Hello,

I have been told that I need to minimise the error between two matrices, like so.


{A_1}x = {A_2}s

Where A_1 and A_2 are convolution folding matrices. x represents the desired filter and s the actual filter. I need to minimise the error between x and s. So using Cholesky, I would do this


x = {({A^T}_1{A_1})^{ - 1}}{A_2}s


but the answer in my book is:


x = {({A^T}_1{A_1})^{ - 1}}{A_1}{A_2}s



Can someone explain how that answer was derived?

Thanks.