Help proving equality of matrix expressions

  • Context: Graduate 
  • Thread starter Thread starter Chuck37
  • Start date Start date
  • Tags Tags
    Expressions Matrix
Click For Summary

Discussion Overview

The discussion revolves around proving the equality of two matrix expressions involving arbitrary-sized matrix J and square matrices P and R. Participants explore various approaches to demonstrate this equality, focusing on algebraic manipulation and properties of matrix inverses.

Discussion Character

  • Debate/contested
  • Mathematical reasoning
  • Technical explanation

Main Points Raised

  • One participant notes that they see numerical equality between the two expressions but struggles to prove it algebraically.
  • Another participant suggests setting the two expressions equal and performing operations on both sides to achieve equality, although they express a preference for a more direct transformation method.
  • A different participant provides a factorization approach, indicating that both sides can be manipulated to show equivalence through specific algebraic steps.
  • One participant emphasizes the importance of using the identity for the inverse of a product of matrices and suggests applying it repeatedly to simplify the left-hand side.
  • Concerns are raised about the necessity of J being invertible for certain methods to work, indicating that the approach may vary depending on the properties of J.

Areas of Agreement / Disagreement

Participants do not reach a consensus on a single method to prove the equality, with multiple approaches and interpretations being discussed. Some participants agree on the utility of algebraic manipulation, while others highlight the limitations based on the properties of J.

Contextual Notes

Limitations include the assumption that J is invertible for certain methods to apply, and the discussion does not resolve the mathematical steps necessary to complete the proof.

Chuck37
Messages
51
Reaction score
0
I've been beating my head against this problem for hours. I see numerically that two expressions are equal, but I can't prove it:

(J^{T}R^{-1}J + P^{-1})^{-1}J^{T}R^{-1}

=

PJ^{T}(JPJ^{T} + R)^{-1}

J is arbitrary size, P and R are square though not necessarily equal. Can anyone help? I thought binomial inverse theorem would save me but I haven't been able to get rid of the extraneous terms.
 
Physics news on Phys.org
Well, right after I posted I tried a new tack of setting them equal and doing operations to both sides until it came out the same. Not quite as satisfying as transforming one into the other, but I guess it works. If someone can see how to make one into the other, I'd love to see it.
 
This is equivalent to

J^{T}R^{-1} JPJ^{T} + J^{T}
=
J^{T}R^{-1} JPJ^{T} + J^{T}

Factorize it two different ways:

J^{T}R^{-1} (JPJ^{T} + R)
=
(J^{T}R^{-1}J + P^{-1}) PJ^{T}

And the final step to get to your equation should be obvious.

(Of course this was invented by working backwards.)
 
Last edited:
Chuck37 said:
Well, right after I posted I tried a new tack of setting them equal and doing operations to both sides until it came out the same. Not quite as satisfying as transforming one into the other, but I guess it works. If someone can see how to make one into the other, I'd love to see it.
As you just said, you don't actually have to "transform one into the other" to prove this result. However, it's not hard to just rewrite your left-hand side until you end up with the right-hand side. You just have to use the identity ##(AB)^{-1}=B^{-1}A^{-1}## repeatedly (and the fact that matrix multiplication is distributive over matrix addition). I recommend that you prove this identity first. It's not hard.

I can't tell you the complete solution because of the forum rules about textbook-style problems, so for now I will only suggest that you start by applying this identity to the factor ##J^TR^{-1}##. To get more help, you will have to show your work up to the point where you get stuck.

By the way, you shouldn't put itex tags around each symbol. Instead, put an opening tag (itex or tex) before each formula and a closing tag after it. See the LaTeX guide for more information.Edit: The method I suggested only works when J is invertible (and therefore square). When J isn't square, I think you have to do something like what AlephZero suggested.
 
Last edited:

Similar threads

Replies
3
Views
2K
  • · Replies 14 ·
Replies
14
Views
4K
  • · Replies 7 ·
Replies
7
Views
3K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 1 ·
Replies
1
Views
4K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 17 ·
Replies
17
Views
7K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 2 ·
Replies
2
Views
2K