Add TetraMeshData class

[munechika-koyo]

Proposal of new feature: TetraMeshData

TL;DR: Reading files in the created tree structure will greatly enhance the speed of the process with this PR.

Description

I would propose a new feature, the TetraMeshData class, modeled after the existing MeshData definition.
(I have already suggested #407 but would like to split the function.)

It was implemented with the following objectives:

• Saving/Loading from the file formatted .rsm, which has constructed a K-D tree structure.
• Offering useful methods like calculating a tetrahedral volume and barycenter.

Additionally, this class will be used when loading mesh files formatted as .vtk or .obj, which contain tetrahedral (3-D unstructured) mesh data in the future.

Examples of the use:

1. Creating the instance of TetraMeshData

>>>fromraysect.primitive.mesh.tetra_meshimportTetraMeshData>>>verts= [[0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 1]]
>>>tets= [[0, 1, 2, 3]]
>>>tetra=TetraMeshData(verts, tets)

2. Calculating the volume of a tetrahedron

>>>tetra.volume(0)
0.166666# == 1 x 1 x 0.5 / 3.0>>>tetra.volume_total()
0.166666

3. Calculating the barycenter of a tetrahedron

>>>tetra.barycenter(0)
Point3D(0.25, 0.25, 0.25)

4. Saving as .rsm

>>>tetra.save("tetra.rsm")

5. Loading from a .rsm file

>>>TetraMeshData.from_file("tetra.rsm")
<raysect.primitive.mesh.tetra_mesh.TetraMeshDataat0x1336e7440>

Speed test

I attempted to compare the instancing of Discrete3DMesh with TetraMeshData loaded from a file.
As a mesh file, I used the stanford_bunny.mesh file located in demos/resources/.

	elapse time (sec)
Create Discrete3DMesh	15.441215
Create TetraMeshData	15.894989
Load TetraMeshData from file	01.766619

I attained a loading speed for tetra mesh that is approximately 9 times faster than generating TetraMeshData using vertices and tetrahedral index arrays.
This difference is expected to widen as the vertex data size increases.
The test script that I used is in the collapsed section below.

Test script

fromdatetimeimporttimedeltafrompathlibimportPathfromtimeitimporttimeitimportnumpyasnpfromraysect.core.math.function.float.function3d.interpolateimportDiscrete3DMeshfromraysect.primitive.mesh.tetra_meshimportTetraMeshDatadefload_mesh(mesh_filepath):
vertices= []
tetrahedra= []
withopen(mesh_filepath, "r") asf:
lines=f.readlines()
i=0whilei<len(lines):
line=lines[i].strip()
ifnotlineorline.startswith("#"):
i+=1continue# Check section headers (case-sensitive)ifline=="Vertices":
i+=1vertex_count=int(lines[i].strip())
i+=1for_inrange(vertex_count):
parts=lines[i].strip().split()
# take only the first 3 coordinatesvertices.append([float(v) forvinparts[:3]])
i+=1continueelifline=="Tetrahedra":
i+=1tet_count=int(lines[i].strip())
i+=1for_inrange(tet_count):
parts=lines[i].strip().split()
# take only the first 4 indices, converting them to int.tetrahedra.append([int(idx) foridxinparts[:4]])
i+=1continuei+=1vertices=np.array(vertices)
tetrahedra=np.array(tetrahedra, dtype=np.int32) -1# 0-based indexingreturnvertices, tetrahedradefcreate_discrete3dmesh(vertices, tetrahedra):
returnDiscrete3DMesh(
vertices, tetrahedra, np.ones((tetrahedra.shape[0])), False, 0
)
defcreate_tetrameshdata(vertices, tetrahedra):
returnTetraMeshData(vertices, tetrahedra)
defload_tetrameshdata(mesh_filepath):
returnTetraMeshData.from_file(mesh_filepath)
if__name__=="__main__":
ROOT=Path(__file__).parentmesh_file=ROOT/"demos"/"resources"/"stanford_bunny.mesh"vertices, tetrahedra=load_mesh(mesh_file)
# Save tetramesh data in advancetetra=TetraMeshData(vertices, tetrahedra)
tetra_file=ROOT/"temp_tetra.rsm"tetra.save(tetra_file)
# === Measure time ===loop=5# create Discrete3DMeshresult=timeit(
"create_discrete3dmesh(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create Discrete3DMesh: {elapsed_time}")
# create TetraMeshDataresult=timeit(
"create_tetrameshdata(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create TetraMeshData: {elapsed_time}")
# load TetraMeshData from fileresult=timeit("load_tetrameshdata(tetra_file)", globals=globals(), number=loop)
elapsed_time=timedelta(seconds=result/loop)
print(f"Load TetraMeshData from file: {elapsed_time}")

Class structure

The structure of the TetraMeshData I implemented is as follows.

classDiagram
class KDTree3DCore {
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
}
class TetraMeshData {
+__init__(object vertices, object tetrahedra, bint tolerant=True)
+__getstate__()
+__setstate__(state)
+__reduce__()
+vertices
+tetrahedra
+Point3D vertex(int index)
+ndarray tetrahedron(int index)
+Point3D barycenter(int index)
+double volume(int index)
+double volume_total()
+BoundingBox3D bounding_box(AffineMatrix3D to_world)
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
+classmethod from_file(cls, file)
}
KDTree3DCore <|-- TetraMeshData

Loading

Details of unit test

I also implemented a unit test for TestMeshData in raysect/primitive/mesh/tests/test_tetra_mesh.py.
Below is a table showing the correspondence between the methods of the TestTetraMeshData class in unit tests and the methods tested in the TestMeshData class.

