Computational relaxations for affine tensor product model transformation

The paper introduces methods that decrease the computational and memory burden of discretisation based affine tensor product model transformation. Their importance comes from the fact, that the computational cost of multi-variate functions' orthonormalization and the amount of accessible memory bound the applicable discretisation density and the number of parameters. The proposed methods can overwhelm these limitations and allow to apply the methodology on LPV/qLPV models of complex, practically relevant systems.