This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: MATH 220: THE FOURIER TRANSFORM BASIC PROPERTIES AND THE INVERSION FORMULA The Fourier transform is the basic and most powerful tool in studying constant coefficient PDE on R n . It is based on the following simple observation: for R n , the functions v ( x ) = e ix = e ix 1 1 e ix n n are joint eigenfunctions of the operators x j , namely for each j , (1) x j v = i j v . It would thus be desirable to decompose an arbitrary function u as an (infinite) linear combination of the v , namely write it as (2) u ( x ) = (2 ) n integraldisplay R n u ( ) e ix d, where u ( ) is the amplitude of the harmonic e ix in u . (The factor (2 ) n is here due to a convention, it could also be moved to other places.) It turns out that this identity, (2), holds provided that we define u ( ) = integraldisplay R n e ix u ( x ) dx ; (2) is then the Fourier inversion formula . Rather than showing this at once, we start with a stepbystep approach. We first define the Fourier transform as ( F u )( ) = integraldisplay R n e ix u ( x ) dx for u C ( R n ) with  x  N  u  bounded for some N > n (i.e.  u ( x )  M  x  N for some M in say  x  > 1); one could instead assume simply that u L 1 ( R n ) if one is familiar with the Lebesgue integral (but actually just working with the special case mentioned above gives everything we want). Note that for such functions  ( F u )( )  integraldisplay R n  e ix  u ( x )  dx = integraldisplay R n  u ( x )  dx, so F u is actually a bounded continuous function. We can similarly define the inverse Fourier transform ( F 1 )( x ) = (2 ) n integraldisplay R n e ix ( ) d ; then F 1 maps u C ( R n ) with  x  N  u  bounded for some N > n (or indeed u L 1 ( R n )) to bounded continuous functions. With these definition it is of course not clear whether F 1 is indeed the inverse of F , and worse, not even clear whether F 1 F makes sense for L 1 ( R n ), since F is then a bounded continuous function, which is not sufficient to ensure that the integral defining F 1 actually converges! We thus proceed to study properties of F and F 1 . First, we note a property of F which is the main reason for its usefulness in studying PDE, and which is an immediate consequence of (1). Namely, suppose that C 1 ( R n ) and both and 1 2 all first derivatives j , j = 1 ,...,n , decay at infinity in the same sense as above (so  x  N j is bounded for some N > n ). Then integration by parts gives ( F ( x j ))( ) = integraldisplay R n e ix x j ( x ) dx = integraldisplay R n x j ( e ix ) ( x ) dx = i j integraldisplay R n e ix ( x ) dx = i j ( F )( ) ....
View
Full
Document
This note was uploaded on 03/16/2010 for the course CME 303 taught by Professor Vasy during the Fall '10 term at Stanford.
 Fall '10
 Vasy

Click to edit the document details