Test Method Name in TestTetraMeshData	Tested Method/Function in TetraMeshData
test_initialization()	__init__()
test_invalid_tetrahedron_indices()	__init__()
test_vertex_method()	vertex()
test_invalid_vertex_index()	vertex()
test_barycenter()	barycenter()
test_compute_volume()	volume(), volume_total()
test_is_contained()	is_contained()
test_bounding_box()	bounding_box()
test_pickle_state ()	__getstate__(), __setstate__(), save(), load()

---

I would appreciate it if you would review it and make any suggestions and comments.

[munechika-koyo]

Proposal of new feature: TetraMeshData

TL;DR: Reading files in the created tree structure will greatly enhance the speed of the process with this PR.

Description

I would propose a new feature, the TetraMeshData class, modeled after the existing MeshData definition.
(I have already suggested #407 but would like to split the function.)

It was implemented with the following objectives:

• Saving/Loading from the file formatted .rsm, which has constructed a K-D tree structure.
• Offering useful methods like calculating a tetrahedral volume and barycenter.

Additionally, this class will be used when loading mesh files formatted as .vtk or .obj, which contain tetrahedral (3-D unstructured) mesh data in the future.

Examples of the use:

1. Creating the instance of TetraMeshData

>>>fromraysect.primitive.mesh.tetra_meshimportTetraMeshData>>>verts= [[0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 1]]
>>>tets= [[0, 1, 2, 3]]
>>>tetra=TetraMeshData(verts, tets)

2. Calculating the volume of a tetrahedron

>>>tetra.volume(0)
0.166666# == 1 x 1 x 0.5 / 3.0>>>tetra.volume_total()
0.166666

3. Calculating the barycenter of a tetrahedron

>>>tetra.barycenter(0)
Point3D(0.25, 0.25, 0.25)

4. Saving as .rsm

>>>tetra.save("tetra.rsm")

5. Loading from a .rsm file

>>>TetraMeshData.from_file("tetra.rsm")
<raysect.primitive.mesh.tetra_mesh.TetraMeshDataat0x1336e7440>

Speed test

I attempted to compare the instancing of Discrete3DMesh with TetraMeshData loaded from a file.
As a mesh file, I used the stanford_bunny.mesh file located in demos/resources/.

	elapse time (sec)
Create Discrete3DMesh	15.441215
Create TetraMeshData	15.894989
Load TetraMeshData from file	01.766619

I attained a loading speed for tetra mesh that is approximately 9 times faster than generating TetraMeshData using vertices and tetrahedral index arrays.
This difference is expected to widen as the vertex data size increases.
The test script that I used is in the collapsed section below.

Test script

fromdatetimeimporttimedeltafrompathlibimportPathfromtimeitimporttimeitimportnumpyasnpfromraysect.core.math.function.float.function3d.interpolateimportDiscrete3DMeshfromraysect.primitive.mesh.tetra_meshimportTetraMeshDatadefload_mesh(mesh_filepath):
vertices= []
tetrahedra= []
withopen(mesh_filepath, "r") asf:
lines=f.readlines()
i=0whilei<len(lines):
line=lines[i].strip()
ifnotlineorline.startswith("#"):
i+=1continue# Check section headers (case-sensitive)ifline=="Vertices":
i+=1vertex_count=int(lines[i].strip())
i+=1for_inrange(vertex_count):
parts=lines[i].strip().split()
# take only the first 3 coordinatesvertices.append([float(v) forvinparts[:3]])
i+=1continueelifline=="Tetrahedra":
i+=1tet_count=int(lines[i].strip())
i+=1for_inrange(tet_count):
parts=lines[i].strip().split()
# take only the first 4 indices, converting them to int.tetrahedra.append([int(idx) foridxinparts[:4]])
i+=1continuei+=1vertices=np.array(vertices)
tetrahedra=np.array(tetrahedra, dtype=np.int32) -1# 0-based indexingreturnvertices, tetrahedradefcreate_discrete3dmesh(vertices, tetrahedra):
returnDiscrete3DMesh(
vertices, tetrahedra, np.ones((tetrahedra.shape[0])), False, 0
)
defcreate_tetrameshdata(vertices, tetrahedra):
returnTetraMeshData(vertices, tetrahedra)
defload_tetrameshdata(mesh_filepath):
returnTetraMeshData.from_file(mesh_filepath)
if__name__=="__main__":
ROOT=Path(__file__).parentmesh_file=ROOT/"demos"/"resources"/"stanford_bunny.mesh"vertices, tetrahedra=load_mesh(mesh_file)
# Save tetramesh data in advancetetra=TetraMeshData(vertices, tetrahedra)
tetra_file=ROOT/"temp_tetra.rsm"tetra.save(tetra_file)
# === Measure time ===loop=5# create Discrete3DMeshresult=timeit(
"create_discrete3dmesh(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create Discrete3DMesh: {elapsed_time}")
# create TetraMeshDataresult=timeit(
"create_tetrameshdata(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create TetraMeshData: {elapsed_time}")
# load TetraMeshData from fileresult=timeit("load_tetrameshdata(tetra_file)", globals=globals(), number=loop)
elapsed_time=timedelta(seconds=result/loop)
print(f"Load TetraMeshData from file: {elapsed_time}")

Class structure

The structure of the TetraMeshData I implemented is as follows.

classDiagram
class KDTree3DCore {
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
}
class TetraMeshData {
+__init__(object vertices, object tetrahedra, bint tolerant=True)
+__getstate__()
+__setstate__(state)
+__reduce__()
+vertices
+tetrahedra
+Point3D vertex(int index)
+ndarray tetrahedron(int index)
+Point3D barycenter(int index)
+double volume(int index)
+double volume_total()
+BoundingBox3D bounding_box(AffineMatrix3D to_world)
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
+classmethod from_file(cls, file)
}
KDTree3DCore <|-- TetraMeshData

Loading

Details of unit test

I also implemented a unit test for TestMeshData in raysect/primitive/mesh/tests/test_tetra_mesh.py.
Below is a table showing the correspondence between the methods of the TestTetraMeshData class in unit tests and the methods tested in the TestMeshData class.

