Frobenius matrix norm vs. 2-norm

1.4k Views Asked by At

From this article about the singular value decomposition:

Let $A$ be an $n \times d$ matrix and think of the rows of $A$ as $n$ points in $d$-dimensional space. The Frobenius norm of $A$ is the square root of the sum of the squared distance of the points to the origin. The 2-norm is the square root of the sum of squared distances to the origin along the direction that maximizes this quantity.

Can someone help me to understand this remark? Thanks

1

There are 1 best solutions below

6
On BEST ANSWER

The first says $$\Vert A \Vert_F^2 = \sum_{k=1}^n \Vert p_k - 0 \Vert_2^2 = \sum_{k=1}^n \sum_{i=1}^d (p_k)_i^2$$ which is true if $(p_k)_i = A_{ki}$.
For the latter a similar thing holds considering $\Vert A \Vert_2^2 = |\lambda_{\max}|$; Mathematically put and implying a "direction" vector to be normalized, it says $$\Vert A \Vert_2^2 = \sup_{\Vert x\Vert_2 = 1} \Vert Ax \Vert_2^2 = \sup_{x\in \mathbb R^d \setminus \{0\}} \frac{\Vert Ax \Vert_2^2}{\Vert x\Vert_2^2}$$ Which coincides with the definition of the $2$-Norm as the induced Operator norm and the last equality holds due to continuity and linearity of the Operator $x\mapsto Ax$.