Parametric transformation of the input space variables. The transformation is obtained coordinatewise by integrating piecewise affine marginal "densities" parametrized by a vector of knots and a matrix of density values at the knots. See references for more detail.
affineScalingFun(X, knots, eta)
an n*d matrix standing for a design of n experiments in d-dimensional space
a (K+1) vector of knots parametrizing the transformation. The knots are here the same in all dimensions.
a d*(K+1) matrix of coefficients parametrizing the d marginal transformations. Each line stands for a set of (K+1) marginal density values at the knots defined above.
The image of X by a scaling transformation of parameters knots and eta
Y. Xiong, W. Chen, D. Apley, and X. Ding (2007), Int. J. Numer. Meth. Engng, A non-stationary covariance-based Kriging method for metamodelling in engineering design.