Test Method Name in TestTetraMeshData	Tested Method/Function in TetraMeshData
test_initialization()	__init__()
test_invalid_tetrahedron_indices()	__init__()
test_vertex_method()	vertex()
test_invalid_vertex_index()	vertex()
test_barycenter()	barycenter()
test_compute_volume()	volume(), volume_total()
test_is_contained()	is_contained()
test_bounding_box()	bounding_box()
test_pickle_state ()	__getstate__(), __setstate__(), save(), load()

---

I would appreciate it if you would review it and make any suggestions and comments.

[munechika-koyo]

Proposal of new feature: TetraMeshData

TL;DR: Reading files in the created tree structure will greatly enhance the speed of the process with this PR.

Description

I would propose a new feature, the TetraMeshData class, modeled after the existing MeshData definition.
(I have already suggested #407 but would like to split the function.)

It was implemented with the following objectives:

• Saving/Loading from the file formatted .rsm, which has constructed a K-D tree structure.
• Offering useful methods like calculating a tetrahedral volume and barycenter.

Additionally, this class will be used when loading mesh files formatted as .vtk or .obj, which contain tetrahedral (3-D unstructured) mesh data in the future.

Examples of the use:

1. Creating the instance of TetraMeshData

>>>fromraysect.primitive.mesh.tetra_meshimportTetraMeshData>>>verts= [[0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 1]]
>>>tets= [[0, 1, 2, 3]]
>>>tetra=TetraMeshData(verts, tets)

2. Calculating the volume of a tetrahedron

>>>tetra.volume(0)
0.166666# == 1 x 1 x 0.5 / 3.0>>>tetra.volume_total()
0.166666

3. Calculating the barycenter of a tetrahedron

>>>tetra.barycenter(0)
Point3D(0.25, 0.25, 0.25)

4. Saving as .rsm

>>>tetra.save("tetra.rsm")

5. Loading from a .rsm file

>>>TetraMeshData.from_file("tetra.rsm")
<raysect.primitive.mesh.tetra_mesh.TetraMeshDataat0x1336e7440>

Speed test

I attempted to compare the instancing of Discrete3DMesh with TetraMeshData loaded from a file.
As a mesh file, I used the stanford_bunny.mesh file located in demos/resources/.

	elapse time (sec)
Create Discrete3DMesh	15.441215
Create TetraMeshData	15.894989
Load TetraMeshData from file	01.766619

I attained a loading speed for tetra mesh that is approximately 9 times faster than generating TetraMeshData using vertices and tetrahedral index arrays.
This difference is expected to widen as the vertex data size increases.
The test script that I used is in the collapsed section below.

Test script

fromdatetimeimporttimedeltafrompathlibimportPathfromtimeitimporttimeitimportnumpyasnpfromraysect.core.math.function.float.function3d.interpolateimportDiscrete3DMeshfromraysect.primitive.mesh.tetra_meshimportTetraMeshDatadefload_mesh(mesh_filepath):
vertices= []
tetrahedra= []
withopen(mesh_filepath, "r") asf:
lines=f.readlines()
i=0whilei<len(lines):
line=lines[i].strip()
ifnotlineorline.startswith("#"):
i+=1continue# Check section headers (case-sensitive)ifline=="Vertices":
i+=1vertex_count=int(lines[i].strip())
i+=1for_inrange(vertex_count):
parts=lines[i].strip().split()
# take only the first 3 coordinatesvertices.append([float(v) forvinparts[:3]])
i+=1continueelifline=="Tetrahedra":
i+=1tet_count=int(lines[i].strip())
i+=1for_inrange(tet_count):
parts=lines[i].strip().split()
# take only the first 4 indices, converting them to int.tetrahedra.append([int(idx) foridxinparts[:4]])
i+=1continuei+=1vertices=np.array(vertices)
tetrahedra=np.array(tetrahedra, dtype=np.int32) -1# 0-based indexingreturnvertices, tetrahedradefcreate_discrete3dmesh(vertices, tetrahedra):
returnDiscrete3DMesh(
vertices, tetrahedra, np.ones((tetrahedra.shape[0])), False, 0
)
defcreate_tetrameshdata(vertices, tetrahedra):
returnTetraMeshData(vertices, tetrahedra)
defload_tetrameshdata(mesh_filepath):
returnTetraMeshData.from_file(mesh_filepath)
if__name__=="__main__":
ROOT=Path(__file__).parentmesh_file=ROOT/"demos"/"resources"/"stanford_bunny.mesh"vertices, tetrahedra=load_mesh(mesh_file)
# Save tetramesh data in advancetetra=TetraMeshData(vertices, tetrahedra)
tetra_file=ROOT/"temp_tetra.rsm"tetra.save(tetra_file)
# === Measure time ===loop=5# create Discrete3DMeshresult=timeit(
"create_discrete3dmesh(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create Discrete3DMesh: {elapsed_time}")
# create TetraMeshDataresult=timeit(
"create_tetrameshdata(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create TetraMeshData: {elapsed_time}")
# load TetraMeshData from fileresult=timeit("load_tetrameshdata(tetra_file)", globals=globals(), number=loop)
elapsed_time=timedelta(seconds=result/loop)
print(f"Load TetraMeshData from file: {elapsed_time}")

Class structure

The structure of the TetraMeshData I implemented is as follows.

classDiagram
class KDTree3DCore {
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
}
class TetraMeshData {
+__init__(object vertices, object tetrahedra, bint tolerant=True)
+__getstate__()
+__setstate__(state)
+__reduce__()
+vertices
+tetrahedra
+Point3D vertex(int index)
+ndarray tetrahedron(int index)
+Point3D barycenter(int index)
+double volume(int index)
+double volume_total()
+BoundingBox3D bounding_box(AffineMatrix3D to_world)
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
+classmethod from_file(cls, file)
}
KDTree3DCore <|-- TetraMeshData

Loading

Details of unit test

I also implemented a unit test for TestMeshData in raysect/primitive/mesh/tests/test_tetra_mesh.py.
Below is a table showing the correspondence between the methods of the TestTetraMeshData class in unit tests and the methods tested in the TestMeshData class.

