Why is the remainder term zero for linear maps in D&K example 2.2.5?

3 replies · 2K views · last reply
Math Amateur
Gold Member
MHB
Messages
3,920
Reaction score
48
I am reading "Multidimensional Real Analysis I: Differentiation" by J. J. Duistermaat and J. A. C. Kolk ...

I am focused on Chapter 2: Differentiation ... ...

I need help with an aspect of Example 2.2.5 ... ...

Duistermaat and Kolk's Example 2.2.5 read as follows:
Printed textbook page showing Example 2.2.5: every linear map A is differentiable with DA(a) = A, and the derivative of addition on R^n x R^n is the addition map itself, with the inner product derivative Dg(a1,a2)(h1,h2) = <a1,h2> + <a2,h1>.

Printed page from Duistermaat and Kolk showing the bound 0 ≤ lim |(h1,h2)|/||(h1,h2)|| ≤ lim ||h1|| ||h2||/||(h1,h2)|| ≤ lim ||h2|| = 0, and the statement that Df(a)h = (Df1(a)h, Df2(a)h).
In the above text by D&K we read the following:

" ... ... Indeed ##A(a+h) - A(a) = A(h)##, for every ##h \in \mathbb{R}^n##; and there is no remainder term. ... ... "Now I can see that

##A(a + h) = A(a) + A(h)## ... ... (1) from the definition of ##A## ...

and in (2.10) we have ...

##A(a +h) - A(a) = DA(a)h + \epsilon_a(h)## ... ... (2)

So ... from (1) and (2) we get

##A(h) = DA(a)h + \epsilon_a(h)##

... BUT ... why, in D&K's terms is "there no remainder term" ...

... in other words ... why is ##\epsilon_a(h) = 0## ...
Hope someone can help ...

Peter
==========================================================================================***NOTE***

The above post refers to equation (2.10) which occurs in Definition 2.2.2 ... so I am providing Definition 2.2.2 and the accompanying text ... as follows:
Text page with Definition 2.2.2: a map f from open U in R^n to R^p is differentiable at a if a linear map Df(a) exists with the limit of ||f(x) - f(a) - Df(a)(x - a)|| / ||x - a|| equal to zero as x tends to a; equivalently f(a+h) = f(a) + Df(a)h + epsilon_a(h) with epsilon_a(h)/||h|| tending to zero as h tends to zero (2.10).

Printed text from a textbook: In the case where n = p = 1, the mapping L ↦ L(1) gives a linear isomorphism End(R) → R. This new definition of the derivative Df(a) ↔ Df(a)1 ∈ R therefore coincides (up to isomorphism) with the original f'(a) ∈ R.
I hope that helps readers understand the context and notation of the above post ...

Peter
 
Attachments
  • Printed textbook page showing Example 2.2.5: every linear map A is differentiable with DA(a) = A, and the derivative of addition on R^n x R^n is the addition map itself, with the inner product derivative Dg(a1,a2)(h1,h2) = <a1,h2> + <a2,h1>.
    D&K - 1 - Example 2.2.5 ... ... PART 1 ... ... png.png
    37.9 KB · Views: 778
  • Printed page from Duistermaat and Kolk showing the bound 0 ≤ lim |(h1,h2)|/||(h1,h2)|| ≤ lim ||h1|| ||h2||/||(h1,h2)|| ≤ lim ||h2|| = 0, and the statement that Df(a)h = (Df1(a)h, Df2(a)h).
    D&K - 2 - Example 2.2.5 ... ... PART 2 ... .png
    32.2 KB · Views: 589
  • Text page with Definition 2.2.2: a map f from open U in R^n to R^p is differentiable at a if a linear map Df(a) exists with the limit of ||f(x) - f(a) - Df(a)(x - a)|| / ||x - a|| equal to zero as x tends to a; equivalently f(a+h) = f(a) + Df(a)h + epsilon_a(h) with epsilon_a(h)/||h|| tending to zero as h tends to zero (2.10).
    D&K - 1 - Defn 2.2.2 ... ... PART 1 ... .png
    57.4 KB · Views: 669
  • Printed text from a textbook: In the case where n = p = 1, the mapping L ↦ L(1) gives a linear isomorphism End(R) → R. This new definition of the derivative Df(a) ↔ Df(a)1 ∈ R therefore coincides (up to isomorphism) with the original f'(a) ∈ R.
    D&K - 2 - Defn 2.2.2 ... ... PART 2 ... .png
    11.6 KB · Views: 503
Last edited:
If we define ##DA(a)(h) := A(h)## then we get a linear mapping which satisfies the conditions of a derivative and for which the remainder is zero by definition. Now Lemma 2.2.3 says, that there is only one and thus ##DA(a)=A## is the derivative.
 
  • Like
Likes   Reactions: Math Amateur
In the quoted text we have that ##DA(a)=A\forall a\in \mathbb R^n##.
hence equation (2) becomes
$$A(a+h)=
A(a)+
DA(a)(h)+\epsilon_a(h)
=A(a)+A(h)+\epsilon_a(h)$$
Subtracting (1) from this gives the desired result.
 
  • Like
Likes   Reactions: Math Amateur
Thanks fresh_42, Andrew ...

... appreciate your help ...

Peter