format

Author: odunbar

docs/Manifest.toml

@@ -211,7 +211,13 @@ u_init = get_u_prior(ekiobj)
 anim_eki_unconst_cloudy = @animate for i in 1:(n_iter - 1)
     u_i = get_u(ekiobj, i)
 
-    p1 = Plots.plot(u_i[1, :], u_i[2, :], seriestype = :scatter, xlims = extrema(u_init[1, :]), ylims = extrema(u_init[2, :]))
+    p1 = Plots.plot(
+        u_i[1, :],
+        u_i[2, :],
+        seriestype = :scatter,
+        xlims = extrema(u_init[1, :]),
+        ylims = extrema(u_init[2, :]),
+    )
     Plots.plot!(
         p1,
         [θ_true[1]],
@@ -226,7 +232,13 @@ anim_eki_unconst_cloudy = @animate for i in 1:(n_iter - 1)
     )
     Plots.plot!(p1, [θ_true[2]], seriestype = "hline", linestyle = :dash, linecolor = :red, label = "optimum")
 
-    p2 = Plots.plot(u_i[2, :], u_i[3, :], seriestype = :scatter, xlims = extrema(u_init[2, :]), ylims = extrema(u_init[3, :]))
+    p2 = Plots.plot(
+        u_i[2, :],
+        u_i[3, :],
+        seriestype = :scatter,
+        xlims = extrema(u_init[2, :]),
+        ylims = extrema(u_init[3, :]),
+    )
     Plots.plot!(
         p2,
         [θ_true[2]],
@@ -242,7 +254,13 @@ anim_eki_unconst_cloudy = @animate for i in 1:(n_iter - 1)
 
     Plots.plot!(p2, [θ_true[3]], seriestype = "hline", linestyle = :dash, linecolor = :red, label = "optimum")
 
-    p3 = Plots.plot(u_i[3, :], u_i[1, :], seriestype = :scatter, xlims = extrema(u_init[3, :]), ylims = extrema(u_init[1, :]))
+    p3 = Plots.plot(
+        u_i[3, :],
+        u_i[1, :],
+        seriestype = :scatter,
+        xlims = extrema(u_init[3, :]),
+        ylims = extrema(u_init[1, :]),
+    )
     Plots.plot!(
         p3,
         [θ_true[3]],
@@ -268,7 +286,13 @@ gif(anim_eki_unconst_cloudy, joinpath(output_directory, "cloudy_eki_unconstr.gif
 anim_eki_cloudy = @animate for i in 1:(n_iter - 1)
     ϕ_i = get_ϕ(priors, ekiobj, i)
 
-    p1 = Plots.plot(ϕ_i[1, :], ϕ_i[2, :], seriestype = :scatter, xlims = extrema(ϕ_init[1, :]), ylims = extrema(ϕ_init[2, :]))
+    p1 = Plots.plot(
+        ϕ_i[1, :],
+        ϕ_i[2, :],
+        seriestype = :scatter,
+        xlims = extrema(ϕ_init[1, :]),
+        ylims = extrema(ϕ_init[2, :]),
+    )
     Plots.plot!(
         p1,
         [ϕ_true[1]],
@@ -284,7 +308,13 @@ anim_eki_cloudy = @animate for i in 1:(n_iter - 1)
 
     Plots.plot!(p1, [ϕ_true[2]], seriestype = "hline", linestyle = :dash, linecolor = :red, label = "optimum")
 
-    p2 = Plots.plot(ϕ_i[2, :], ϕ_i[3, :], seriestype = :scatter, xlims = extrema(ϕ_init[2, :]), ylims = extrema(ϕ_init[3, :]))
+    p2 = Plots.plot(
+        ϕ_i[2, :],
+        ϕ_i[3, :],
+        seriestype = :scatter,
+        xlims = extrema(ϕ_init[2, :]),
+        ylims = extrema(ϕ_init[3, :]),
+    )
 
     Plots.plot!(
         p2,
@@ -301,7 +331,13 @@ anim_eki_cloudy = @animate for i in 1:(n_iter - 1)
 
     Plots.plot!(p2, [ϕ_true[3]], seriestype = "hline", linestyle = :dash, linecolor = :red, label = "optimum")
 
