make exp a little harder..

examples/Darcy/calibrate.jl

@@ -49,13 +49,13 @@ function main()
     dim = 2
     N, L = 80, 1.0
     pts_per_dim = LinRange(0, L, N)
-    obs_ΔN = 10
+    obs_ΔN = 20
 
     # To provide a simple test case, we assume that the true function parameter is a particular sample from the function space we set up to define our prior. More precisely we choose a value of the truth that doesnt have a vanishingly small probability under the prior defined by a probability distribution over functions; here taken as a family of Gaussian Random Fields (GRF). The function distribution is characterized by a covariance function - here a Matern kernel which assumes a level of smoothness over the samples from the distribution. We define an appropriate expansion of this distribution, here based on the Karhunen-Loeve expansion (similar to an eigenvalue-eigenfunction expansion) that is truncated to a finite number of terms, known as the degrees of freedom (`dofs`). The `dofs` define the effective dimension of the learning problem, decoupled from the spatial discretization. Explicitly, larger `dofs` may be required to represent multiscale functions, but come at an increased dimension of the parameter space and therefore a typical increase in cost and difficulty of the learning problem.
 
     smoothness = 1.0
     corr_length = 0.25
-    dofs = 5
+    dofs = 6
 
     grf = GRF.GaussianRandomField(
         GRF.CovarianceFunction(dim, GRF.Matern(smoothness, corr_length)),
@@ -84,7 +84,7 @@ function main()
     println(" Number of observation points: $(darcy.N_y)")
     h_2d_true = solve_Darcy_2D(darcy, κ_true)
     y_noiseless = compute_obs(darcy, h_2d_true)
-    obs_noise_cov = 0.05^2 * I(length(y_noiseless)) * (maximum(y_noiseless) - minimum(y_noiseless))
+    obs_noise_cov = 0.2^2 * I(length(y_noiseless)) * (maximum(y_noiseless) - minimum(y_noiseless))
     truth_sample = vec(y_noiseless + rand(rng, MvNormal(zeros(length(y_noiseless)), obs_noise_cov)))
 
 
@@ -93,7 +93,7 @@ function main()
 
 
     # We define some algorithm parameters, here we take ensemble members larger than the dimension of the parameter space
-    N_ens = 50 # number of ensemble members
+    N_ens = 30 # number of ensemble members
     N_iter = 5 # number of EKI iterations
 
     # We sample the initial ensemble from the prior, and create the EKP object as an EKI algorithm using the `Inversion()` keyword

examples/Darcy/emulate_sample.jl

@@ -22,7 +22,6 @@ function main()
 
     cases = [
         "GP", # diagonalize, train scalar GP, assume diag inputs
-        "RF-vector-svd-nonsep",
     ]
 
     #### CHOOSE YOUR CASE: 
@@ -33,25 +32,6 @@ function main()
         println("case: ", case)
         min_iter = 1
         max_iter = 5 # number of EKP iterations to use data from is at most this
-        overrides = Dict(
-            "verbose" => true,
-            "scheduler" => DataMisfitController(terminate_at = 100.0),
-            "cov_sample_multiplier" => 0.5,
-            "n_iteration" => 20,
-            "n_ensemble" => 120,
-            "localization" => Localizers.SECNice(1000, 0.1, 1.0),
-        )
-        # we do not want termination, as our priors have relatively little interpretation
-
-        # Should be loaded:
-        dim = 2
-        N, L = 80, 1.0
-        pts_per_dim = LinRange(0, L, N)
-        obs_ΔN = 10
-        smoothness = 1.0
-        corr_length = 0.25
-        dofs = 20 # i.e. the input space dimension
-        ####
 
         exp_name = "darcy"
         rng_seed = 940284
@@ -93,18 +73,6 @@ function main()
             #            gppackage = Emulators.SKLJL()
             gppackage = Emulators.GPJL()
             mlt = GaussianProcess(gppackage; noise_learn = false)
-        elseif case ∈ ["RF-vector-svd-nonsep"]
-            kernel_structure = NonseparableKernel(LowRankFactor(1, nugget))
-            n_features = 100
-
-            mlt = VectorRandomFeatureInterface(
-                n_features,
-                n_params,
-                output_dim,
-                rng = rng,
-                kernel_structure = kernel_structure,
-                optimizer_options = overrides,
-            )
         end