13.1. FOURIER TRANSFORMS OF FUNCTIONS IN G 377

The following lemma follows from the dominated convergence theorem and routinemanipulations.

Lemma 13.1.7 For ψ ∈ G ,

Dαt F (ψ) = (−i)|α|F (xα

ψ (x)) , Dαt F−1 (ψ) = (i)|α|F−1 (xα

ψ (x))

In this lemma, xα ψ (x) denotes the function x→ xα ψ (x).One reason for using the functions G is that it is very easy to compute the Fourier

transform of these functions. The first thing to do is to verify F and F−1 map G to G andthat F−1 ◦F (ψ) = ψ. Next is a simple lemma from calculus which is about the Fouriertransforms of e−c|t|2 .

Lemma 13.1.8 The following hold. (c > 0)(1

)n/2 ∫Rn

e−c|t|2e−is·tdt =(

12π

)n/2 ∫Rn

e−c|t|2eis·tdt

=

(1

)n/2

e−|s|24c

(√π√c

)n

=

(12c

)n/2

e−14c |s|

2. (13.1)

That is, F(

e−c|t|2)=( 1

2c

)n/2e−

14c |s|

2and F−1

(e−c|t|2

)=( 1

2c

)n/2e−

14c |s|

2.

Proof: Consider first the case of one dimension. Let H (s) be given by

H (s)≡∫R

e−ct2e−istdt =

∫R

e−ct2cos(st)dt

Then H (0)2 =∫R∫R e−c(t2+s2)dtds=

∫∞

0∫ 2π

0 e−cr2rdθdr = 1

c π so H (0)=√

π

c . Then using

the dominated convergence theorem to differentiate, H ′ (s) + s2c H (s) = 0,H (0) =

√π

c .Thus

dds

(es2/4cH (s)

)= 0 and so es2/4cH (s)−

√π

c= 0, so H (s) =

√π

ce−s2/4c

Hence1√2π

∫R

e−ct2e−istdt =

√π

c1√2π

e−s24c =

(12c

)1/2

e−s24c .

This proves the formula in the case of one dimension. The case of the inverse Fouriertransform is similar. The n dimensional formula follows from Fubini’s theorem. ■

With these formulas, it is easy to verify F,F−1 map G to G and F ◦F−1 = F−1 ◦F = id.

Theorem 13.1.9 Each of F and F−1 map G to G . Also F−1 ◦ F (ψ) = ψ andF ◦F−1 (ψ) = ψ .

Proof: It is obvious that F,F−1 map G to G using Lemmas 13.1.7, 13.1.8. Indeed, forψ (x) = xα e−c|x|2 ,

F (ψ) = (−i)|α|Dα F(

e−c|x|2)= (−i)|α|Dα

((12c

)n/2

e−14c |s|

2

)

= (−i)|α| (−1)|α|(

12c

)|α|+n/2

sα e−14c |s|

2

13.1. FOURIER TRANSFORMS OF FUNCTIONS IN Y 377The following lemma follows from the dominated convergence theorem and routinemanipulations.Lemma 13.1.7 For yeZ,D¢F (w) = (i) F (a y(x)), DPF! (y) = (i) Fo! (a y(a))In this lemma, x” y (x) denotes the function x > 2% y (a).One reason for using the functions Y is that it is very easy to compute the Fouriertransform of these functions. The first thing to do is to verify F and F~' map Y to Y andthat F~! o F (w) = y. Next is a simple lemma from calculus which is about the Fourier2transforms of e~¢l#l’.Lemma 13.1.8 The following hold. (c > 0)n/2_, n/2.,(=z) i oct! gist pp — (=) / on cltl gist ay2 Ni 2(a) (Gree om2a Jc 2c , ‘That is, F (ew) = (2)"? e-aels! and F~! (e“""’) = (2)? eal,Proof: Consider first the case of one dimension. Let H (s) be given byH()= fee mara [ee cos (st) dtThen H (0)” = fe fe el) dtds = So So” e- rdO0dr=!ns0H (0) = ,/. Then usingcthe dominated convergence theorem to differentiate, H’ (s) + aeThusda s*/4c _ s*/4c _ ax _ _ a —s*/4c<(e H(s)) =Oand soe H(s) ~ 9 so H (s) = ae1/21 —ct? —ist x 1 2 ( 1 ) _2— | e“%e“dt=,/—-— =e *®& =[|— e «,aa hh Vc 2a 2cThis proves the formula in the case of one dimension. The case of the inverse Fouriertransform is similar. The n dimensional formula follows from Fubini’s theorem. MfWith these formulas, it is easy to verify F,F~' map Y to Y and FoF~! =F —!oF =id.=So|=)_o—(a)SSllenHenceTheorem 13.1.9 Each of F and F~' map Y to Y. Also F~!oF (w) = w andFoF '(w)=y.Proof: It is obvious that F,F7! map Y to Y using Lemmas 13.1.7, 13.1.8. Indeed, fory(e) =a%e le",(—)|*| D“F (e“#") = (—i)!@! p® (3) *")|a|-+n/2_ (—1)!@' (—1)'' (5:) s%e kis”F(y)2c