Can someone help me to understand the last two steps in Hadamard transpose times Hardamard in the given image? I don’t understand how/why the $8I$’s got multiplied by $Q_8$.
Gilbert Strang, page 242:
Then $Q=H/2$ has orthonormal columns. Dividing by $2$ gives unit vectors in $Q$. A $5$-by-$5$ Hadamard matrix is impossible because the dot product of columns would have five $1$’s and/or $-1$’s and could not add to zero. $H_8$ has orthogonal columns of length $\sqrt 8$. $$\def\T{\mathrm T} H^\T_8 H_8 = \left[ \begin{array}{lr} H^\T & H^\T \\ H^\T & -H^\T \end{array} \right] \left[ \begin{array}{lr} H & H \\ H & -H \end{array} \right] = \begin{bmatrix} 2H^\T H & 0 \\ 0 & 2H^\T H \end{bmatrix} = \begin{bmatrix} 8I & 0 \\ 0 & 8I \end{bmatrix} . Q_8 = \frac {H_8} {\sqrt 8}$$
is the next Hadamard matrix with 1's -1's. What is the product of $H^T_8H_8$? Here $H_4$ is defined in terms of $H_2$.
$H_4$=\begin{bmatrix} {H_2}& {H_2} \ {H_2}& {-H_2} \ \end{bmatrix} and $H_2$=\begin{bmatrix} 1& 1 \ 1& -1 \ \end{bmatrix}
– user82039 Feb 04 '24 at 04:31