Isaac 2025 Work Package #15

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open

iomags2211 wants to merge 49 commits into main from dev2

.gitignore

iomags2211 marked this conversation as resolved.

Show resolved Hide resolved

-Original file line number
+Diff line change
@@ Expand Up / @@ -158,3 +158,9 @@ cython_debug/ @@
     #  and can be added to the global gitignore or merged into this file.  For a more nuclear
     #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
     #.idea/
+    # Icarus-specific files
+    examples/example_outputs
+    src/icarus.egg-info/PKG-INFO
+    .vscode
+    moose-config.json

LICENSE

Large diffs are not rendered by default.

README.md

-Original file line number
+Diff line change
@@ -1,2 +1,62 @@
-    # icarus
-    Machine learning for fusion simulation validation.
+    # Icarus
+    A fully customisable set of simulation-driven machine learning tools that interacts with MOOSE and can be used to assess the agreement between an experiment and simulation; that is, to validate the simulation with experimental data and provide the engineer with a probable reason for any mismatches to allow further investigation and diagnosis.
+    ## Installation
+    ### Virtual Environment
+    Icarus is intended for use in a virtual environment. You can create a new virtual environment with
+    ```
+    python -m venv venv
+    ```
+    and activate it with
+    ```
+    source venv/bin/activate
+    ```
+    ### Standard & Developer Installation
+    Clone `icarus` to your local system along with submodules using
+    ```
+    git clone --recurse-submodules git@github.com:aurora-multiphysics/icarus.git
+    ```
+    `cd` to the root directory of `icarus` and run
+    ```
+    pip install .
+    ```
+    Alternatively, to create an editable/developer installation of `icarus`
+    ```
+    pip install -e .
+    ```
+    ### MOOSE App
+    `icarus` is intended for use with any MOOSE app
+    It has been developed and tested using the `proteus` MOOSE app which can be found here: https://github.com/aurora-multiphysics/proteus. Follow the build instructions found on this page to install `proteus`.
+    ## Getting Started
+    The examples folder shows how to use `icarus` for different input files and with different levels of user control. Make sure to include a moose_config.json file wherever you run it, and change the values contained to match your system.
+    Icarus requires a moose_config.json file to be present wherever it is run; if there is not one already present, it will create one with the following values:
+    ```
+    {
+        "main_path": "path/to/moose",
+        "app_path": "path/to/YourMooseApp",
+        "app_name": "YourMooseApp-opt"
+    }
+    ```
+    Make sure to change these values to match your system before running Icarus.
+    ## Contributors
+    - Isaac Magee, UK Atomic Energy Authority, (iomags2211)
+    - Lloyd Fletcher, UK Atomic Energy Authority, (TheScepticalRabbit)
+    - Luke Humphrey, UK Atomic Energy Authority, (lukethehuman)

examples/ex1_2d_full_control.py

iomags2211 marked this conversation as resolved.

Show resolved Hide resolved

-Original file line number
+Diff line change
@@ -0,0 +1,144 @@
+    """
+    Example file showing how to use Icarus and detailing customisable options.
+    This example runs an input file to produce a 2D temperature field with
+    different parameters perturbed. Any results from the default input file
+    are considered valid, while any other results are invalid. The datasets
+    generated are split into training and validation datasets and labelled,
+    and then a machine learning model is trained using the training dataset
+    to distinguish between valid and invalid datasets (and classify the
+    invalid sets depending on which parameter has been altered, if desired).
+    The model is then validated using the validation datasets to see if it
+    correctly predicted the labels.
+    (c) Copyright UKAEA 2025.
+    """
+    import sklearn
+    from icarus import (DatasetGenerator,
+                        ModelBuilder)
+    def main():
+        """Run all of the other functions.
+        """
+        # Required parameters - change to desired values
+        # Input and output paths
+        input_file_path = "scripts/moose/plate_2d_thermal.i"
+        output_file_path = "examples/example_outputs/ex1_outputs/"
+        # Perturbation parameters, leave blank (ie parameters = {}) to use
+        # auto-generated tkinter interface
+        # Format as {param_name, [param_class, [param_values]]}
+        # where param_name is the name of the parameter from the input file
+        # param_class is the classification of the parameter
+        #   geom - geometry, bc - boundary condtion, mat_prop - material property
+        # and param_values is a list of values the parameter should take on
+        parameters = {
+            "max_temp": ["bc", [750, 1000, 1250, 1500, 1750]],
+            "thermal_conductivity": ["mat_prop", [55, 65, 76, 85, 95]]
+        }
+        # Number of validation values to use for each parameter, respectively.
+        # This defines how many different perturbed values each parameter will
+        # take to create validation datasets to assess the quality of the model.
+        num_validation_values = [3, 3]
+        # Parallelisation options
+        n_tasks, n_threads = 1, 2  # Parallelisation of each simulation
+        num_para_runs = 2  # Number of concurrent simulations
+        # Ratio of invalid:valid datasets
+        # This defines how many datasets will be generated from the default
+        # input file (no perturbations -> labelled 0 for valid dataset),
+        # as the right ratio is important to ensure the model can recognise
+        # a valid dataset without assuming all datasets are valid.
+        datasets_per_ground_truth = 3
+        # Modelling parameters - change to desired values:
+        # Classifier framework
+        # For a comprehensive list of available classifiers, please refer
+        # to the relevant documentation:
+        # https://scikit-learn.org/stable/supervised_learning.html
+        classifier_framework = sklearn.ensemble.RandomForestClassifier
+        # Classifier parameters
+        # Note that not all parameters are required for every classifier.
+        # Any irrelevant or unsupported parameters for the selected model
+        # will be automatically ignored during model creation.
+        # For a comprehensive list of available parameters for each classifier,
+        # please refer to the relevant documentation:
+        # https://scikit-learn.org/stable/modules/classes.html#classifier
+        classifier_params = {
+            "n_estimators": 100,
+            "random_state": 42,
+            "kernel": "linear",
+            "C": 0.025,
+            "max_depth": 5
+        }
+        # The field being analysed as used by your input script
+        field_key = "temperature"
+        # Analysis sensor type
+        # (thermocouples for temperature, disp_sensors for displacement,
+        # or strain_gauges for strain)
+        sensor_type = "thermocouples"
+        # Sensor arrangement (x_sensors, y_sensors, z_sensors)
+        # Defines how many sensors are present in each dimension,
+        # with sensors being uniformly distributed.
+        # e.g. (3,2,1) would result in a 2D arrangement as follows:
+        #  _______________
+        # |               |
+        # |   x   x   x   |
+        # |   x   x   x   |
+        # |_______________|
+        sensors = (3, 2, 1)
+        # Number of spatial dimensions being used
+        dims = 2
+        # Whether the sensors should include errors or give exact values.
+        # Including errors means the model is more likely to miscategorise,
+        # but a successful model will be much better in reality, where errors
+        # are unavoidable.
+        errors = False
+        # Whether the model should be a multi-classifier rather than binary,
+        # so it can distinguish between perturbations to different classes of
+        # invalid parameters (geometry, BCs, material properties) rather than
+        # just valid and invalid results
+        multi = True
+        # Whether the unlabelled data should be deleted
+        delete_datasets = False
+        # Whether the model should be saved as a .pkl file
+        # and what it should be called
+        save = False
+        model_name = "ex1_2d_model"
+        # Initialise dataset generator to generate unlabelled datasets
+        dataset_generator = DatasetGenerator(input_file_path,
+                                             n_tasks, n_threads,
+                                             parameters, num_validation_values,
+                                             datasets_per_ground_truth,
+                                             output_file_path,
+                                             num_para_runs)
+        # Generate the unlabelled perturbed, validation and ground truths datasets
+        dataset_generator.generate_datasets()
+        # Initialise model builder to set up, run, and (optionally)
+        # save the chosen model
+        model = ModelBuilder(output_file_path, field_key, sensor_type, sensors,
+                             dims, errors, multi, delete_datasets, save)
+        # Set up, run, and (optionally) save the chosen model
+        model.build_model(classifier_framework, classifier_params, model_name)
+    if __name__ == "__main__":
+        main()

pyproject.toml

-Original file line number
+Diff line change
@@ Expand Up / @@ -5,22 +5,28 @@ build-backend = "setuptools.build_meta" @@
     [project]
     name = "icarus"
     version = "0.1.0"
-    description = "A machine learning tool for fusion simulation validation."
+    description = "Machine learning for fusion simulation validation"
     authors = [
-      { name = "arjavp-ukaea", email = "arjav.poudel@ukaea.uk" },
+      { name = "iomags2211", email = "isaac.magee@ukaea.uk" },
+      { name = "arjavp-ukaea", email = "arjav.poudel@ukaea.uk"},
       { name = "barisc-ukaea", email = "baris.cavusoglu@ukaea.uk"},
-      { name = "lukethehuman", email = "luke.humphrey@ukaea.uk"},
+      { name = "lukethehuman", email = "luke.humphrey@ukaea.uk"}
     ]
     license = { file = "LICENSE" }
     readme = "README.md"
     classifiers = [
-        "Programming Language :: Python :: 3",
-        "License :: OSI Approved :: GNU General Public License v3 (GPLv3)",
+        "Programming Language :: Python :: 3",
+        "License :: GNU LGPL",
         "Operating System :: OS Independent",
     ]
     requires-python = ">=3.11"
     dependencies = [
       "numpy>=1.26.0",
+      "sklearn>=1.6.0",
+      "mooseherder>=0.1.0",
+      "pyvale>=2025.5.3",
+      "joblib>=1.4.0",
+      "pytest>=8.2.0"
     ]
     [project.urls]
@@ Expand Down @@

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Isaac 2025 Work Package #15

Uh oh!

Diff view

Diff view

There are no files selected for viewing

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Isaac 2025 Work Package #15

Are you sure you want to change the base?

Uh oh!

Isaac 2025 Work Package #15

Uh oh!

Uh oh!

Diff view

Diff view

There are no files selected for viewing

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!