Deriving transformation matrix from clues

  • Thread starter GooseMunch
  • Start date
  • #1

Main Question or Discussion Point

I have PTV=R where P and T are square matrices (4x4) and V and R are non-square (4x3).
P and V are known, T is unknown, and R is partially known (3 unknown elements).
Seems impossible, but T is a transformation matrix (ie upperleft 3x3 is a rotation matrix) which gives me additional clues.

I'm trying to find T, but I can't figure out how to approach this problem. It's been almost 10 years since I took linear algebra course and I feel very lost.
Seems like I can just rearrange it like so:
T = P'RV' where P' and V' are inverses of P and V respectively but V is non-square so I can only get left-inverse. I don't think pseudoinverse of V is the answer because I'm afraid I'll end up with overdetermined system and the whole thing becomes unsolvable.

Can anyone nudge me in the right direction? Thanks.
 

Answers and Replies

  • #2
12,892
9,533
I would write out the entire system of linear equations, and see what can be said about kernel and rank of its coefficient matrix.
 

Related Threads on Deriving transformation matrix from clues

  • Last Post
Replies
14
Views
2K
Replies
3
Views
2K
  • Last Post
Replies
14
Views
2K
Top