171k views
3 votes
(a) Consider an n×m matrix A such that Aᵀ A= Iₙ .Is it necessarily true that AAᵀ =Iₙ ? Explain or give a counter-example.

(b) Consider an n×n matrix A such that Aᵀ A=Iₙ . Is it necessarily true that AAᵀ = Iₙ ? Explain or give a counter-example.
(c) Consider an n×n matrix A which has orthonormal columns. Compute Aᵀ A , what is it equal to?
Is it the same as AAᵀ ? (Hint: refer to part b). Can we say that Aᵀ = A−¹ ?

User Jcreignou
by
7.6k points

1 Answer

2 votes

Final answer:

In (a), it is not necessarily true that AAᵀ=Iₙ. In (b), it is necessarily true that AAᵀ=Iₙ. In (c), AᵀA equals Iₙ when the columns of A are orthonormal.

Step-by-step explanation:

(a) Consider an n×m matrix A such that Aᵀ A= Iₙ. Is it necessarily true that AAᵀ =Iₙ? Explain or give a counter-example.

No, it is not necessarily true that AAᵀ = Iₙ. A counter-example can be shown by considering a 2x3 matrix A such that A = [1 0 0; 0 1 0]. In this case, Aᵀ A = I₂ but AAᵀ ≠ I₂, because the dimensions do not match.

(b) Consider an n×n matrix A such that Aᵀ A=Iₙ. Is it necessarily true that AAᵀ = Iₙ? Explain or give a counter-example.

Yes, it is necessarily true that AAᵀ = Iₙ. This can be proven by considering the singular value decomposition (SVD) of matrix A. From the SVD, it can be shown that AAᵀ = UΣ²Uᵀ = Iₙ, where U is a unitary matrix and Σ is a diagonal matrix containing the singular values of A.

(c) Consider an n×n matrix A which has orthonormal columns. Compute Aᵀ A, what is it equal to? Is it the same as AAᵀ?

Aᵀ A is equal to the identity matrix Iₙ when the columns of A are orthonormal. This is because the transpose of a matrix with orthonormal columns is the inverse of the matrix. However, AAᵀ is not necessarily equal to Iₙ.

User Mark Adams
by
8.0k points