How can I prove the property of ranks for an n x m matrix with n < m?

  • Context: Graduate 
  • Thread starter Thread starter dabd
  • Start date Start date
  • Tags Tags
    Property
Click For Summary

Discussion Overview

The discussion revolves around proving a property of ranks for an n x m matrix A where n < m, specifically that there exists some λ such that rank(A^T A + λ I) = m. Participants explore various aspects of matrix properties, eigenvalues, and conditions for invertibility.

Discussion Character

  • Exploratory
  • Technical explanation
  • Mathematical reasoning
  • Debate/contested

Main Points Raised

  • One participant suggests that if A is an n x m matrix with n < m, there exists some λ such that rank(A^T A + λ I) = m, but they struggle to prove it for more than two columns being independent.
  • Another participant questions how many eigenvalues transpose(A)*A has, noting that A is any n x m matrix.
  • A later reply indicates that the rank will be full if the determinant is non-zero, stating that the determinant of A^T A + λ I is a polynomial in λ and will be non-zero for all but finitely many λ.
  • One participant provides hints, stating that A^T A + λ I is full rank if it is invertible and questions the diagonal elements of A^T A + λ I.
  • Another participant mentions that A^T A is symmetric and positive semi-definite, and by taking any λ > 0, the matrix A^T A + λ I becomes positive definite, hence non-singular and invertible.
  • There is a query about whether -B is positive semi-definite if B is symmetric, while also noting that it is possible to choose λ such that the matrix is positive definite.
  • One participant clarifies that they meant A^T A is symmetric in their previous post.

Areas of Agreement / Disagreement

Participants express differing views on the conditions under which the rank can be proven to be full, and there is no consensus on the implications of eigenvalues or the conditions for positive definiteness.

Contextual Notes

Some assumptions regarding the properties of matrices and eigenvalues are not fully explored, and the discussion includes unresolved mathematical steps related to the rank and invertibility of the matrix.

dabd
Messages
25
Reaction score
0
if A is an n x m matrix where n < m I would like to prove that there exists some [tex]\lambda[/tex] such that [tex]rank(A^T A + \lambda I) = m[/tex]

I know that if two of the columns of [tex]A^T A[/tex] are linearly dependent, they are scalar multiples of each other and by adding some [tex]\lambda[/tex] to two different positions, those colums will become independent but I can't prove it for more than two columns.

Any tips?
 
Last edited:
Physics news on Phys.org
How many eigen-values does transpose(A)*A has?
 
chingkui said:
How many eigen-values does transpose(A)*A has?

That is not given. A is any n x m matrix.
 
It seems to me that you're taking the pessimistic viewpoint. The rank will be full if the determinant is non-zero. The determinant of [tex]A^T A + \lambda I[/tex] is a polynomial in [tex]\lambda[/tex] , so will have non-zero determinant for all but finitely many [tex]\lambda[/tex].
 
A few hints:

(1) [itex]A^T A + \lambda I[/itex] is full rank iff it's invertible.

(2) Any strictly Diagonally dominant matrix is invertible.

What are the diagonal elements of [itex]A^T A + \lambda I[/itex]?
 
I got the answer in simpler terms [tex]A^T A[/tex] is symmetric, so it is positive semi-definite and by taking any [tex]\lambda > 0[/tex] the matrix [tex]A^T A + \lambda I[/tex] is positive definite, hence non-singular and invertible.
 
Last edited:
If B is symmetric, then -B is positive-semidefinite? Notwithstanding this, you can choose [itex]\lambda[/itex] such that your matrix is indeed positive definite.
 
JSuarez said:
If B is symmetric, then -B is positive-semidefinite? Notwithstanding this, you can choose [itex]\lambda[/itex] such that your matrix is indeed positive definite.

I meant [tex]A^T A[/tex] is symmetric.
 

Similar threads

  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 20 ·
Replies
20
Views
4K
  • · Replies 5 ·
Replies
5
Views
2K
Replies
2
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 3 ·
Replies
3
Views
4K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 17 ·
Replies
17
Views
7K