Scribd
Upload
195 views3 pages

Theorem

This document discusses the Fourier transformation and the stationary phase method. It presents theorems and lemmas about estimating integrals involving oscillatory functions. It analyzes when integrals of this type are rapidly decreasing functions of a parameter and describes how contributions come from points where the phase is real and stationary.

Uploaded by

Dana Bman
Copyright
© © All Rights Reserved
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
195 views3 pages

Theorem

This document discusses the Fourier transformation and the stationary phase method. It presents theorems and lemmas about estimating integrals involving oscillatory functions. It analyzes when integrals of this type are rapidly decreasing functions of a parameter and describes how contributions come from points where the phase is real and stationary.

Uploaded by

Dana Bman
Copyright
© © All Rights Reserved
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

216 VII.

The Fourier Transformation

where f and u are smooth, Imf 0 and w + 00. If UE Cg', f E COO is


real valued and f' =1= 0 in supp u, then
Jueiwf dx=O(w- k), 00,

for every k. This follows from Lemma 7.1.3 if u has so small support
that f can be taken as a coordinate in a new local coordinate system
in a neighborhood of supp u. We can reduce to this situation by
decomposing u with a partition of unity. In the following theorem we
elaborate the integration by parts in Lemma 7.1.3 to obtain uniform
bounds and to cover the case of complex valued f also.

Theorem 7.7.1. Let K e1R." be a compact set, X an open neighborhood


of K and j, k non-negative integers. If uEq(K), fEC k +1 (X) and
in X, then
(7.7.1) Wi+klJ u(x)(lmf(x))ieiwf(X)dxl

L sup ID a ul(lf'1 2 +Imf)l aI!2-k, w>O.

Here C is bounded when f stays in a bounded set in Ck +l(X). Jilihen f


is real valued, the estimate (7.7.1) reduces to

(7.7.1), L supIDa ullf'l'a l-2k, w>O.

Proof. When k=O the assertion is obvious since tir! is bounded for
t>O. In the proof we may therefore assume that k>O and that (7.7.1)
is already proved for smaller values of k as well as for smaller values
of j and the same k if j > O. It may also be assumed that
N=1f'12+Imf
is positive in K, for if (7.7.1) is proved then we can always replace f
by f+ei in (7.7.1) and let +0.

To be able to raise j or integrate by parts we observe that


n
Nu=u L loflox vI2+ulmf,
1
hence
n
(7.7.2) U= Luvofloxv+u o Imf
1

if NUv=uollox v when v =1=0 and Nuo=u. Since


iwofloxveiwf =ov eiwf
7.7. The Method of Stationary Phase 217

we obtain after an integration by parts


j u(Imf)ieiwf dx = j uo(Imf)i+ 1 ffwf dx
+i/w L(j(o.uv)(Imf)ieiWf dx+ jju.(ovlmf)(ImfY-leiwf dx) .
•
With the notation
lul",= L 1000ui
1"1=,,,
we have by the inductive hypothesis
(7.7.3) Wi+klj u(Imf)iffWf dxl

C sup (Iuol", + .tl1uvl",+ 1) N"'/2-k+ 1


+ ",ta lu.o v Im fl",N",/2-k).

To estimate the right-hand side we need a lemma.

Lemma 7.7.2. If gEC 2 ( -(j)j) is non-negative, then


(j2Ig'(OW + 2 sup (j2Ig"(X)I).
Ixld

Proof The proof is reduced to the case (j = 1 if x/(j is introduced as a


new variable. We may also assume g(O) = 1. By Taylor's formula
Ixl<l; M=suplg"l.
If we take x 2 =1 and obtain since
and if M>2 we take x 2 =2/M and obtain
=(2M)t. This proves the lemma.

End of proof of Theorem 7.7.1. Lemma 7.7.2 applied to Imf gives


(7.7.4) in K.
We shall now prove that when we have

(7.7.5) Nt L IUvl" + N IUol"


n
C L lulrN(r-
'"
",)/2 •
v= 1 r= 0

This follows from the definition of Nand u. if JJ =0 so we may


assume that 0 < JJ k and that the estimate is proved for smaller
values of JJ. Let letl = JJ and apply 0" to the equations
Nu.=uoJjox v when V=FO, Nuo=u.
218 VII. The Fourier Transformation

Estimating all terms except N 8a uv in the left-hand side by means of


(7.7.5) we obtain when v=l=O in view of (7.7.4)

N IUvl1' c(INI1Iuvll'_ 1+ IUvll'_ 2 + ... + IUvi o + lull' If I1+ lull'_ 1+ ... + lul o)
C(Ntluvll'_l + IUvll'_ 2 + ... + lull' Nt + lull'_l + ... )
L lulrN(r-l'+ 1)/2
I'
C"
o

which gives (7.7.5) then. Similarly

I'
CL lul r N(r-I')/2
o
which completes the proof of (7.7.5). From (7.7.5) and (7.7.4) we obtain
I'
(7.7.6) luv8vImfll' CL lul r N(r-I')/2, v =1=0, k.
o

If the estimates (7.7.5) and (7.7.6) are used in the right-hand side of
(7.7.3) we obtain (7.7.1).

Theorem 7.7.1 shows that the integral


Sueirof dx, W > 0,
is a rapidly decreasing function of w if UECg', fECoo, and
there is no point in supp u with Imf = 0 and f' = o. Thus the essential
contributions must always come from points where the phase f is real
and stationary. This is the basis for the stationary phase method which
in addition describes the contributions from the simplest types of such
critical points. The special case where f is a quadratic form was
discussed in Section 7.6 but we restate the result in a form which is
more convenient here.

Lemma 7.7.3. Let A be a symmetric non-degenerate matrix with


1m A O. Then we have for every integer k > 0 and integer s> n/2
(7.7.7) IS u(x) eiro (Ax,x)/2dx -(det(wA/2ni))-t 7;.(w)1
L IIDaullu, UE[/,

k-1
(7.7.8) 7;.(w)= L (2iw)-i<A- 1 D,D)i u(0)/j!.
o

You might also like