compgen 0.1.0__tar.gz

compgen-0.1.0/LICENSE

@@ -0,0 +1,23 @@
+Copyright 2026, UChicago Argonne, LLC
+All Rights Reserved
+Software Name: compgen
+By: Argonne National Laboratory
+OPEN SOURCE LICENSE (MIT)
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in
+all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+THE SOFTWARE.

compgen-0.1.0/PKG-INFO

@@ -0,0 +1,105 @@
+Metadata-Version: 2.4
+Name: compgen
+Version: 0.1.0
+Summary: Python toolkit for generating simulation data for two phase composite materials.
+Author: Mark Messner, Tianchen Hu, Fabio Sozio
+License-Expression: MIT
+Project-URL: Homepage, https://github.com/applied-material-modeling/compgen
+Project-URL: Repository, https://github.com/applied-material-modeling/compgen
+Project-URL: Issues, https://github.com/applied-material-modeling/compgen/issues
+Keywords: composite materials,microstructure,finite element,homogenization,MOOSE,dataset generation
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Science/Research
+Classifier: Operating System :: POSIX
+Classifier: Operating System :: MacOS
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Topic :: Scientific/Engineering
+Classifier: Topic :: Scientific/Engineering :: Physics
+Requires-Python: >=3.13
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: numpy
+Requires-Dist: scipy
+Requires-Dist: h5py
+Requires-Dist: scikit-image
+Requires-Dist: gstools
+Requires-Dist: multiprocess
+Requires-Dist: pyvista
+Requires-Dist: exodusii==26.10.29
+Provides-Extra: mcrpy
+Requires-Dist: mcrpy; extra == "mcrpy"
+Provides-Extra: demos
+Requires-Dist: matplotlib; extra == "demos"
+Provides-Extra: dev
+Requires-Dist: black==26.3.0; extra == "dev"
+Requires-Dist: pytest==9.0.2; extra == "dev"
+Requires-Dist: pylint==4.0.5; extra == "dev"
+Requires-Dist: pre-commit==4.7.0; extra == "dev"
+Dynamic: license-file
+
+# compgen
+
+## Summary
+compgen is a Python toolkit for generating simulation data for two phase composite materials.  It will generate structures, invoke physics codes to simulate the response for a variety of material properties, and write summaries of the results.
+
+## Install
+compgen requires Python 3.13 or newer.
+
+1. (Optional) Create and activate a virtual environment.
+2. Install the package:
+   - `pip install compgen`
+   - To use the reconstruction generator, `pip install compgen[mcrpy]`.  mcrpy depends on TensorFlow, which currently supports Python up to 3.13, so use Python 3.13 for this option.
+   - To develop compgen, or to run the tests and demos, clone this repository and run `pip install -e .[dev,demos]` from its root.
+3. Install the mesher and physics codes you want to use:
+    1. [MOOSE](https://mooseframework.inl.gov/)
+    2. [SCULPT](https://www.sandia.gov/files/cubit/documents/SculptSANDReport.pdf)
+
+## Inputs
+
+compgen pipes together several steps to generate results:
+1. Generate a composite structure -- the distribution of two materials A and B over a periodic domain.  We use pixelated fixed grids to define this distribution.
+2. Make an unstructured finite element mesh of the fixed grid representation, smoothing interfaces.  This might involve first filtering the fixed grid representation, then calling a mesh generator.
+3. Setup input files for the physics codes describing the problem physics and boundary conditions.
+4. Actually run the physics code.  This might require multiple calls, for example if we're interested in properties in different loading directions.
+5. Post process the output from the physics code into a common format.  We use an [HDF5 format](https://github.com/applied-material-modeling/compgen/blob/main/doc/data.md).
+
+compgen provides a simple API (`runner.run`) for doing all of this based on json input files.  You need two input files to invoke the runner:
+1. A problem file defining the steps outlined above -- which structure generator, mesh, physics, solver, etc.
+2. A system file letting compgen know how to run external tools (meshers and physics codes).
+
+There are examples of each file type to make up the 2D and 3D demos, described below.
+
+## Tests and demos
+
+First set environment variables:
+1. `MOOSE_DIR` to the root directory of your moose installation
+2. `CUBIT_DIR` to the root directory of your CUBIT directory, if you want to run the demos using sculpt.  This is not necessary for the verification tests.
+
+## Run the tests
+
+Run the tests with `pytest`.  The general tests in the `tests` directory do not require MOOSE and are quick and cheap to run.  The verification tests in `tests/verification` are more expensive and require MOOSE.
+
+## Run the demos
+
+There are complete demos that generate random structures, call MOOSE to solve for some material properties, and store the results.  These demos live in `demos/complete`.  The subdirectories first split
+the problems into different physics/types of material properties and then into 2D and 3D examples.
+
+## License headers
+
+Every source file (`*.py` and the MOOSE `*.i` templates) must start with the contents of `LICENSE` as a comment header.  CI enforces this.
+
+- `python scripts/add_license.py` adds the header to any tracked source file that is missing it (or pass specific files as arguments).
+- `python scripts/check_license.py` reports files missing the header and exits non-zero if there are any.
+
+## Pre-commit hooks
+
+The license check, black, and pylint run as [pre-commit](https://pre-commit.com) hooks on the files staged for each commit.  To enable them, install the dev dependencies and then install the hooks once per clone:
+
+```
+pip install -e .[dev]
+pre-commit install
+```
+
+To run the hooks by hand against every file, use `pre-commit run --all-files`.

