Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more ➡
Standard view
Full view
of .
×
0 of .
Results for:
P. 1
phase plane analysis

# phase plane analysis

Ratings: 0|Views: 182|Likes:

### Availability:

See More
See less

01/17/2013

pdf

text

original

Chapter 1
Basic Notions of Phase SpaceAnalysis
1.1 Introduction to pseudodi
ﬀ
erential operators
1.1.1 Prolegomena
A di
ﬀ
erential operator of order
m
on
R
n
can be written as
a
(
x,D
) =
|
α
|
m
a
α
(
x
)
D
α
x
,
where we have used the notation (4.1.4) for the multi-indices. Its
symbol
is apolynomial in the variable
ξ
and is deﬁned as
a
(
x,
ξ
) =
|
α
|
m
a
α
(
x
)
ξ
α
,
ξ
α
=
ξ
α
1
1
...
ξ
α
n
n
.
We have the formula
(
a
(
x,D
)
u
)(
x
) =

R
n
e
2
i
π
x
·
ξ
a
(
x,
ξ
u
(
ξ
)
d
ξ
,
(1.1.1)where
ˆ
u
is the Fourier transform as deﬁned in (4.1.1). It is possible to generalizethe previous formula to the case where
a
is a tempered distribution on
R
2
n
.Let
u,v
be in the Schwartz class
(
R
n
)
. Then the function
R
n
×
R
n
(
x,
ξ
)
ˆ
u
(
ξ
v
(
x
)
e
2
i
π
x
·
ξ
=
u,v
(
x,
ξ
)
(1.1.2)belongs to
(
R
2
n
)
and the mapping
(
u,v
)
u,v
is sesquilinear continuous.Using these notations, we can provide the following deﬁnition.

2
Chapter 1. Basic Notions of Phase Space Analysis
Deﬁnition 1.1.1.
Let
a
(
R
2
n
)
be a tempered distribution. We deﬁne theoperator
a
(
x,D
) :
(
R
n
)
(
R
n
)
by the formula
a
(
x,D
)
u,v
(
R
n
)
,
(
R
n
)
=
a,
u,v
(
R
2
n
)
,
(
R
2
n
)
,
where
(
R
n
)
is the antidual of
(
R
n
)
(continuous antilinear forms). The dis-tribution
a
is called the symbol of the operator
a
(
x,D
)
.
N.B.
The duality product
u,v
(
R
2
n
)
,
(
R
2
n
)
,
is linear in the variable
u
and an-tilinear in the variable
v
. We shall use the same notation for the dot product inthe
complex
Hilbert space
L
2
with the notation
u,v
L
2
=

u
(
x
)
v
(
x
)
dx.
The general rule that we shall follow is to always use the sesquilinear duality asabove, except if speciﬁed otherwise. For the real duality, as in the left-hand-side of the formula in Deﬁnition 1.1.1, we shall use the notation
u,v
=

u
(
x
)
v
(
x
)
dx
,e.g. for
u,v
(
R
n
)
.Although the previous formula is quite general, since it allows us to
quantize
1
any tempered distribution on
R
2
n
, it is not very useful, since we cannot composethis type of operators. We are in fact looking for an algebra of operators and thefollowing theorem is providing a simple example.In the sequel we shall denote by
b
(
R
2
n
)
the (Fréchet) space of
functionson
R
2
n
which are bounded as well as all their derivatives.
Theorem 1.1.2.
Let
a
b
(
R
2
n
)
. Then the operator
a
(
x,D
)
is continuous from
(
R
n
)
into itself.Proof.
Using Deﬁnition 1.1.1, we have for
u,v
(
R
n
)
,a
b
(
R
2
n
)
,
a
(
x,D
)
u,v
(
R
n
)
,
(
R
n
)
=

e
2
i
π
x
·
ξ
a
(
x,
ξ
u
(
ξ
v
(
x
)
dxd
ξ
.
On the other hand the function
(
x
) =

e
2
i
π
x
·
ξ
a
(
x,
ξ
u
(
ξ
)
d
ξ
is smooth and suchthat, for any multi-indices
α
,
β
,
x
β
D
α
x
(
x
) = (
1)
|
β
|
α
+
α

=
α
α
!
α
!
α

!

e
2
i
π
x
·
ξ
D
βξ
ξ
α
(
D
α

x
a
)(
x,
ξ
)ˆ
u
(
ξ
)
d
ξ
= (
1)
|
β
|
α
+
α

=
α
α
!
α
!
α

!

e
2
i
π
x
·
ξ
D
βξ
(
D
α

x
a
)(
x,
ξ
)

D
α
u
(
ξ
)
d
ξ
and thus
sup
x
R
n
|
x
β
D
α
x
(
x
)
|
α
+
α

=
αβ
+
β

=
β
α
!
α
!
α

!
β
!
β
!
β

!
D
β
ξ
D
α

x
a
L
(
R
2
n
)
D
β

D
α
u
L
1
(
R
n
)
.
1
We mean simply here that we are able to deﬁne a linear mapping from
(
R
2
n
)
to the setof continuous operators from
(
R
n
)
to
(
R
n
)
.