-    p3 = Plots.plot(ϕ_i[3, :], ϕ_i[1, :], seriestype = :scatter, xlims = extrema(ϕ_init[3, :]), ylims = extrema(ϕ_init[1, :]))
+    p3 = Plots.plot(
+        ϕ_i[3, :],
+        ϕ_i[1, :],
+        seriestype = :scatter,
+        xlims = extrema(ϕ_init[3, :]),
+        ylims = extrema(ϕ_init[1, :]),
+    )
     Plots.plot!(
         p3,
         [ϕ_true[3]],

examples/Cloudy/Cloudy_calibrate.jl

@@ -96,14 +96,7 @@ function main()
 
 
     # Emulate
-    cases = [
-        "GP",
-        "RF-prior",
-        "RF-lr-scalar",
-        "rf-diag-scalar",
-        "RF-lr-lr",
-        "RF-nonsep",
-    ]
+    cases = ["GP", "RF-prior", "RF-lr-scalar", "rf-diag-scalar", "RF-lr-lr", "RF-nonsep"]
 
     case = cases[4] # 7
 

examples/Emulator/L63/emulate.jl

@@ -29,7 +29,7 @@ function main()
     ]
 
     #### CHOOSE YOUR CASE: 
-    mask =[2] # e.g. 1:2 or [3]
+    mask = [2] # e.g. 1:2 or [3]
     for (case) in cases[mask]
 
 
@@ -92,7 +92,7 @@ function main()
                 "n_features_opt" => 160,
             )
             n_features = 200
-#            kernel_structure = SeparableKernel(LowRankFactor(1,nugget), OneDimFactor())
+            #            kernel_structure = SeparableKernel(LowRankFactor(1,nugget), OneDimFactor())
             kernel_structure = SeparableKernel(DiagonalFactor(nugget), OneDimFactor())
             mlt = ScalarRandomFeatureInterface(
                 n_features,
@@ -112,7 +112,7 @@ function main()
             )
             kernel_structure = NonseparableKernel(LowRankFactor(1, nugget))
             n_features = 200
-            
+
             mlt = VectorRandomFeatureInterface(
                 n_features,
                 n_params,
@@ -165,9 +165,7 @@ function main()
 
         # data processing configuration
         retain_var = 0.95
-        encoder_schedule = [
-            (decorrelate_structure_mat(retain_var = retain_var), "in_and_out"),
-        ]
+        encoder_schedule = [(decorrelate_structure_mat(retain_var = retain_var), "in_and_out")]
 
         emulator = Emulator(
             mlt,

examples/Lorenz/emulate_sample_spatial_dep.jl

@@ -212,7 +212,7 @@ function hyperparameters_from_flat(x::V, lrf::LowRankFactor) where {V <: Abstrac
 
     D = Diagonal(x[1:d]) # D acts like sing.vals.^2
     U = reshape(x[(d + 1):end], d, r)
-    return U * U' + D + get_eps(lrf)*I
+    return U * U' + D + get_eps(lrf) * I
 
 end
 
@@ -259,11 +259,11 @@ function build_default_prior(name::SS, n_hp::Int, lrf::LowRankFactor) where {SS
         Dict(
             "name" => "$(name)_lowrank_diagonal",
             "distribution" => VectorOfParameterized(repeat([Normal(log(sqrt(0.1 / d)), 2)], d)),
-#            "distribution" => VectorOfParameterized(repeat([Normal(log(1.0 / d), 2)], d)),
+            #            "distribution" => VectorOfParameterized(repeat([Normal(log(1.0 / d), 2)], d)),
             "constraint" => repeat([bounded_below(1e-8 / d)], d),
         ),
     )
-#    U = constrained_gaussian("$(name)_lowrank_U", 0.0, 10.0 / (d * r), -Inf, Inf, repeats = Int(d * r))
+    #    U = constrained_gaussian("$(name)_lowrank_U", 0.0, 10.0 / (d * r), -Inf, Inf, repeats = Int(d * r))
     U = constrained_gaussian("$(name)_lowrank_U", 0.0, sqrt(0.1 / d) / r, -Inf, Inf, repeats = Int(d * r))
     # D scales with sqrt(1/d), U scales with 1/r,sqrt(1/d) 
     return combine_distributions([D, U])
@@ -298,10 +298,10 @@ function calculate_n_hyperparameters(
 
     # if both in and output dim are "OneDimFactors" we still learn 1 hp
     if n_hp_in + n_hp_out == 0
-        return 1 
+        return 1
     end
 
