# Inversion problem for big dataset

Suppose my goal is to solve for $$x$$ in $$Ax = b$$ using CG method and $$A$$ is Toeplitz. The problem is that $$b$$ is extremely large in size and I can't just read in the whole vector $$b$$ and then solve for $$x$$.

I know that if $$A$$ is comes from a convolution process, method such as OLS or OLA can work on big dataset by doing it piecewise.

However if $$A$$ isn't from a convolution process but is Toeplitz, is there a general approach to this kind of problem? Some way to break $$b$$ up and solve for individual pieces and put it back together in a meaningful way ?

• Is $A$ Toeplitz, but not circulant? – Marcus Müller Jul 23 '19 at 11:22
• Unfortunately, not a circulant. – Ong Beng Seong Jul 24 '19 at 3:04