compgen-0.1.0/README.md

@@ -0,0 +1,64 @@
+# compgen
+
+## Summary
+compgen is a Python toolkit for generating simulation data for two phase composite materials.  It will generate structures, invoke physics codes to simulate the response for a variety of material properties, and write summaries of the results.
+
+## Install
+compgen requires Python 3.13 or newer.
+
+1. (Optional) Create and activate a virtual environment.
+2. Install the package:
+   - `pip install compgen`
+   - To use the reconstruction generator, `pip install compgen[mcrpy]`.  mcrpy depends on TensorFlow, which currently supports Python up to 3.13, so use Python 3.13 for this option.
+   - To develop compgen, or to run the tests and demos, clone this repository and run `pip install -e .[dev,demos]` from its root.
+3. Install the mesher and physics codes you want to use:
+    1. [MOOSE](https://mooseframework.inl.gov/)
+    2. [SCULPT](https://www.sandia.gov/files/cubit/documents/SculptSANDReport.pdf)
+
+## Inputs
+
+compgen pipes together several steps to generate results:
+1. Generate a composite structure -- the distribution of two materials A and B over a periodic domain.  We use pixelated fixed grids to define this distribution.
+2. Make an unstructured finite element mesh of the fixed grid representation, smoothing interfaces.  This might involve first filtering the fixed grid representation, then calling a mesh generator.
+3. Setup input files for the physics codes describing the problem physics and boundary conditions.
+4. Actually run the physics code.  This might require multiple calls, for example if we're interested in properties in different loading directions.
+5. Post process the output from the physics code into a common format.  We use an [HDF5 format](https://github.com/applied-material-modeling/compgen/blob/main/doc/data.md).
+
+compgen provides a simple API (`runner.run`) for doing all of this based on json input files.  You need two input files to invoke the runner:
+1. A problem file defining the steps outlined above -- which structure generator, mesh, physics, solver, etc.
+2. A system file letting compgen know how to run external tools (meshers and physics codes).
+
+There are examples of each file type to make up the 2D and 3D demos, described below.
+
+## Tests and demos
+
+First set environment variables:
+1. `MOOSE_DIR` to the root directory of your moose installation
+2. `CUBIT_DIR` to the root directory of your CUBIT directory, if you want to run the demos using sculpt.  This is not necessary for the verification tests.
+
+## Run the tests
+
+Run the tests with `pytest`.  The general tests in the `tests` directory do not require MOOSE and are quick and cheap to run.  The verification tests in `tests/verification` are more expensive and require MOOSE.
+
+## Run the demos
+
+There are complete demos that generate random structures, call MOOSE to solve for some material properties, and store the results.  These demos live in `demos/complete`.  The subdirectories first split
+the problems into different physics/types of material properties and then into 2D and 3D examples.
+
+## License headers
+
+Every source file (`*.py` and the MOOSE `*.i` templates) must start with the contents of `LICENSE` as a comment header.  CI enforces this.
+
+- `python scripts/add_license.py` adds the header to any tracked source file that is missing it (or pass specific files as arguments).
+- `python scripts/check_license.py` reports files missing the header and exits non-zero if there are any.
+
+## Pre-commit hooks
+
+The license check, black, and pylint run as [pre-commit](https://pre-commit.com) hooks on the files staged for each commit.  To enable them, install the dev dependencies and then install the hooks once per clone:
+
+```
+pip install -e .[dev]
+pre-commit install
+```
+
+To run the hooks by hand against every file, use `pre-commit run --all-files`.