-    return n_hp_in + n_hp_out 
+    return n_hp_in + n_hp_out
 end
 
 function calculate_n_hyperparameters(input_dim::Int, kernel_structure::KST) where {KST <: KernelStructureType}
@@ -319,10 +319,10 @@ function calculate_n_hyperparameters(
 
     # if both in and output dim are "OneDimFactors" we still learn 1 hp
     if n_hp == 0
-        return 1 
+        return 1
     end
 
-    return n_hp 
+    return n_hp
 end
 
 
@@ -352,7 +352,7 @@ function build_default_prior(input_dim::Int, output_dim::Int, kernel_structure::
     else
         return combine_distributions([input_prior, output_prior])
     end
-    
+
 end
 
 function build_default_prior(input_dim::Int, kernel_structure::KST) where {KST <: KernelStructureType}
@@ -370,7 +370,7 @@ function build_default_prior(input_dim::Int, output_dim::Int, kernel_structure::
     end
 
     return pd_kern
-    
+
 end
 
 function hyperparameters_from_flat(

src/RandomFeature.jl

@@ -277,7 +277,7 @@ function RFM_from_hyperparameters(
     MT <: MultithreadType,
 }
 
-    xi_hp= isa(l, AbstractVecOrMat) ? l[:] : [l]
+    xi_hp = isa(l, AbstractVecOrMat) ? l[:] : [l]
     kernel_structure = get_kernel_structure(srfi)
     pd = hyperparameter_distribution_from_flat(xi_hp, input_dim, kernel_structure, prior_in_scale)
 
@@ -437,10 +437,10 @@ function build_models!(
         end
 
         # scale up the prior so that default priors are always "reasonable"
-        prior_in_scale = 1.0 ./ std(input_values, dims = 2) 
+        prior_in_scale = 1.0 ./ std(input_values, dims = 2)
         prior_out_scale = std(output_values[i, :])
-        
-        
+
+
         prior = build_default_prior(input_dim, kernel_structure)
 
         # where prior space has changed we need to rebuild the priors
@@ -508,12 +508,12 @@ function build_models!(
         scheduler = optimizer_options["scheduler"]
         accelerator = optimizer_options["accelerator"]
         localization = optimizer_options["localization"]
-        
+
         initial_params = construct_initial_ensemble(rng, prior, n_ensemble)
         # bug with scalar mean o/w
         prior_mean = isa(mean(prior), AbstractVector) ? mean(prior) : [mean(prior)]
         prior_cov = cov(prior)
-        
+
         ekiobj = EKP.EnsembleKalmanProcess(
             initial_params,
             observation,

src/ScalarRandomFeature.jl

@@ -330,7 +330,7 @@ function RFM_from_hyperparameters(
     MT <: MultithreadType,
 }
 
-    xi_hp= isa(l, AbstractVecOrMat) ? l[:] : [l]
+    xi_hp = isa(l, AbstractVecOrMat) ? l[:] : [l]
     prior_out_scale_scalar = maximum(prior_out_scale) # most conservative scaling
 
     kernel_structure = get_kernel_structure(vrfi)
@@ -566,7 +566,7 @@ function build_models!(
 
     initial_params = construct_initial_ensemble(rng, prior, n_ensemble)
     # bug with scalar mean o/w
-    prior_mean = isa(mean(prior), AbstractVector) ? mean(prior) : [mean(prior) ]
+    prior_mean = isa(mean(prior), AbstractVector) ? mean(prior) : [mean(prior)]
     prior_cov = cov(prior)
 
     ekiobj = EKP.EnsembleKalmanProcess(

src/VectorRandomFeature.jl

@@ -104,7 +104,7 @@ rng = Random.MersenneTwister(seed)
               ndims(build_default_prior("input", calculate_n_hyperparameters(d, c_in), c_in)) +
               ndims(build_default_prior("output", calculate_n_hyperparameters(p, c_out), c_out))
         @test ndims(build_default_prior(d, p, k_nonsep)) ==
-              ndims(build_default_prior("full", calculate_n_hyperparameters(d * p, c_in), c_in)) 
+              ndims(build_default_prior("full", calculate_n_hyperparameters(d * p, c_in), c_in))
 
         @test ndims(build_default_prior(1, 1, k_sep1d)) == 1
         @test ndims(build_default_prior(1, 1, k_nonsep1d)) == 1