an extended space Lpe with
Lp ⊂ Lpe . In addition to the class of functions in Lp this extended space also
includes truncated functions of the form:

⎧ f (t ) t≤T
f T (t ) = ⎨ (7)
⎩0 t>T
such that f T p
< ∞ ; ∀T .

Induced Matrix Norms

For a matrix A ∈ℜ n× n a norm can be defined by

Ax
A p
=$ sup p
∀x ∈ℜ n (8)
x ≠0 x p

The norm is called the induced (matrix) norm of A corresponding to the vector
norm x p .
For the induced norm it can be shown that

AB p
≤ A p
B p, ∀ A, B ∈ℜ n×n (9)
Ax p
≤ A p
x p, ∀ A ∈ℜ n×n , x ∈ℜ n (10)

Page 2 of 3
29. October 2010, TOA

For p = 2 the induced norm can be given as A 2 = max λ i ( A T A) .


i

Gain of Linear Operators

Let H be a causal operator H : f → g . The Lp gain of H is denoted H p


and is
defined as the smallest value of γ such that

g p
= Hf p
≤γ f p
+β ∀ f ∈ Lp (11)

Consider the convolution-type operator H : f → g given by


g (t ) = ∫ h(t − τ ) f (τ ) dτ , ∀ t ∈ℜ (12)
−∞

Then for p = 2 the L2 gain is given by

H = max h$( jω ) (13)


2 ω ∈ℜ

where h$ ( s) is the Laplace transform of the impulse response of H .


For p = ∞ the L∞ is given by


H ∞
= ∫ h(t ) dt (14)
0
------
Having defined the Lp spaces the following lemma is useful in analyzing the
input-output properties of systems with exponentially stable transfer functions.

Lemma 1: (Desoer and Vidyasagar [1])

Let the transfer function H ( s) be exponentially stable and strictly proper.


Then f ∈ L2 implies that g = H ∗ f ∈ L2 I L∞ , g& ∈ L2 , g is continuous, and
g → 0 as t → ∞ . If, in addition, f → 0 as t → ∞ , then g& → 0 .

References

[1] Desoer, C., and Vidyasagar, M., “Feedback Systems: Input-Output Properties”,
Academic Press: New York, 1975.
[2] Vidyasagar, M., “Nonlinear Systems Analysis”, Prentice-Hall: Englewood Cliffs, N.J.,
1993.

Page 3 of 3

You might also like