Merge pull request #89 from KratosMultiphysics/MLMC/correct-problem-0

[MLMC] Update problems

Author: riccardotosi

README.md

@@ -140,9 +140,6 @@ Unit tests should *not* be uploaded to this repository. Please put them in the `
 - [Compressible potential flow problem](multilevel_monte_carlo/use_cases/compressible_potential_flow/README.md)
 - [Fluid dynamics building problem](multilevel_monte_carlo/use_cases/fluid_dynamics_building)
 - [Wind engineering rectangle problem](multilevel_monte_carlo/use_cases/wind_engineering_rectangle)
-    - [Deterministic wind engineering rectangle problem with ensemble average approach](multilevel_monte_carlo/use_cases/wind_engineering_rectangle/deterministic_ensemble_average)
-    - [Stochastic wind engineering rectangle problem](multilevel_monte_carlo/use_cases/wind_engineering_rectangle/stochastic_MC)
-    - [Stochastic wind engineering rectangle problem with ensemble average approach](multilevel_monte_carlo/use_cases/wind_engineering_rectangle/stochastic_MC_ensemble_average)
 - [Wind engineering CAARC problem](multilevel_monte_carlo/use_cases/wind_engineering_CAARC)
     - [Steady inlet wind engineering CAARC problem with ensemble average approach](multilevel_monte_carlo/use_cases/wind_engineering_CAARC/steady_inlet_monte_carlo)
     - [Turbulent inlet wind engineering CAARC problem with ensemble average approach](multilevel_monte_carlo/use_cases/wind_engineering_CAARC/turbulent_inlet_monte_carlo)

multilevel_monte_carlo/README.md

@@ -4,9 +4,6 @@
 - [Compressible potential flow problem](use_cases/compressible_potential_flow/README.md)
 - [Fluid dynamics building problem](use_cases/fluid_dynamics_building)
 - [Wind engineering rectangle problem](use_cases/wind_engineering_rectangle)
-    - [Deterministic wind engineering rectangle problem with ensemble average approach](use_cases/wind_engineering_rectangle/deterministic_ensemble_average)
-    - [Stochastic wind engineering rectangle problem](use_cases/wind_engineering_rectangle/stochastic_MC)
-    - [Stochastic wind engineering rectangle problem with ensemble average approach](use_cases/wind_engineering_rectangle/stochastic_MC_ensemble_average)
 - [Wind engineering CAARC problem](use_cases/wind_engineering_CAARC)
     - [Steady inlet wind engineering CAARC problem with ensemble average approach](use_cases/wind_engineering_CAARC/steady_inlet_monte_carlo)
     - [Turbulent inlet wind engineering CAARC problem with ensemble average approach](use_cases/wind_engineering_CAARC/turbulent_inlet_monte_carlo)

multilevel_monte_carlo/use_cases/README.md

@@ -5,9 +5,6 @@ This folder contains the use cases:
 - [Compressible potential flow problem](compressible_potential_flow/README.md)
 - [Fluid dynamics building problem](fluid_dynamics_building)
 - [Wind engineering rectangle problem](wind_engineering_rectangle)
-    - [Deterministic wind engineering rectangle problem with ensemble average approach](wind_engineering_rectangle/deterministic_ensemble_average)
-    - [Stochastic wind engineering rectangle problem](wind_engineering_rectangle/stochastic_MC)
-    - [Stochastic wind engineering rectangle problem with ensemble average approach](wind_engineering_rectangle/stochastic_MC_ensemble_average)
 - [Wind engineering CAARC problem](wind_engineering_CAARC)
     - [Steady inlet wind engineering CAARC problem with ensemble average approach](wind_engineering_CAARC/steady_inlet_monte_carlo)
     - [Turbulent inlet wind engineering CAARC problem with ensemble average approach](wind_engineering_CAARC/turbulent_inlet_monte_carlo)

multilevel_monte_carlo/use_cases/compressible_potential_flow/README.md

@@ -1,6 +1,6 @@
 # Compressible Potential Flow problem
 
-**Author:** [Riccardo Tosi](https://github.com/riccardotosi)
+**Author:** [Riccardo Tosi](https://riccardotosi.github.io)
 
 **Kratos version:** 8.1
 
@@ -22,7 +22,7 @@ The problem can be run with four different algorithms:
 * Synchronous Multilevel Monte Carlo (SMLMC),
 * Asynchronous Multilevel Monte Carlo (AMLMC),
 
-and by default AMC and AMLMC are selected. If one is interested in running SMC or SMLMC, it is needed to select `asynchronous = false` in the solver wrapper settings.
+and by default AMC and AMLMC are selected. If one is interested in running SMC or SMLMC, it is needed to select `asynchronous = false` in the solver wrapper settings. Please observe that for running you may want to increase the number of realizations per level.
 