Test Method Name in TestTetraMeshData	Tested Method/Function in TetraMeshData
test_initialization()	__init__()
test_invalid_tetrahedron_indices()	__init__()
test_vertex_method()	vertex()
test_invalid_vertex_index()	vertex()
test_barycenter()	barycenter()
test_compute_volume()	volume(), volume_total()
test_is_contained()	is_contained()
test_bounding_box()	bounding_box()
test_pickle_state ()	__getstate__(), __setstate__(), save(), load()

---

I would appreciate it if you would review it and make any suggestions and comments.

[munechika-koyo]

Proposal of new feature: TetraMeshData

TL;DR: Reading files in the created tree structure will greatly enhance the speed of the process with this PR.

Description

I would propose a new feature, the TetraMeshData class, modeled after the existing MeshData definition.
(I have already suggested #407 but would like to split the function.)

It was implemented with the following objectives:

• Saving/Loading from the file formatted .rsm, which has constructed a K-D tree structure.
• Offering useful methods like calculating a tetrahedral volume and barycenter.

Additionally, this class will be used when loading mesh files formatted as .vtk or .obj, which contain tetrahedral (3-D unstructured) mesh data in the future.

Examples of the use:

1. Creating the instance of TetraMeshData

>>>fromraysect.primitive.mesh.tetra_meshimportTetraMeshData>>>verts= [[0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 1]]
>>>tets= [[0, 1, 2, 3]]
>>>tetra=TetraMeshData(verts, tets)

2. Calculating the volume of a tetrahedron

>>>tetra.volume(0)
0.166666# == 1 x 1 x 0.5 / 3.0>>>tetra.volume_total()
0.166666

3. Calculating the barycenter of a tetrahedron

>>>tetra.barycenter(0)
Point3D(0.25, 0.25, 0.25)

4. Saving as .rsm

>>>tetra.save("tetra.rsm")

5. Loading from a .rsm file

>>>TetraMeshData.from_file("tetra.rsm")
<raysect.primitive.mesh.tetra_mesh.TetraMeshDataat0x1336e7440>

Speed test

I attempted to compare the instancing of Discrete3DMesh with TetraMeshData loaded from a file.
As a mesh file, I used the stanford_bunny.mesh file located in demos/resources/.

	elapse time (sec)
Create Discrete3DMesh	15.441215
Create TetraMeshData	15.894989
Load TetraMeshData from file	01.766619

I attained a loading speed for tetra mesh that is approximately 9 times faster than generating TetraMeshData using vertices and tetrahedral index arrays.
This difference is expected to widen as the vertex data size increases.
The test script that I used is in the collapsed section below.

Test script

fromdatetimeimporttimedeltafrompathlibimportPathfromtimeitimporttimeitimportnumpyasnpfromraysect.core.math.function.float.function3d.interpolateimportDiscrete3DMeshfromraysect.primitive.mesh.tetra_meshimportTetraMeshDatadefload_mesh(mesh_filepath):
vertices= []
tetrahedra= []
withopen(mesh_filepath, "r") asf:
lines=f.readlines()
i=0whilei<len(lines):
line=lines[i].strip()
ifnotlineorline.startswith("#"):
i+=1continue# Check section headers (case-sensitive)ifline=="Vertices":
i+=1vertex_count=int(lines[i].strip())
i+=1for_inrange(vertex_count):
parts=lines[i].strip().split()
# take only the first 3 coordinatesvertices.append([float(v) forvinparts[:3]])
i+=1continueelifline=="Tetrahedra":
i+=1tet_count=int(lines[i].strip())
i+=1for_inrange(tet_count):
parts=lines[i].strip().split()
# take only the first 4 indices, converting them to int.tetrahedra.append([int(idx) foridxinparts[:4]])
i+=1continuei+=1vertices=np.array(vertices)
tetrahedra=np.array(tetrahedra, dtype=np.int32) -1# 0-based indexingreturnvertices, tetrahedradefcreate_discrete3dmesh(vertices, tetrahedra):
returnDiscrete3DMesh(
vertices, tetrahedra, np.ones((tetrahedra.shape[0])), False, 0
)
defcreate_tetrameshdata(vertices, tetrahedra):
returnTetraMeshData(vertices, tetrahedra)
defload_tetrameshdata(mesh_filepath):
returnTetraMeshData.from_file(mesh_filepath)
if__name__=="__main__":
ROOT=Path(__file__).parentmesh_file=ROOT/"demos"/"resources"/"stanford_bunny.mesh"vertices, tetrahedra=load_mesh(mesh_file)
# Save tetramesh data in advancetetra=TetraMeshData(vertices, tetrahedra)
tetra_file=ROOT/"temp_tetra.rsm"tetra.save(tetra_file)
# === Measure time ===loop=5# create Discrete3DMeshresult=timeit(
"create_discrete3dmesh(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create Discrete3DMesh: {elapsed_time}")
# create TetraMeshDataresult=timeit(
"create_tetrameshdata(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create TetraMeshData: {elapsed_time}")
# load TetraMeshData from fileresult=timeit("load_tetrameshdata(tetra_file)", globals=globals(), number=loop)
elapsed_time=timedelta(seconds=result/loop)
print(f"Load TetraMeshData from file: {elapsed_time}")