1.1. Introduction to pseudodi
ﬀ
erential operators
3Since the Fourier transform and
∂
x
j
are continuous on
(
R
n
)
, we get that themapping
u
is continuous from
(
R
n
)
into itself. The above deﬁning formulafor
a
(
x,D
)
ensures that
a
(
x,D
)
u
=
.
The Schwartz space
(
R
2
n
)
is not dense in the Fréchet space
b
(
R
2
n
)
(e.g.
ϕ
(
R
2
n
)
,
sup
x
R
2
n
|
1
ϕ
(
x
)
|
1
) but, in somewhat pedantic terms, one maysay that this density is true for the bornology on
b
(
R
2
n
)
; in simpler terms, let
a
be a function in
b
(
R
2
n
)
and take for instance
a
k
(
x,
ξ
) =
a
(
x,
ξ
)
e
(
|
x
|
2
+
|
ξ
|
2
)
k
2
.
It is easy to see that each
a
k
belongs to
(
R
2
n
)
, that the sequence
(
a
k
)
is boundedin
b
(
R
2
n
)
and converges in
(
R
2
n
)
to
a
. This type of density will be enoughfor the next lemma.
Lemma 1.1.3.
Let
(
a
k
)
be a sequence in
(
R
2
n
)
such that
(
a
k
)
is bounded in the Fréchet space
b
(
R
2
n
)
and
(
a
k
)
is converging in
(
R
2
n
)
to a function
a
. Then
a
belongs to
b
(
R
2
n
)
and for any
u
(
R
n
)
, the sequence
(
a
k
(
x,D
)
u
)
converges to
a
(
x,D
)
u
in
(
R
n
)
.Proof.
The fact that
a
belongs to
b
(
R
2
n
)
is obvious. Using the identities in theproof of Theorem 1.1.2 we see that
x
β
D
α
x
a
k
(
x,D
)
u
a
(
x,D
)
u
=
x
β
D
α
x
(
a
k
a
)(
x,D
)
u
= (
1)
|
β
|
α
+
α

=
αβ
+
β

=
β
α
!
α
!
α

!
β
!
β
!
β

!

e
2
i
π
x
·
ξ
D
β
ξ
D
α

x
(
a
k
a
)
(
x,
ξ
)
D
β

ξ

D
α
u
(
ξ
)
d
ξ
=
α
+
α

=
αβ
+
β

=
β
α
!
α
!
α

!
β
!
β
!
β

!(1 +
|
x
|
2
)
1
×

(1 +
|
D
ξ
|
2
)
e
2
i
π
x
·
ξ

D
β
ξ
D
α

x
(
a
k
a
)
(
x,
ξ
)
D
β

ξ

D
α
u
(
ξ
)
d
ξ
,
that is a (ﬁnite) sum of terms of type
k
(
x
) = (1+
|
x
|
2
)
1

e
2
i
π
x
·
ξ
b
k
(
x,
ξ
)
w
u
(
ξ
)
d
ξ
with the sequence
(
b
k
)
bounded in
b
(
R
2
n
)
and converging to 0 in
(
R
2
n
)
,
u
w
u
linear continuous from
(
R
n
)
into itself. As a consequence we get that,with
R
1
,R
2
positive parameters,
|
k
(
x
)
|
sup
|
x
|
R
1
|
ξ
|
R
2
|
b
k
(
x,
ξ
)
|

|
ξ
|
R
2
|
w
u
(
ξ
)
|
d
ξ
1
{|
x
|
R
1
}
+

|
ξ
|
R
2
|
w
u
(
ξ
)
|
d
ξ
sup
k
N
b
k
L
(
R
2
n
)
1
{|
x
|
R
1
}
+
R
21
1
{|
x
|
R
1
}
sup
k
N
b
k
L
(
R
2
n
)

|
w
u
(
ξ
)
|
d
ξ
,

## Activity (2)

### Showing

AllMost RecentReviewsAll NotesLikes