Research Article
Tensorial Model for Photolithography Aerial Image Simulation
Algorithm 1
Lower rank-(
) tensor approximation (LRTA-(
))
(1) Input: data tensor , and dimensions of all -mode signal subspaces. | (2) Initialization : For to , calculate the projectors given by HOSVD-(): | (a) -mode unfold into matrix ; | (b) Compute the SVD of ; | (c) Compute matrix formed by the eigenvectors associated with the largest singular values of | . is the initial matrix of the -mode signal subspace orthogonal basis vectors; | (d) Form the initial orthogonal projector on the -mode signal subspace; | (e) Compute the HOSVD-() of tensor given by | ; | (3) ALS loop: Repeat until convergence, that is, for example, while , being a prior | fixed threshold, | (a) For to : | (i) Form : ; | (ii) -mode unfold tensor into matrix ; | (iii) Compute matrix ; | (iv) Compute matrix composed of the eigenvectors associated with the largest eigenvalues | of . is the matrix of the -mode signal subspace orthogonal basis vectors at the | iteration; | (v) Compute ; | (b) Compute ; | (c) Increment . | (4) Output: the estimated signal tensor is obtained through . is | the rank- approximation of , where is the index of the last iteration after the | convergence of TUCKALS3 algorithm. |
|