 The Quantities of Interest of the problem are the lift coefficient and the pressure coefficient. Statistical convergence is assessed for the lift coefficient.
 

multilevel_monte_carlo/use_cases/compressible_potential_flow/source/run_mc_Kratos.py

@@ -95,7 +95,7 @@
     with open(parameters["solverWrapperInputDictionary"]["projectParametersPath"],'r') as parameter_file:
         project_parameters = json.load(parameter_file)
     pickled_model = algo.monteCarloSampler.indices[0].sampler.solvers[0].pickled_model[0]
-    serialized_model = pickle.loads(pickled_model)
+    serialized_model = pickle.loads(get_value_from_remote(pickled_model))
     current_model = KratosMultiphysics.Model()
     serialized_model.Load("ModelSerialization",current_model)
     model_part_of_interest = "model.Body2D_Body"

multilevel_monte_carlo/use_cases/compressible_potential_flow/source/run_mlmc_Kratos.py

@@ -96,7 +96,7 @@
     with open(parameters["solverWrapperInputDictionary"]["projectParametersPath"][0],'r') as parameter_file:
         project_parameters = json.load(parameter_file)
     pickled_model = algo.monteCarloSampler.indices[0].sampler.solvers[0].pickled_model[0]
-    serialized_model = pickle.loads(pickled_model)
+    serialized_model = pickle.loads(get_value_from_remote(pickled_model))
     current_model = KratosMultiphysics.Model()
     serialized_model.Load("ModelSerialization",current_model)
     model_part_of_interest = "model.Body2D_Body"

multilevel_monte_carlo/use_cases/compressible_potential_flow/source/run_runcompss.sh

@@ -1,8 +1,34 @@
 #!/usr/bin/env bash
 
-path_to_folder=$(pwd)
+path_to_examples_folder=$(pwd)
+
+materials_path_to_replace_poisson="problem_settings/materials.json"
+materials_new_path_poisson="$path_to_examples_folder/problem_settings/materials.json"
+mdpa_path_to_replace_0="problem_settings/naca0012Mesh4"
+mdpa_new_path_0="$path_to_examples_folder/problem_settings/naca0012Mesh4"
+mdpa_path_to_replace_1="problem_settings/CPS_MONTECARLO_MeshInterpError1e-2"
+mdpa_new_path_1="$path_to_examples_folder/problem_settings/CPS_MONTECARLO_MeshInterpError1e-2"
+mdpa_path_to_replace_2="problem_settings/CPS_MONTECARLO_MeshInterpError5e-3"
+mdpa_new_path_2="$path_to_examples_folder/problem_settings/CPS_MONTECARLO_MeshInterpError5e-3"
+
+# set absolute path in Kratos parameters
+sed -i "s|$materials_path_to_replace_poisson|$materials_new_path_poisson|g" "problem_settings/parameters_potential_naca_lev0.json"
+sed -i "s|$materials_path_to_replace_poisson|$materials_new_path_poisson|g" "problem_settings/parameters_potential_naca_lev1.json"
+sed -i "s|$materials_path_to_replace_poisson|$materials_new_path_poisson|g" "problem_settings/parameters_potential_naca_lev2.json"
+sed -i "s|$mdpa_path_to_replace_0|$mdpa_new_path_0|g" "problem_settings/parameters_potential_naca_lev0.json"
+sed -i "s|$mdpa_path_to_replace_1|$mdpa_new_path_1|g" "problem_settings/parameters_potential_naca_lev1.json"
+sed -i "s|$mdpa_path_to_replace_2|$mdpa_new_path_2|g" "problem_settings/parameters_potential_naca_lev2.json"
+
 runcompss \
     --lang=python \
     --python_interpreter=python3 \
     --pythonpath=$path_to_folder \
     ./run_mc_Kratos.py problem_settings/parameters_xmc_asynchronous_mc_potentialFlow.json
+
+# revert change in Kratos parameters
+sed -i "s|$materials_new_path_poisson|$materials_path_to_replace_poisson|g" "problem_settings/parameters_potential_naca_lev0.json"
+sed -i "s|$materials_new_path_poisson|$materials_path_to_replace_poisson|g" "problem_settings/parameters_potential_naca_lev1.json"
+sed -i "s|$materials_new_path_poisson|$materials_path_to_replace_poisson|g" "problem_settings/parameters_potential_naca_lev2.json"
+sed -i "s|$mdpa_new_path_0|$mdpa_path_to_replace_0|g" "problem_settings/parameters_potential_naca_lev0.json"
+sed -i "s|$mdpa_new_path_1|$mdpa_path_to_replace_1|g" "problem_settings/parameters_potential_naca_lev1.json"
+sed -i "s|$mdpa_new_path_2|$mdpa_path_to_replace_2|g" "problem_settings/parameters_potential_naca_lev2.json"

multilevel_monte_carlo/use_cases/fluid_dynamics_building/README.md

@@ -1,12 +1,12 @@
 # Fluid dynamics building problem
 
-**Author:** [Riccardo Tosi](https://github.com/riccardotosi)
+**Author:** [Riccardo Tosi](https://riccardotosi.github.io)
 
 **Kratos version:** 8.1
 
-**XMC version:** 2.0
+**XMC version:** Kratos default version
 
-**PyCOMPSs version:** 2.7
+**PyCOMPSs version:** Kratos default version to run in serial, >2.8 to run with `runcompss`
 
 **Source files:** [Asynchronous Monte Carlo and Asynchronous Multilevel Monte Carlo](source)
 
@@ -22,7 +22,7 @@ The problem can be run with four different algorithms:
 * Synchronous Multilevel Monte Carlo (SMLMC),
 * Asynchronous Multilevel Monte Carlo (AMLMC),
 
-and by default AMC and AMLMC are selected. If one is interested in running SMC or SMLMC, it is needed to select `asynchronous = false` in the solver wrapper settings.
+and by default AMC and AMLMC are selected. If one is interested in running SMC or SMLMC, it is needed to select `asynchronous = false` in the solver wrapper settings. Please observe that for running you may want to increase the number of realizations per level and the time horizon of each realization.
 
 The Quantities of Interest of the problem are the drag force and the pressure field. Statistical convergence is assessed for the drag force.