ostrich: Surrogate modeling using PCA and Gaussian process interpolation
Abstract
Ostrich emulates surrogate models for complex and expensive functions using Principal Component Analysis (PCA) to decompose a signal, then interpolate the PCA weights over the parameters θ using a Gaussian Process interpolator. The code is trained on samples from the expensive functions, recreating and interpolating between those training samples with reduced computational cost, and recalculating for each use.
- Publication:
-
Astrophysics Source Code Library
- Pub Date:
- January 2024
- Bibcode:
- 2024ascl.soft01011C
- Keywords:
-
- Software