Adjoint of transformation proof

Click For Summary

Homework Help Overview

The discussion revolves around proving properties related to the adjoint of a linear operator T in an inner product space V. Specifically, participants are tasked with demonstrating relationships between the range and nullspace of T and its adjoint T*.

Discussion Character

  • Exploratory, Assumption checking, Mathematical reasoning

Approaches and Questions Raised

  • Participants explore the implications of the inner product relationships and the definitions of range and nullspace. There are attempts to establish subset relationships between N(T) and R(T*)⊥, as well as discussions on the significance of finite dimensionality in part b).

Discussion Status

Several participants are actively engaging with the problem, offering various approaches and questioning assumptions. There is a recognition of the need to clarify definitions and the implications of certain steps, particularly regarding the relationship between vectors in the nullspace and their orthogonality to the range of T*.

Contextual Notes

Participants note that the problem may impose constraints on the use of matrix representations and that the definitions of range (R) and nullspace (N) are critical to the discussion. The mention of finite dimensionality in part b) raises questions about its relevance to the proof.

zcd
Messages
197
Reaction score
0

Homework Statement


Let V be an inner product space and let T be a linear operator on V. Prove the following results:
a)R(T*)=N(T)
b)R(T*)=N(T) if V is finite dimensional

Homework Equations


<Tx,y>=<x,T*y>

The Attempt at a Solution


pick x and y ≠0 and <Tx,y>=<x,T*y>=0
this implies x∈N(T) and x ⊥ R(T*) , so N(T)⊇R(T*)

This half of the subset relationship I got, but how do I prove the other way? And as for part b), do I just prove that (W)=W? If so, why did they mention that V is finite dimensional?
 
Physics news on Phys.org
zcd said:

The Attempt at a Solution


pick x and y ≠0 and <Tx,y>=<x,T*y>=0
this implies x∈N(T) and x ⊥ R(T*) , so N(T)⊇R(T*)
i don't think this implies x∈N(T) and x ⊥ R(T*)...

but that (x∈N(T) OR Tx ⊥ y) AND x ⊥ R(T*)
 
how about starting by either assuming you have vector in the nullspace, and then for the other direction a vector perpindicular to the row vectors of T*
 
With your suggestion, I tried:

pick x and y ≠0 (the 0 case is readily proven) and x∈N(T)
then <Tx,y>=<x,T*y>=0, which implies x ⊥ R(T*) i.e x∈R(T*)
N(T)⊆R(T*)

for the other way, pick x∈R(T*) (x and y ≠0 again)
<Tx,y>=<x,T*y>=0 only when Tx=0, so x∈N(T)
N(T)⊇R(T*)

is that a sufficient proof?
 
first part pretty much, however I would write it as follows (some of it probably just diffenrent style)
-->
first pick x∈N(T)
then Tx=0
and <Tx,y>=0 for all y
but <Tx,y>=<x,T*y>=0 for all y
T*y represents a linear combination of the columns of T*
x ⊥column space of T*
x ⊥ R(T*)
so x∈R(T*)
-----

actually now that I'm working thorugh it, what exactly do you mean by R & N?
 
R is the range of transformation T and N is the nullspace of transformation T. I don't think we're supposed to use columns of T as that implies we pick a basis to represent T as a matrix.
 
ok well in the last post as y is arbitrary, you have shown x is perp to any element in the image of T (which depending on terminology can be the same as the range)
 

Similar threads

Replies
4
Views
1K
  • · Replies 24 ·
Replies
24
Views
4K
Replies
2
Views
1K
Replies
8
Views
2K
  • · Replies 11 ·
Replies
11
Views
2K
  • · Replies 7 ·
Replies
7
Views
1K
Replies
10
Views
3K
Replies
1
Views
2K
  • · Replies 4 ·
Replies
4
Views
1K
  • · Replies 1 ·
Replies
1
Views
3K