Exercise 10.3

Exercise 3: (a) If F ˇ is as in Theorem 10.7, put Ǎ = F ˇ ( 0 ˇ ) , F ˇ 1 ( x ˇ ) = Ǎ 1 F ˇ ( x ˇ ) . Then F ˇ 1 ( 0 ˇ ) = I . Show that

F ˇ 1 ( x ˇ ) = Ǧ n Ǧ 1 ( x ˇ )

in some neighborhood of 0 ˇ , for certain primitive mappings Ǧ 1 , , Ǧ n . This gives another version of Theorem 10.7:

F ˇ ( x ˇ ) = F ˇ ( 0 ˇ ) Ǧ n Ǧ 1 ( x ˇ ) .

(b) Prove that the mapping ( x , y ) ( y , x ) of 2 onto 2 is not the composition of any two primitive mappings, in any neighborhood of the origin. This shows that the flips B i cannot be omitted from the statement of Theorem 10.7.

Answers

(a) (Much of this solution just repeats the proof of Theorem 10.7.) Assume 1 m n 1 , and make the following induction hypothesis (which evidently holds for m = 1 ):

V m is a neighborhood of 0 ˇ , F ˇ m C ( V m ) , F ˇ m ( 0 ˇ ) = 0 ˇ , F ˇ m ( 0 ˇ ) = I , and for x ˇ V m ,

( ) P m 1 F ˇ m ( x ˇ ) = P m 1 x ˇ

By (∗) , we have

F ˇ m ( x ˇ ) = P m 1 x ˇ + i = m n α i ( x ˇ ) ě i ,

where α m , , α n are real C -functions in V m . Hence, for j = m , , n ,

ě j = F ˇ m ( 0 ˇ ) ě j = i = m n ( D j α i ) ( 0 ˇ ) ě i .

Since ě m , ě n are independent, we must have

( ) ( D m α m ) ( 0 ˇ ) = 1 ( D m + 1 α m ) ( 0 ˇ ) = = ( D n α m ) ( 0 ˇ ) = 0 .

Define, for x ˇ V m ,

Ǧ m ( x ˇ ) = x ˇ + ( α m ( x ˇ ) x m ) ě m

Then Ǧ m C ( V m ) , Ǧ m is primitive, and Ǧ m ( 0 ˇ ) = I by (∗∗) . The inverse function theorem shows therefore that there is an open set U m , with 0 ˇ U m V m , such that Ǧ m is a 1-1 mapping of U m onto a neighborhood V m + 1 of 0 ˇ , in which Ǧ m 1 is continuously differentiable, and

Ǧ m 1 ( 0 ˇ ) = Ǧ m ( 0 ˇ ) 1 = I .

Define F ˇ m + 1 ( y ˇ ) , for y ˇ V m + 1 , by

F ˇ m + 1 ( y ˇ ) = F ˇ m Ǧ m 1 ( y ˇ ) .

Then F ˇ m + 1 C ( V m + 1 ) , F ˇ m + 1 ( 0 ˇ ) = 0 ˇ , and F ˇ m + 1 ( 0 ˇ ) = I by the chain rule. Also, for x ˇ U m ,

P m F ˇ m + 1 ( Ǧ m ( x ˇ ) ) = P m F ˇ m ( x ˇ ) = P m ( P m 1 x ˇ + α m ( x ˇ ) ě m + ) = P m 1 x ˇ + α m ( x ˇ ) ě m = P m Ǧ m ( x ˇ )

so that, for y ˇ V m + 1 , P m F ˇ m + 1 ( y ˇ ) = P m y ˇ . Our induction hypothesis holds therefore with m + 1 in place of m .

Note that, for y ˇ = Ǧ m ( x ˇ ) , we have

F ˇ m + 1 ( Ǧ m ( x ˇ ) ) = F ˇ m ( x ˇ ) .

If we apply this with m = 1 , , n 1 , we successively obtain

F ˇ 1 = F ˇ 2 Ǧ 1 = F ˇ 3 Ǧ 2 Ǧ 1 = = F ˇ n Ǧ n 1 Ǧ 1

in some neighborhood of 0 ˇ . By (∗) , F ˇ n is primitive, so we can let Ǧ n = F ˇ n .

(b) Let F ˇ be the mapping ( x , y ) ( y , x ) and suppose F ˇ = Ǧ 2 Ǧ 1 in some neighborhood of the origin, where

Ǧ 1 ( x , y ) = ( f ( x , y ) , y ) Ǧ 2 ( u , v ) = ( u , g ( u , v ) )

are primitive mappings. Then we would have

( y , x ) = Ǧ 2 Ǧ 1 ( x , y ) = Ǧ 2 ( f ( x , y ) , y ) = ( f ( x , y ) , g ( f ( x , y ) , y ) )

so that

y = f ( x , y ) x = g ( f ( x , y ) , y ) = g ( y , y )

which is impossible. Trying F ˇ = Ǧ 1 Ǧ 2 leads to a similar contradiction.

User profile picture
2023-08-07 00:00
Comments