Matrix Derivative Homework: Show Differentiability and Find Derivative

In summary, the conversation discusses finding the derivative of the function f(A) = A^-1, which is the inverse of a nxn matrix. The attempt at a solution involves using the relation f(A)A = I and differentiating both sides to find f'(A). It is then suggested to determine the existence of f'(A) using the definition of the derivative. The conversation also discusses the nature of the derivative of an inverse matrix and how it is not an n×n matrix. It is then suggested to express the inverse of (A+δ) as (A-1+ε) and derive an expression for ε. This leads to an understanding of the desired derivative and the need to validate certain assumptions.
  • #1
Gavins
27
0

Homework Statement


Let f: R^nxn -> R^nxn be the function f(A) = A^-1. ie f is the inverse function for some nxn matrix. Show it is differentiable and find the derivative.


Homework Equations





The Attempt at a Solution


I need to do something along the lines of (A + H)^-1 = A^-1 + f'(A)H + p(H) where p is a remainder term. I'm not sure how to expand this. I've looked in my notes and our lecturer did something along the lines of (A + H)^-1 - A^-1 = (A+H)^-1(I - (A+H)A^-1) = (A+H)^-1(-HA^-1). I can't really figure out what to do next since it doesn't seem easy to expand (A+H)^-1.
 
Physics news on Phys.org
  • #2
It might be easier to start with the relation f(A) A = I and differentiate both sides.
 
  • #3
f'(A)A + f(A) = 0?
f'(A) = -A^-2

That's assuming the derivative exists. Check if the remainder term goes to 0 to show it is differentiable. Let |.| be the norm.

(A + H)^-1 - A^-1 + A^-2H = p(H)
|p(H)|/|H| = |(A + H)^-1 - A^-1 + A^-2H|/|H|
=< |(A+H)^-1|/|H| - |A^-1|/|H| + |A^-2|

I don't think this is right since this definitely doesn't go to zero when H goes to zero. Sorry for the really bad format.
 
  • #4
You can write f'(A) in terms of f(A). Determine the existence of f'(A) in terms of the existence of f(A).
 
  • #5
I'm not sure I understand. Is this easy to do from first principles? I think that's what my lecturer wants. Specifically, using the definition of the derivative, f(x+h) = f(x) + f'(x)h + p(h).

Thanks.
 
  • #6
You might want to step back and ask "What is the nature of this beast?" regarding the derivative of the inverse of some matrix A with respect to A.

Hint: It is not an n×n matrix.Given some invertible matrix A and a small perturbation δ of this matrix, assume that (A+δ) is invertible. Without loss of generality you can express the inverse of (A+δ) in the form (A-1+ε). Derive an expression for this matrix ε.

Hint: Solve for (A-1+ε)(A+δ)=I, assuming that the term εδ is small.What does this tell you about the desired derivative? What assumptions did you make that need to be validated?
 

1. What is a matrix derivative?

A matrix derivative is a mathematical concept that describes how a matrix changes with respect to a certain variable. It is similar to a regular derivative, but instead of a single variable, it involves multiple variables and results in a matrix of derivatives.

2. How do you show differentiability of a matrix?

To show differentiability of a matrix, you need to use the definition of differentiability and check if the limit of the difference quotient exists for each entry in the matrix. If all the entries have a limit, then the matrix is differentiable.

3. What is the process for finding the derivative of a matrix?

The process for finding the derivative of a matrix involves taking the partial derivatives of each entry in the matrix with respect to the variable, and then arranging these partial derivatives into a new matrix. This new matrix is the derivative of the original matrix.

4. Can you provide an example of finding the derivative of a matrix?

Sure, let's say we have the matrix A = [[x^2, 3x], [2x, x+1]]. To find its derivative with respect to x, we take the partial derivatives of each entry: d/dx(x^2) = 2x, d/dx(3x) = 3, d/dx(2x) = 2, and d/dx(x+1) = 1. These partial derivatives form the derivative matrix: [[2x, 3], [2, 1]].

5. How are matrix derivatives used in real-world applications?

Matrix derivatives have many applications in fields such as physics, engineering, economics, and computer science. They are used to optimize functions, solve optimization problems, and model complex systems. For example, in machine learning, matrix derivatives are used to calculate the gradient of a cost function, which is then used to update the parameters of a model during training.

Similar threads

  • Calculus and Beyond Homework Help
Replies
8
Views
476
  • Calculus and Beyond Homework Help
Replies
3
Views
574
  • Calculus and Beyond Homework Help
Replies
4
Views
119
  • Calculus and Beyond Homework Help
Replies
6
Views
1K
  • Calculus and Beyond Homework Help
Replies
22
Views
2K
  • Calculus and Beyond Homework Help
Replies
11
Views
2K
  • Calculus and Beyond Homework Help
Replies
4
Views
1K
  • Calculus and Beyond Homework Help
Replies
2
Views
672
  • Calculus and Beyond Homework Help
Replies
2
Views
250
  • Calculus and Beyond Homework Help
Replies
0
Views
171
Back
Top