Space of Solutions for least square

43 Views Asked by At

Hi I am stuck at a problem , I know the answer but I need a proper mathematical way of writing it - How do I write the solution space of

$\underset{x}{\mathrm{arg min}}$$ \frac{1}{2}\|y-A_1\times A_2\times ....\times A_k \times x\|^2$

$A_1,A_2,....A_k$ are of any size and any rank.

I know it depends on the matrix with the least ranks but I need it proper mathematical defination , like we define for a vector space(with btw it will be in this case)

$V= \{x : x$ satisfy something $\}$

I am thinking answer should be

$W = \{ x: x = \{(A_1\times A_2\times ....\times A_k)^+\times y + v\}\}$ where $v\in V$ and $V$ is a vector space span by null space of $(A_1\times A_2\times ....\times A_k)$ and $(A_1\times A_2\times ....\times A_k)^+$ is pseudo inverse of $(A_1\times A_2\times ....\times A_k)$.

But I am not sure if this is the right way to write it , should I include $r$, where $r$ is the least rank of $A_i$ in $A_1,A_2,....A_k$ which is basically rank of $(A_1\times A_2\times ....\times A_k)$