Class structure

The structure of the TetraMeshData I implemented is as follows.

classDiagram
class KDTree3DCore {
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
}
class TetraMeshData {
+__init__(object vertices, object tetrahedra, bint tolerant=True)
+__getstate__()
+__setstate__(state)
+__reduce__()
+vertices
+tetrahedra
+Point3D vertex(int index)
+ndarray tetrahedron(int index)
+Point3D barycenter(int index)
+double volume(int index)
+double volume_total()
+BoundingBox3D bounding_box(AffineMatrix3D to_world)
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
+classmethod from_file(cls, file)
}
KDTree3DCore <|-- TetraMeshData

Loading

Details of unit test

I also implemented a unit test for TestMeshData in raysect/primitive/mesh/tests/test_tetra_mesh.py.
Below is a table showing the correspondence between the methods of the TestTetraMeshData class in unit tests and the methods tested in the TestMeshData class.

Test Method Name in TestTetraMeshData	Tested Method/Function in TetraMeshData
test_initialization()	__init__()
test_invalid_tetrahedron_indices()	__init__()
test_vertex_method()	vertex()
test_invalid_vertex_index()	vertex()
test_barycenter()	barycenter()
test_compute_volume()	volume(), volume_total()
test_is_contained()	is_contained()
test_bounding_box()	bounding_box()
test_pickle_state ()	__getstate__(), __setstate__(), save(), load()

---

I would appreciate it if you would review it and make any suggestions and comments.

[munechika-koyo]

Proposal of new feature: TetraMeshData

TL;DR: Reading files in the created tree structure will greatly enhance the speed of the process with this PR.

Description

I would propose a new feature, the TetraMeshData class, modeled after the existing MeshData definition.
(I have already suggested #407 but would like to split the function.)

It was implemented with the following objectives:

• Saving/Loading from the file formatted .rsm, which has constructed a K-D tree structure.
• Offering useful methods like calculating a tetrahedral volume and barycenter.

Additionally, this class will be used when loading mesh files formatted as .vtk or .obj, which contain tetrahedral (3-D unstructured) mesh data in the future.

Examples of the use:

1. Creating the instance of TetraMeshData

>>>fromraysect.primitive.mesh.tetra_meshimportTetraMeshData>>>verts= [[0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 1]]
>>>tets= [[0, 1, 2, 3]]
>>>tetra=TetraMeshData(verts, tets)

2. Calculating the volume of a tetrahedron

>>>tetra.volume(0)
0.166666# == 1 x 1 x 0.5 / 3.0>>>tetra.volume_total()
0.166666

3. Calculating the barycenter of a tetrahedron

>>>tetra.barycenter(0)
Point3D(0.25, 0.25, 0.25)

4. Saving as .rsm

>>>tetra.save("tetra.rsm")

5. Loading from a .rsm file

>>>TetraMeshData.from_file("tetra.rsm")
<raysect.primitive.mesh.tetra_mesh.TetraMeshDataat0x1336e7440>

Speed test

I attempted to compare the instancing of Discrete3DMesh with TetraMeshData loaded from a file.
As a mesh file, I used the stanford_bunny.mesh file located in demos/resources/.

	elapse time (sec)
Create Discrete3DMesh	15.441215
Create TetraMeshData	15.894989
Load TetraMeshData from file	01.766619

I attained a loading speed for tetra mesh that is approximately 9 times faster than generating TetraMeshData using vertices and tetrahedral index arrays.
This difference is expected to widen as the vertex data size increases.
The test script that I used is in the collapsed section below.

Test script

fromdatetimeimporttimedeltafrompathlibimportPathfromtimeitimporttimeitimportnumpyasnpfromraysect.core.math.function.float.function3d.interpolateimportDiscrete3DMeshfromraysect.primitive.mesh.tetra_meshimportTetraMeshDatadefload_mesh(mesh_filepath):
vertices= []
tetrahedra= []
withopen(mesh_filepath, "r") asf:
lines=f.readlines()
i=0whilei<len(lines):
line=lines[i].strip()
ifnotlineorline.startswith("#"):
i+=1continue# Check section headers (case-sensitive)ifline=="Vertices":
i+=1vertex_count=int(lines[i].strip())
i+=1for_inrange(vertex_count):
parts=lines[i].strip().split()
# take only the first 3 coordinatesvertices.append([float(v) forvinparts[:3]])
i+=1continueelifline=="Tetrahedra":
i+=1tet_count=int(lines[i].strip())
i+=1for_inrange(tet_count):
parts=lines[i].strip().split()
# take only the first 4 indices, converting them to int.tetrahedra.append([int(idx) foridxinparts[:4]])
i+=1continuei+=1vertices=np.array(vertices)
tetrahedra=np.array(tetrahedra, dtype=np.int32) -1# 0-based indexingreturnvertices, tetrahedradefcreate_discrete3dmesh(vertices, tetrahedra):
returnDiscrete3DMesh(
vertices, tetrahedra, np.ones((tetrahedra.shape[0])), False, 0
)
defcreate_tetrameshdata(vertices, tetrahedra):
returnTetraMeshData(vertices, tetrahedra)
defload_tetrameshdata(mesh_filepath):
returnTetraMeshData.from_file(mesh_filepath)
if__name__=="__main__":
ROOT=Path(__file__).parentmesh_file=ROOT/"demos"/"resources"/"stanford_bunny.mesh"vertices, tetrahedra=load_mesh(mesh_file)
# Save tetramesh data in advancetetra=TetraMeshData(vertices, tetrahedra)
tetra_file=ROOT/"temp_tetra.rsm"tetra.save(tetra_file)
# === Measure time ===loop=5# create Discrete3DMeshresult=timeit(
"create_discrete3dmesh(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create Discrete3DMesh: {elapsed_time}")
# create TetraMeshDataresult=timeit(
"create_tetrameshdata(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create TetraMeshData: {elapsed_time}")
# load TetraMeshData from fileresult=timeit("load_tetrameshdata(tetra_file)", globals=globals(), number=loop)
elapsed_time=timedelta(seconds=result/loop)
print(f"Load TetraMeshData from file: {elapsed_time}")