compgen-0.1.0/compgen/__init__.py

@@ -0,0 +1,29 @@
+# Copyright 2026, UChicago Argonne, LLC
+# All Rights Reserved
+# Software Name: compgen
+# By: Argonne National Laboratory
+# OPEN SOURCE LICENSE (MIT)
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+# THE SOFTWARE.
+
+"""Top-level package for composite structure generation and simulation workflows."""
+
+from importlib.metadata import version
+
+__version__ = version("compgen")

compgen-0.1.0/compgen/exoutil.py

@@ -0,0 +1,43 @@
+# Copyright 2026, UChicago Argonne, LLC
+# All Rights Reserved
+# Software Name: compgen
+# By: Argonne National Laboratory
+# OPEN SOURCE LICENSE (MIT)
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+# THE SOFTWARE.
+
+"""Various helpers for reading Exodus II files."""
+
+import numpy as np
+
+
+def get_element_field(exo, field):
+    """Retrieve an element field from an Exodus mesh.
+
+    Args:
+        exo (exodusii.File): Open Exodus file object.
+        field (str): Name of the element field to retrieve.
+    """
+    var_names = list(exo.get_element_variable_names())
+    if field not in var_names:
+        raise ValueError(f"Field '{field}' not found in output Exodus file.")
+
+    return np.concatenate(
+        [exo.get_element_variable_values(blk, field) for blk in [1, 2]], axis=-1
+    )

compgen-0.1.0/compgen/filters.py

@@ -0,0 +1,120 @@
+# Copyright 2026, UChicago Argonne, LLC
+# All Rights Reserved
+# Software Name: compgen
+# By: Argonne National Laboratory
+# OPEN SOURCE LICENSE (MIT)
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+# THE SOFTWARE.
+
+"""Image filtering utilities for preprocessing generated microstructures."""
+
+import numpy as np
+import skimage
+from compgen import utils
+
+
+def mean_filter(img, radius=3, periodic=True):
+    """Apply a mean filter to an image.
+
+    Args:
+        img (np.ndarray): 2D or 3D image.
+        radius (int): Radius of disk/ball footprint.
+        periodic (bool): If True, use periodic wrapping at the boundaries.
+
+    Returns:
+        np.ndarray: Smoothed binary image.
+    """
+    ndim = img.ndim
+
+    if ndim == 3:
+        footprint = skimage.morphology.ball(radius)
+    elif ndim == 2:
+        footprint = skimage.morphology.disk(radius)
+    else:
+        raise ValueError("mean_filter only supports 2D and 3D images")
+
+    def smooth(i):
+        i_u8 = skimage.util.img_as_ubyte(i.astype(float))
+
+        if periodic:
+            pad_width = [(radius, radius)] * ndim
+            i_u8 = np.pad(i_u8, pad_width=pad_width, mode="wrap")
+
+        out = skimage.filters.rank.mean(i_u8, footprint=footprint)
+        out = skimage.util.img_as_float(out)
+
+        if periodic:
+            slices = tuple(slice(radius, -radius) for _ in range(ndim))
+            out = out[slices]
+
+        return out
+
+    return utils.binary_smoother(img, smooth)
+
+
+def make_orthogonal(img):
+    """Mirror and tile the input array to give it orthogonal periodicity.
+
+    This works for arrays of arbitrary dimensionality by concatenating the
+    array with its flipped version along each axis (processing axes from
+    last to first).
+    """
+    arr = np.asarray(img)
+    for axis in reversed(range(arr.ndim)):
+        arr = np.concatenate([arr, np.flip(arr, axis=axis)], axis=axis)
+    return arr
+
+
+def make_periodic(img: np.ndarray, margin: int = 2) -> np.ndarray:
+    """
+    Make periodic for Sculpt/MOOSE workflow.
+
+    Overwrite the first and last `margin` layers of each axis with the layer at
+    index `margin`, for 2D or 3D arrays.
+
+    A value `margin` = 2 seems to work consistently with Sculpt/MOOSE.
+    """
+    arr = np.array(img, copy=True)
+
+    if arr.ndim not in (2, 3):
+        raise RuntimeError(f"Expected 2D or 3D array, got {arr.ndim}D")
+
+    if not isinstance(margin, int) or margin < 1:
+        raise ValueError(f"margin must be a positive integer, got {margin}")
+
+    if any(2 * margin >= s for s in arr.shape):
+        raise ValueError(f"margin={margin} is too large for shape {arr.shape}")
+
+    # Loop over each axis
+    for ax in range(arr.ndim):
+        # Slice selecting the master layer at index = margin
+        master = [slice(None)] * arr.ndim
+        master[ax] = slice(margin, margin + 1)
+
+        # Slice for lower and upper boundary block
+        lo = [slice(None)] * arr.ndim
+        lo[ax] = slice(0, margin)
+        hi = [slice(None)] * arr.ndim
+        hi[ax] = slice(-margin, None)
+
+        # Overwrite both boundary blocks using the master layer
+        arr[tuple(lo)] = arr[tuple(master)]
+        arr[tuple(hi)] = arr[tuple(master)]
+
+    return arr