where the subscripts m and n denote the number of
process states and inputs, respectively.
Now, the highest possible amplification for each
state can be found as the H1-norms of the related
row vectors in the transfer function matrix. The scaling
vector is then defined as