Simplify EKP config slightly

docs/src/experiment_setup_guide.md

@@ -215,15 +215,13 @@ The EKP configuration file must contain the following:
 - `n_iterations`, the number of iterations to run
 - `ensemble_size`, the ensemble size
 - `prior_path`, the path to the TOML file with the prior parameter distributions
-- `parameter_names`, a list of the long names of the parameters being calibrated 
 - `truth_data`, the truth data
 - `truth_noise`, the covariance of the truth data
 - `output_dir`, the folder where you want calibration data and logs to be output. This must be the same as the `output_dir` in the model configuration file.
 Example:
 ```
 output_dir: output/sphere_held_suarez_rhoe_equilmoist
 prior_path: experiments/sphere_held_suarez_rhoe_equilmoist/prior.toml
-parameter_names: ["equator_pole_temperature_gradient_wet"]
 ensemble_size: 10
 n_iterations: 3
 truth_data: experiments/sphere_held_suarez_rhoe_equilmoist/obs_mean.jld2

src/atmos_interface.jl

@@ -17,8 +17,7 @@ Turns off default diagnostics and sets the TOML parameter file to the member's p
 This assumes that the config dictionary has `output_dir` and `restart_file` keys.
 """
 function get_atmos_config(member, iteration, config_dict::AbstractDict)
-    # Specify member path for output_dir
-    # Set TOML to use EKP parameter(s)
+    # Specify member path for output_dir and set TOML to use EKP parameter(s)
     output_dir = config_dict["output_dir"]
     member_path =
         EKP.TOMLInterface.path_to_ensemble_member(output_dir, iteration, member)

src/ekp_interface.jl

@@ -52,15 +52,11 @@ function initialize(
     rng_ekp = Random.MersenneTwister(rng_seed)
 
     output_dir = config["output_dir"]
-    prior_path = config["prior_path"]
-    param_names = config["parameter_names"]
     ensemble_size = config["ensemble_size"]
     # Save in EKI object in iteration_000 folder
     eki_path = joinpath(output_dir, "iteration_000", "eki_file.jld2")
 
-    param_dict = TOML.parsefile(prior_path)
-    prior_vec = [get_parameter_distribution(param_dict, n) for n in param_names]
-    prior = combine_distributions(prior_vec)
+    prior = get_prior(config["prior_path"])
 
     initial_ensemble =
         EKP.construct_initial_ensemble(rng_ekp, prior, ensemble_size)
@@ -100,7 +96,6 @@ function update_ensemble(
     config = YAML.load_file("experiments/$experiment_id/ekp_config.yml"),
 )
     output_dir = config["output_dir"]
-    names = config["parameter_names"]
     # Load EKI object from iteration folder
     iter_path = path_to_iteration(output_dir, iteration)
     eki_path = joinpath(iter_path, "eki_file.jld2")
@@ -114,10 +109,8 @@ function update_ensemble(
     iteration += 1
 
     # Update and save parameters for next iteration
-    prior_path = config["prior_path"]
-    param_dict = TOML.parsefile(prior_path)
-    prior_vec = [get_parameter_distribution(param_dict, n) for n in names]
-    prior = combine_distributions(prior_vec)
+    prior = get_prior(config["prior_path"])
+
     save_parameter_ensemble(
         EKP.get_u_final(eki),  # constraints applied when saving
         prior,

test/test_init.jl

@@ -12,7 +12,6 @@ param_names = ["one", "two"]
 config = Dict(
     "output_dir" => output_dir,
     "prior_path" => prior_path,
-    "parameter_names" => param_names,
     "ensemble_size" => 10,
 )
 Γ = 0.1 * I