Conditional random fields by bramvandeneijnden · Pull Request #3 · StemVibrations/RandomFields

bramvandeneijnden · 2024-06-28T15:25:48Z

extension of random field generator with random field conditioning.

Uses sklear as additional requirement for efficient kriging, which is not available efficiently enough in gstools.

new file: random_fields/generate_conditioned_field.py

modified: random_fields/generate_conditioned_field.py

Changes to be committed: modified: ../README.md new file: Makefile renamed: static/random_field.png -> _static/random_field.png new file: _static/random_field_1.png new file: _static/random_field_2.png new file: _static/random_field_3.png renamed: static/random_field_3D.png -> _static/random_field_3D.png new file: authors.rst new file: conf.py new file: contributions.rst new file: index.rst new file: installation.rst new file: make.bat new file: requirements.txt new file: theory.rst new file: tutorials.rst modified: ../random_fields/generate_conditioned_field.py modified: ../random_fields/generate_field.py modified: ../random_fields/utils.py

Changes committed: new file: _static/conditioned_random_field_3D.png new file: _static/kriging_mean_2D.png new file: _static/kriging_mean_3D.png new file: _static/kriging_std_2D.png new file: _static/kriging_std_3D.png new file: _static/random_field_4.png new file: _static/random_field_5.png modified: index.rst modified: tutorials.rst modified: ../random_fields/utils.py

Changes to be committed: modified: tutorials.rst modified: ../requirements.txt

aronnoordam

few comments about style and some stuff what is unclear for me.

Furthermore can you add tests for the added functionalities?

aronnoordam · 2024-07-03T13:19:48Z

+    def create_solver_settings_dictionary(self, model: Model, mesh_file_name: str, materials_file_name: str) -> Dict[str, Any]:
+        """
+        Creates a dictionary containing the solver settings.
+
+        Args:
+            - model (:class:`random_fields.model.Model`): The model object containing the solver data and model parts.
+            - mesh_file_name (str): The name of the mesh file.
+            - materials_file_name (str): The name of the materials parameters json file.
+
+        Raises:
+            - ValueError: if solver_settings in model are not initialised.
+
+        Returns:
+            - Dict[str, Any]: dictionary containing the part of the project parameters
+                dictionary related to problem data and solver settings.
+        """
+
+        if model.project_parameters is None:
+            raise ValueError("Solver settings are not initialised in model.")
+
+        return self.solver_io.create_settings_dictionary(
+            model.project_parameters,
+            Path(mesh_file_name).stem,
+            materials_file_name,
+            model.get_all_model_parts(),


this function and the class above do not exists within the RF repos, can you replace it with an existing function?

updated with existing function

aronnoordam · 2024-07-03T13:22:44Z

 numpy==1.24.2
 gstools==1.5.0
 matplotlib>=3.7.0
+sklearn


missing version, also sklearn is deprecated, i think it should be scikit-learn

also the dependency should be added to requirements_dev.txt

scikit-learn corrected ==1.5.1

I tried to set matplotlib version to the one used in the STEM package (3.7.4), but this breaks one of the tests on the plotting utils...

aronnoordam · 2024-07-12T09:50:28Z

+        self.kriging_std = None
+
+
+    def generate_conditioned(self, nodes: np.ndarray) -> None:


Suggested change

def generate_conditioned(self, nodes: np.ndarray) -> None:

def generate_conditioned(self, nodes: npt.NDArray[np.float64]) -> None:

corrected, but hot to properly document arrays in the docstrings? see my attempt in the docstrings

aronnoordam · 2024-07-12T09:56:00Z

+        self.kriging_mean = self.mean + np.sqrt(self.variance) * z_kriged_field
+        #
+        # create single random field at nodes and conditioning points
+        nodes_cpoints = np.vstack([nodes,self.conditioning_points])


can you write out the variable names a bit more, e.g. nodes_cond_points. Also for example z_cond_rf_nodes instead of z_crf_nodes. This makes the code a bit more understandable

Co-authored-by: aronnoordam <51492202+aronnoordam@users.noreply.github.com>

modified: docs/contributions.rst modified: docs/installation.rst modified: random_fields/generate_field.py modified: random_fields/utils.py modified: requirements_dev.txt new file: tests/data/kriging_mean_3D.txt new file: tests/data/kriging_std_3D.txt modified: tests/test_random_field.py

- style changes - update docstrings - extended test modified: docs/contributions.rst modified: random_fields/generate_field.py modified: random_fields/utils.py modified: requirements.txt modified: requirements_dev.txt new file: tests/data/conditioned_rf_3D.txt deleted: tests/data/kriging_mean_3D.txt deleted: tests/data/kriging_std_3D.txt modified: tests/test_random_field.py

modified: requirements_dev.txt

bramvandeneijnden · 2024-08-23T10:13:20Z

+        - z_kriged_field (array): standard-normal kriged random field
+        - gaussian_process (:class:sklearn.gaussian_process.GaussianProcessRegressor)
+        - conditioning_points (array): coordinates of the conditioning points, shape (:,`self.n_dims`)
+        - conditioning_values (array): values of the conditioning points, shape (:,)


How to document arrays in the doc strings? I've seen "Sequence(float)" here and there...

and what about shapes and defaults?

bramvandeneijnden added 8 commits June 13, 2024 17:07

added first version of conditioned random field generator

8015988

new file: random_fields/generate_conditioned_field.py

corrections, clean-up, efficiency and commenting

dfe8826

modified: random_fields/generate_conditioned_field.py

added sklearn to the requirements

6b54e89

Changes to be committed: modified: tutorials.rst modified: ../requirements.txt

deleted: random_fields/generate_conditioned_field.py

78643d3

modified: setup.cfg

5fadc51

modified: setup.cfg (sklearn --> scikit-learn)

04170dd

bramvandeneijnden requested review from anoordam and aronnoordam June 28, 2024 15:25

bramvandeneijnden assigned aronnoordam and anoordam Jun 28, 2024

aronnoordam reviewed Jul 12, 2024

View reviewed changes

bramvandeneijnden and others added 3 commits August 15, 2024 16:32

Update docs/installation.rst

3b9df07

Co-authored-by: aronnoordam <51492202+aronnoordam@users.noreply.github.com>

Update random_fields/generate_field.py

72e157b

Co-authored-by: aronnoordam <51492202+aronnoordam@users.noreply.github.com>

bramvandeneijnden requested a review from brunozc as a code owner August 20, 2024 07:48

aronnoordam reviewed Aug 22, 2024

View reviewed changes

Comment thread random_fields/generate_field.py Outdated

Comment thread tests/test_random_field.py Outdated

Comment thread tests/test_random_field.py Outdated

bramvandeneijnden added 2 commits August 23, 2024 11:15

modified: requirements.txt

f76e0f0

modified: requirements_dev.txt

bramvandeneijnden commented Aug 23, 2024

View reviewed changes

		self.kriging_std = None


		def generate_conditioned(self, nodes: np.ndarray) -> None:

Uh oh!

Conversation

bramvandeneijnden commented Jun 28, 2024

Uh oh!

aronnoordam left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

bramvandeneijnden Aug 23, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

bramvandeneijnden Aug 23, 2024 •

edited

Loading