English

Tensor numerical method for optimal control problems constrained by an elliptic operator with general rank-structured coefficients

Numerical Analysis 2021-05-28 v1 Numerical Analysis

Abstract

We introduce tensor numerical techniques for solving optimal control problems constrained by elliptic operators in Rd\mathbb{R}^d, d=2,3d=2,3, with variable coefficients, which can be represented in a low rank separable form. We construct a preconditioned iterative method with an adaptive rank truncation for solving the equation for the control function, governed by a sum of the elliptic operator and its inverse M=A+A1M=A + A^{-1}, both discretized over large ndn^{\otimes d}, d=2,3d=2,3, spatial grids. Two basic solution schemes are proposed and analyzed. In the first approach, one solves iteratively the initial linear system of equations with the matrix MM such that the matrix vector multiplication with the elliptic operator inverse, y=A1u,y=A^{-1} u, is performed as an embedded iteration by using a rank-structured solver for the equation of the form Ay=uA y=u. The second numerical scheme avoids the embedded iteration by reducing the initial equation to an equivalent one with the polynomial system matrix of the form A2+IA^2 +I. For both schemes, a low Kronecker rank spectrally equivalent preconditioner is constructed by using the corresponding matrix valued function of the anisotropic Laplacian diagonalized in the Fourier basis. Numerical tests for control problems in 2D setting confirm the linear-quadratic complexity scaling of the proposed method in the univariate grid size nn. Further, we numerically demonstrate that for our low rank solution method, a cascadic multigrid approach reduces the number of PCG iterations considerably, however the total CPU time remains merely the same as for the unigrid iteration.

Keywords

Cite

@article{arxiv.2105.13206,
  title  = {Tensor numerical method for optimal control problems constrained by an elliptic operator with general rank-structured coefficients},
  author = {Boris N. Khoromskij and Britta Schmitt and Volker Schulz},
  journal= {arXiv preprint arXiv:2105.13206},
  year   = {2021}
}
R2 v1 2026-06-24T02:31:58.437Z