Error Models

We provide error models valid to characterize priors and observation errors as well as model errors. All implementations provided here consider a design variable which enables activating/deactivating entries/dimensions when possible. Comprehensive work is done on the Gaussian Error Model since most observational errors are modeled as Gaussians. We also added Laplacian error models valid for Dolfin-based simulations to broaden the audience of the package.

• Gaussian Error Models
  • Standard Gaussian Error Models with Binary Design
  • Gaussian Error Models with Pre-Post-Weighted Design
  • Gaussian Error Models with Pointwise-Weighted Design
  • Gaussian Error Models for Fenics (Dolfin) Developments
  • Helper Functions

• Laplacian Error Models