1: \begin{abstract}
2: Inexpensive surrogates are useful for reducing the cost of science and engineering studies
3: involving large-scale, complex computational models with many input parameters.
4: A ridge approximation is one class of surrogate
5: that models a quantity of interest
6: as a nonlinear function of a few linear combinations of the input parameters.
7: When used in parameter studies (e.g., optimization or uncertainty quantification),
8: ridge approximations allow the low-dimensional structure to be exploited,
9: reducing the effective dimension.
10: We introduce a new, fast algorithm for constructing a
11: ridge approximation where the nonlinear function is a polynomial.
12: This polynomial ridge approximation is chosen to minimize least squared mismatch between
13: the surrogate and the quantity of interest on a given set of inputs.
14: Naively, this would require optimizing both the polynomial coefficients and
15: the linear combination of weights; the latter of which define a low-dimensional subspace of the input space.
16: However, given a fixed subspace the optimal polynomial can be found by solving a linear least-squares problem.
17: Hence using \emph{variable projection} the polynomial can be implicitly defined,
18: leaving an optimization problem over the subspace alone.
19: Here we develop an algorithm that finds this polynomial ridge approximation
20: by minimizing over the Grassmann manifold of low-dimensional subspaces
21: using a Gauss-Newton method.
22: %We provide details of this optimization algorithm
23: %and demonstrate its performance on several numerical examples.
24: Our Gauss-Newton method has superior theoretical guarantees and faster convergence
25: on our numerical examples than the alternating approach for polynomial ridge approximation earlier proposed by
26: Constantine, Eftekhari, Hokanson, and Ward
27: [{\tt https://doi.org/10.1016/j.cma.2017.07.038}]
28: that alternates between (i) optimizing the polynomial coefficients given the subspace and
29: (ii) optimizing the subspace given the coefficients.
30: \end{abstract}
31: