.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "_examples/00_basic_simai_ex/02-run_predictions.py"
.. LINE NUMBERS ARE GIVEN BELOW.

.. only:: html

    .. note::
        :class: sphx-glr-download-link-note

        :ref:`Go to the end <sphx_glr_download__examples_00_basic_simai_ex_02-run_predictions.py>`
        to download the full example code.

.. rst-class:: sphx-glr-example-title

.. _sphx_glr__examples_00_basic_simai_ex_02-run_predictions.py:

.. _ref_basic_run_predictions:

Running Physics Predictions
================================================

This example demonstrates how to upload geometries to a workspace and run predictions
using a trained SimAI model. It also shows how to extract and save prediction results,
including global coefficients and confidence scores.

Before you begin
----------------------------------------------------

- Complete ":ref:`ref_basic_build_model`" to train a SimAI model.
- Ensure the model training completed successfully.
- Have a dataset folder with subdirectories containing geometry files. These geometry files can come from a Generative Design model.
- (Optional) Prepare boundary condition JSON files if your model requires them.

.. GENERATED FROM PYTHON SOURCE LINES 43-45

Import necessary libraries
----------------------------------------------------

.. GENERATED FROM PYTHON SOURCE LINES 45-53

.. code-block:: Python


    import json
    import os

    import ansys.simai.core as asc
    from ansys.simai.core.data.geometries import Geometry
    from ansys.simai.core.data.predictions import Prediction


.. GENERATED FROM PYTHON SOURCE LINES 54-57

Configure your settings
----------------------------------------------------
Update these variables with your specific settings:

.. GENERATED FROM PYTHON SOURCE LINES 57-62

.. code-block:: Python


    ORGANIZATION_NAME = "<your_organization>"  # Replace with your organization name
    WORKSPACE_NAME = "<your_workspace_name>"  # Replace with your workspace name
    DATASET_PATH = "<your_dataset>"  # Path to your dataset directory


.. GENERATED FROM PYTHON SOURCE LINES 63-66

Initialize the client and set workspace
----------------------------------------------------
Connect to SimAI and set the workspace to use for predictions:

.. GENERATED FROM PYTHON SOURCE LINES 66-69

.. code-block:: Python


    simai_client = asc.SimAIClient(organization=ORGANIZATION_NAME)


.. GENERATED FROM PYTHON SOURCE LINES 70-71

Set the current workspace:

.. GENERATED FROM PYTHON SOURCE LINES 71-76

.. code-block:: Python


    simai_client.set_current_workspace(WORKSPACE_NAME)
    current_workspace = simai_client.current_workspace
    print(f"Using workspace: {current_workspace.name}")


.. GENERATED FROM PYTHON SOURCE LINES 77-78

A workspace contains a trained model and is created when a model build completes successfully.

.. GENERATED FROM PYTHON SOURCE LINES 80-83

Prepare data structures
----------------------------------------------------
Create lists to store geometries and their boundary conditions:

.. GENERATED FROM PYTHON SOURCE LINES 83-87

.. code-block:: Python


    geometries: list[Geometry] = []
    boundary_conditions: list[dict] = []


.. GENERATED FROM PYTHON SOURCE LINES 88-91

Upload geometries and load boundary conditions
----------------------------------------------------
Process each subdirectory in the dataset:

.. GENERATED FROM PYTHON SOURCE LINES 91-132

.. code-block:: Python


    for dir in os.listdir(f"{DATASET_PATH}"):
        print(f"Processing directory: {dir}")

        # Define the geometry file path
        # The file is uploaded with a specific name for easy identification
        file = (
            f"{DATASET_PATH}/{dir}/surface.vtp",  # Local file path
            f"{dir}.vtp",  # Name to use in SimAI
        )

        # Check if this geometry already exists in the workspace
        geoms = simai_client.geometries.list(
            workspace=current_workspace,
        )

        if f"{dir}.vtp" in [geom.name for geom in geoms]:
            print(f"Geometry {dir}.vtp already uploaded, skipping upload.")
            # Find and use the existing geometry
            existing_geom = [geom for geom in geoms if geom.name == f"{dir}.vtp"][0]
            geometries.append(existing_geom)
        else:
            print(f"Uploading geometry {dir}.vtp...")
            geom = simai_client.geometries.upload(file, workspace=current_workspace)
            geometries.append(geom)

        ###########################################################################
        # Load boundary conditions (if applicable)
        # ----------------------------------------------------
        # This section is optional and only needed if your model requires boundary conditions.
        # If your model does not use boundary conditions, you can skip this section.

        try:
            with open(f"{DATASET_PATH}/{dir}/boundary_conditions.json", "r") as f:
                print(f"Loading boundary conditions for {dir}...")
                boundary_conditions_for_geom = json.load(f)
                boundary_conditions.append(boundary_conditions_for_geom)
        except FileNotFoundError:
            print(f"No boundary conditions file found for {dir}, using empty boundary conditions.")
            boundary_conditions.append({})


.. GENERATED FROM PYTHON SOURCE LINES 133-136

Run predictions on all geometries
----------------------------------------------------
Wait for all geometries to be processed, then run predictions:

.. GENERATED FROM PYTHON SOURCE LINES 136-156

.. code-block:: Python


    preds: list[Prediction] = []

    print("\nWaiting for geometries to be processed...")
    for geom in geometries:
        print(f"Waiting for geometry {geom.name}...")
        geom.wait()
        print(f"Geometry {geom.name} is processed.")

        # Run prediction with boundary conditions if available
        if boundary_conditions[geometries.index(geom)]:
            preds.append(
                geom.run_prediction(boundary_conditions=boundary_conditions[geometries.index(geom)])
            )
            print(f"Prediction started for geometry {geom.name} with boundary conditions.")
        else:
            # Run prediction without boundary conditions
            preds.append(geom.run_prediction())
            print(f"Prediction started for geometry {geom.name}.")


.. GENERATED FROM PYTHON SOURCE LINES 157-160

Collect prediction results
----------------------------------------------------
Wait for all predictions to complete and extract results:

.. GENERATED FROM PYTHON SOURCE LINES 160-174

.. code-block:: Python


    results = {}

    print("\nWaiting for predictions to complete...")
    for pred in preds:
        geom_name = geometries[preds.index(pred)].name
        print(f"Waiting for prediction on geometry {geom_name}...")
        pred.wait()
        print(f"Prediction completed for geometry {geom_name}.")

        # Extract global coefficients and confidence score
        results[geom_name] = pred.post.global_coefficients().data
        results[geom_name]["confidence_score"] = pred.raw_confidence_score


.. GENERATED FROM PYTHON SOURCE LINES 175-178

Save results to file
----------------------------------------------------
Export all prediction results to a JSON file for further analysis:

.. GENERATED FROM PYTHON SOURCE LINES 178-185

.. code-block:: Python


    print("\nSaving results to results.json...")
    with open("results.json", "w") as f:
        json.dump(results, f, indent=4)

    print("Results saved successfully!")


.. GENERATED FROM PYTHON SOURCE LINES 186-197

Understanding the results
----------------------------------------------------
The results dictionary contains:

- Global coefficients: Scalar values computed from the prediction,
- Confidence score: A measure of how familiar this prediction is for the model.

You can also download full field results (VTP files) for visualization:

- Use ``pred.post.surface_vtp()`` to get surface fields.
- Use ``pred.post.volume_vtu()`` to get volume fields.

.. GENERATED FROM PYTHON SOURCE LINES 199-202

Example: Download surface VTP for visualization
----------------------------------------------------
Run the following code to download the first prediction's surface VTP:

.. GENERATED FROM PYTHON SOURCE LINES 202-210

.. code-block:: Python


    if preds:
        first_pred = preds[0]
        first_geom_name = geometries[0].name
        output_path = f"{first_geom_name}_result.vtp"
        first_pred.post.surface_vtp().data.download(output_path)
        print(f"Downloaded surface VTP to {output_path}")


.. GENERATED FROM PYTHON SOURCE LINES 211-219

Next steps
----------------------------------------------------
With your prediction results, you can:

- Analyze global coefficients to evaluate design performance.
- Visualize field results in ParaView or similar tools.
- Compare predictions against validation data.
- Use results to guide design optimization.


.. _sphx_glr_download__examples_00_basic_simai_ex_02-run_predictions.py:

.. only:: html

  .. container:: sphx-glr-footer sphx-glr-footer-example

    .. container:: sphx-glr-download sphx-glr-download-jupyter

      :download:`Download Jupyter notebook: 02-run_predictions.ipynb <02-run_predictions.ipynb>`

    .. container:: sphx-glr-download sphx-glr-download-python

      :download:`Download Python source code: 02-run_predictions.py <02-run_predictions.py>`

    .. container:: sphx-glr-download sphx-glr-download-zip

      :download:`Download zipped: 02-run_predictions.zip <02-run_predictions.zip>`


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_