ERT and Everest settings#

History matching requires cores, min_realizations_success, and at least one facies or thickness parameter.

ERT#

ERT additionally requires ertargs, ensembles, enkf_alpha, and positive errors. Optional random_seed controls reproducibility.

Everest#

Everest requires max_function_evaluations and popsize. Supported differential-evolution options include strategy, maxiter, tol, mutation, recombination, rng, disp, polish, init, atol, updating, workers, x0, integrality, and vectorized.

History-matching parameters#

ERT parameters use a distribution name and two parameters. Everest parameters use [initial, minimum, maximum, scale]. Facies properties may be selected by their numbered key, isotropic permeability may use permN, and thicknessmult controls the thickness multiplier.

monotonic = true only has an effect when at least two optimized variables form a comparable same-property facies sequence.