Class structure

The structure of the TetraMeshData I implemented is as follows.

classDiagram
class KDTree3DCore {
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
}
class TetraMeshData {
+__init__(object vertices, object tetrahedra, bint tolerant=True)
+__getstate__()
+__setstate__(state)
+__reduce__()
+vertices
+tetrahedra
+Point3D vertex(int index)
+ndarray tetrahedron(int index)
+Point3D barycenter(int index)
+double volume(int index)
+double volume_total()
+BoundingBox3D bounding_box(AffineMatrix3D to_world)
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
+classmethod from_file(cls, file)
}
KDTree3DCore <|-- TetraMeshData

Loading

Details of unit test

I also implemented a unit test for TestMeshData in raysect/primitive/mesh/tests/test_tetra_mesh.py.
Below is a table showing the correspondence between the methods of the TestTetraMeshData class in unit tests and the methods tested in the TestMeshData class.

Test Method Name in TestTetraMeshData	Tested Method/Function in TetraMeshData
test_initialization()	__init__()
test_invalid_tetrahedron_indices()	__init__()
test_vertex_method()	vertex()
test_invalid_vertex_index()	vertex()
test_barycenter()	barycenter()
test_compute_volume()	volume(), volume_total()
test_is_contained()	is_contained()
test_bounding_box()	bounding_box()
test_pickle_state ()	__getstate__(), __setstate__(), save(), load()

---

I would appreciate it if you would review it and make any suggestions and comments.

[munechika-koyo]

Proposal of new feature: TetraMeshData

TL;DR: Reading files in the created tree structure will greatly enhance the speed of the process with this PR.

Description

I would propose a new feature, the TetraMeshData class, modeled after the existing MeshData definition.
(I have already suggested #407 but would like to split the function.)

It was implemented with the following objectives:

• Saving/Loading from the file formatted .rsm, which has constructed a K-D tree structure.
• Offering useful methods like calculating a tetrahedral volume and barycenter.

Additionally, this class will be used when loading mesh files formatted as .vtk or .obj, which contain tetrahedral (3-D unstructured) mesh data in the future.

Examples of the use:

1. Creating the instance of TetraMeshData

>>>fromraysect.primitive.mesh.tetra_meshimportTetraMeshData>>>verts= [[0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 1]]
>>>tets= [[0, 1, 2, 3]]
>>>tetra=TetraMeshData(verts, tets)

2. Calculating the volume of a tetrahedron

>>>tetra.volume(0)
0.166666# == 1 x 1 x 0.5 / 3.0>>>tetra.volume_total()
0.166666

3. Calculating the barycenter of a tetrahedron

>>>tetra.barycenter(0)
Point3D(0.25, 0.25, 0.25)

4. Saving as .rsm

>>>tetra.save("tetra.rsm")

5. Loading from a .rsm file

>>>TetraMeshData.from_file("tetra.rsm")
<raysect.primitive.mesh.tetra_mesh.TetraMeshDataat0x1336e7440>

Speed test

I attempted to compare the instancing of Discrete3DMesh with TetraMeshData loaded from a file.
As a mesh file, I used the stanford_bunny.mesh file located in demos/resources/.

	elapse time (sec)
Create Discrete3DMesh	15.441215
Create TetraMeshData	15.894989
Load TetraMeshData from file	01.766619

I attained a loading speed for tetra mesh that is approximately 9 times faster than generating TetraMeshData using vertices and tetrahedral index arrays.
This difference is expected to widen as the vertex data size increases.
The test script that I used is in the collapsed section below.

Test script

fromdatetimeimporttimedeltafrompathlibimportPathfromtimeitimporttimeitimportnumpyasnpfromraysect.core.math.function.float.function3d.interpolateimportDiscrete3DMeshfromraysect.primitive.mesh.tetra_meshimportTetraMeshDatadefload_mesh(mesh_filepath):
vertices= []
tetrahedra= []
withopen(mesh_filepath, "r") asf:
lines=f.readlines()
i=0whilei<len(lines):
line=lines[i].strip()
ifnotlineorline.startswith("#"):
i+=1continue# Check section headers (case-sensitive)ifline=="Vertices":
i+=1vertex_count=int(lines[i].strip())
i+=1for_inrange(vertex_count):
parts=lines[i].strip().split()
# take only the first 3 coordinatesvertices.append([float(v) forvinparts[:3]])
i+=1continueelifline=="Tetrahedra":
i+=1tet_count=int(lines[i].strip())
i+=1for_inrange(tet_count):
parts=lines[i].strip().split()
# take only the first 4 indices, converting them to int.tetrahedra.append([int(idx) foridxinparts[:4]])
i+=1continuei+=1vertices=np.array(vertices)
tetrahedra=np.array(tetrahedra, dtype=np.int32) -1# 0-based indexingreturnvertices, tetrahedradefcreate_discrete3dmesh(vertices, tetrahedra):
returnDiscrete3DMesh(
vertices, tetrahedra, np.ones((tetrahedra.shape[0])), False, 0
)
defcreate_tetrameshdata(vertices, tetrahedra):
returnTetraMeshData(vertices, tetrahedra)
defload_tetrameshdata(mesh_filepath):
returnTetraMeshData.from_file(mesh_filepath)
if__name__=="__main__":
ROOT=Path(__file__).parentmesh_file=ROOT/"demos"/"resources"/"stanford_bunny.mesh"vertices, tetrahedra=load_mesh(mesh_file)
# Save tetramesh data in advancetetra=TetraMeshData(vertices, tetrahedra)
tetra_file=ROOT/"temp_tetra.rsm"tetra.save(tetra_file)
# === Measure time ===loop=5# create Discrete3DMeshresult=timeit(
"create_discrete3dmesh(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create Discrete3DMesh: {elapsed_time}")
# create TetraMeshDataresult=timeit(
"create_tetrameshdata(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create TetraMeshData: {elapsed_time}")
# load TetraMeshData from fileresult=timeit("load_tetrameshdata(tetra_file)", globals=globals(), number=loop)
elapsed_time=timedelta(seconds=result/loop)
print(f"Load TetraMeshData from file: {elapsed_time}")

