Modified toom-cook algorithm

1. For these transformation matrices generated using modified toom-cook algorithm , why are they different compared to the following transformation matrices in the picture below ?
2. In section 5 on page 7, why does author claim that his modified toom-cook algorithm only requires n-1 points ? How exactly deg(M′(x)) = deg(M(x))−1 is relevant in proving modified toom-cook algorithm ? There is not much explanation (Algorithm 8.2.2) on page 249/234 inside reference [10] : VLSI Digital Signal Processing Systems: Design and Implementation

We can solve the smaller problem using M′(x) where deg(M′(x)) = deg(M(x))−1 and just add the missing value. With this approach we need n−1 points instead of n used in Section 3.1 for generating matrices suitable for the same size of computed convolution

Click to expand...

Click to expand...

promach said:

1. For these transformation matrices generated using modified toom-cook algorithm , why are they different compared to the following transformation matrices in the picture below ?

Click to expand...

promach said:

In section 5 on page 7,

Click to expand...