Class structure

The structure of the TetraMeshData I implemented is as follows.

classDiagram
class KDTree3DCore {
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
}
class TetraMeshData {
+__init__(object vertices, object tetrahedra, bint tolerant=True)
+__getstate__()
+__setstate__(state)
+__reduce__()
+vertices
+tetrahedra
+Point3D vertex(int index)
+ndarray tetrahedron(int index)
+Point3D barycenter(int index)
+double volume(int index)
+double volume_total()
+BoundingBox3D bounding_box(AffineMatrix3D to_world)
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
+classmethod from_file(cls, file)
}
KDTree3DCore <|-- TetraMeshData

Loading

Details of unit test

I also implemented a unit test for TestMeshData in raysect/primitive/mesh/tests/test_tetra_mesh.py.
Below is a table showing the correspondence between the methods of the TestTetraMeshData class in unit tests and the methods tested in the TestMeshData class.

Test Method Name in TestTetraMeshData	Tested Method/Function in TetraMeshData
test_initialization()	__init__()
test_invalid_tetrahedron_indices()	__init__()
test_vertex_method()	vertex()
test_invalid_vertex_index()	vertex()
test_barycenter()	barycenter()
test_compute_volume()	volume(), volume_total()
test_is_contained()	is_contained()
test_bounding_box()	bounding_box()
test_pickle_state ()	__getstate__(), __setstate__(), save(), load()

---

I would appreciate it if you would review it and make any suggestions and comments.

[munechika-koyo]

Proposal of new feature: TetraMeshData

TL;DR: Reading files in the created tree structure will greatly enhance the speed of the process with this PR.

Description

I would propose a new feature, the TetraMeshData class, modeled after the existing MeshData definition.
(I have already suggested #407 but would like to split the function.)

It was implemented with the following objectives:

• Saving/Loading from the file formatted .rsm, which has constructed a K-D tree structure.
• Offering useful methods like calculating a tetrahedral volume and barycenter.

Additionally, this class will be used when loading mesh files formatted as .vtk or .obj, which contain tetrahedral (3-D unstructured) mesh data in the future.

Examples of the use:

1. Creating the instance of TetraMeshData

>>>fromraysect.primitive.mesh.tetra_meshimportTetraMeshData>>>verts= [[0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 1]]
>>>tets= [[0, 1, 2, 3]]
>>>tetra=TetraMeshData(verts, tets)

2. Calculating the volume of a tetrahedron

>>>tetra.volume(0)
0.166666# == 1 x 1 x 0.5 / 3.0>>>tetra.volume_total()
0.166666

3. Calculating the barycenter of a tetrahedron

>>>tetra.barycenter(0)
Point3D(0.25, 0.25, 0.25)

4. Saving as .rsm

>>>tetra.save("tetra.rsm")

5. Loading from a .rsm file

>>>TetraMeshData.from_file("tetra.rsm")
<raysect.primitive.mesh.tetra_mesh.TetraMeshDataat0x1336e7440>

Speed test

I attempted to compare the instancing of Discrete3DMesh with TetraMeshData loaded from a file.
As a mesh file, I used the stanford_bunny.mesh file located in demos/resources/.

	elapse time (sec)
Create Discrete3DMesh	15.441215
Create TetraMeshData	15.894989
Load TetraMeshData from file	01.766619

I attained a loading speed for tetra mesh that is approximately 9 times faster than generating TetraMeshData using vertices and tetrahedral index arrays.
This difference is expected to widen as the vertex data size increases.
The test script that I used is in the collapsed section below.

Test script

fromdatetimeimporttimedeltafrompathlibimportPathfromtimeitimporttimeitimportnumpyasnpfromraysect.core.math.function.float.function3d.interpolateimportDiscrete3DMeshfromraysect.primitive.mesh.tetra_meshimportTetraMeshDatadefload_mesh(mesh_filepath):
vertices= []
tetrahedra= []
withopen(mesh_filepath, "r") asf:
lines=f.readlines()
i=0whilei<len(lines):
line=lines[i].strip()
ifnotlineorline.startswith("#"):
i+=1continue# Check section headers (case-sensitive)ifline=="Vertices":
i+=1vertex_count=int(lines[i].strip())
i+=1for_inrange(vertex_count):
parts=lines[i].strip().split()
# take only the first 3 coordinatesvertices.append([float(v) forvinparts[:3]])
i+=1continueelifline=="Tetrahedra":
i+=1tet_count=int(lines[i].strip())
i+=1for_inrange(tet_count):
parts=lines[i].strip().split()
# take only the first 4 indices, converting them to int.tetrahedra.append([int(idx) foridxinparts[:4]])
i+=1continuei+=1vertices=np.array(vertices)
tetrahedra=np.array(tetrahedra, dtype=np.int32) -1# 0-based indexingreturnvertices, tetrahedradefcreate_discrete3dmesh(vertices, tetrahedra):
returnDiscrete3DMesh(
vertices, tetrahedra, np.ones((tetrahedra.shape[0])), False, 0
)
defcreate_tetrameshdata(vertices, tetrahedra):
returnTetraMeshData(vertices, tetrahedra)
defload_tetrameshdata(mesh_filepath):
returnTetraMeshData.from_file(mesh_filepath)
if__name__=="__main__":
ROOT=Path(__file__).parentmesh_file=ROOT/"demos"/"resources"/"stanford_bunny.mesh"vertices, tetrahedra=load_mesh(mesh_file)
# Save tetramesh data in advancetetra=TetraMeshData(vertices, tetrahedra)
tetra_file=ROOT/"temp_tetra.rsm"tetra.save(tetra_file)
# === Measure time ===loop=5# create Discrete3DMeshresult=timeit(
"create_discrete3dmesh(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create Discrete3DMesh: {elapsed_time}")
# create TetraMeshDataresult=timeit(
"create_tetrameshdata(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create TetraMeshData: {elapsed_time}")
# load TetraMeshData from fileresult=timeit("load_tetrameshdata(tetra_file)", globals=globals(), number=loop)
elapsed_time=timedelta(seconds=result/loop)
print(f"Load TetraMeshData from file: {elapsed_time}")

Class structure

The structure of the TetraMeshData I implemented is as follows.

classDiagram
class KDTree3DCore {
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
}
class TetraMeshData {
+__init__(object vertices, object tetrahedra, bint tolerant=True)
+__getstate__()
+__setstate__(state)
+__reduce__()
+vertices
+tetrahedra
+Point3D vertex(int index)
+ndarray tetrahedron(int index)
+Point3D barycenter(int index)
+double volume(int index)
+double volume_total()
+BoundingBox3D bounding_box(AffineMatrix3D to_world)
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
+classmethod from_file(cls, file)
}
KDTree3DCore <|-- TetraMeshData

Loading

Details of unit test

I also implemented a unit test for TestMeshData in raysect/primitive/mesh/tests/test_tetra_mesh.py.
Below is a table showing the correspondence between the methods of the TestTetraMeshData class in unit tests and the methods tested in the TestMeshData class.

Test Method Name in TestTetraMeshData	Tested Method/Function in TetraMeshData
test_initialization()	__init__()
test_invalid_tetrahedron_indices()	__init__()
test_vertex_method()	vertex()
test_invalid_vertex_index()	vertex()
test_barycenter()	barycenter()
test_compute_volume()	volume(), volume_total()
test_is_contained()	is_contained()
test_bounding_box()	bounding_box()
test_pickle_state ()	__getstate__(), __setstate__(), save(), load()

---

I would appreciate it if you would review it and make any suggestions and comments.

[munechika-koyo]

Proposal of new feature: TetraMeshData

TL;DR: Reading files in the created tree structure will greatly enhance the speed of the process with this PR.

Description

I would propose a new feature, the TetraMeshData class, modeled after the existing MeshData definition.
(I have already suggested #407 but would like to split the function.)

It was implemented with the following objectives:

• Saving/Loading from the file formatted .rsm, which has constructed a K-D tree structure.
• Offering useful methods like calculating a tetrahedral volume and barycenter.

Additionally, this class will be used when loading mesh files formatted as .vtk or .obj, which contain tetrahedral (3-D unstructured) mesh data in the future.

Examples of the use:

1. Creating the instance of TetraMeshData

>>>fromraysect.primitive.mesh.tetra_meshimportTetraMeshData>>>verts= [[0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 1]]
>>>tets= [[0, 1, 2, 3]]
>>>tetra=TetraMeshData(verts, tets)

2. Calculating the volume of a tetrahedron

>>>tetra.volume(0)
0.166666# == 1 x 1 x 0.5 / 3.0>>>tetra.volume_total()
0.166666

3. Calculating the barycenter of a tetrahedron

>>>tetra.barycenter(0)
Point3D(0.25, 0.25, 0.25)

4. Saving as .rsm

>>>tetra.save("tetra.rsm")

5. Loading from a .rsm file

>>>TetraMeshData.from_file("tetra.rsm")
<raysect.primitive.mesh.tetra_mesh.TetraMeshDataat0x1336e7440>

Speed test

I attempted to compare the instancing of Discrete3DMesh with TetraMeshData loaded from a file.
As a mesh file, I used the stanford_bunny.mesh file located in demos/resources/.

	elapse time (sec)
Create Discrete3DMesh	15.441215
Create TetraMeshData	15.894989
Load TetraMeshData from file	01.766619

I attained a loading speed for tetra mesh that is approximately 9 times faster than generating TetraMeshData using vertices and tetrahedral index arrays.
This difference is expected to widen as the vertex data size increases.
The test script that I used is in the collapsed section below.

Test script

fromdatetimeimporttimedeltafrompathlibimportPathfromtimeitimporttimeitimportnumpyasnpfromraysect.core.math.function.float.function3d.interpolateimportDiscrete3DMeshfromraysect.primitive.mesh.tetra_meshimportTetraMeshDatadefload_mesh(mesh_filepath):
vertices= []
tetrahedra= []
withopen(mesh_filepath, "r") asf:
lines=f.readlines()
i=0whilei<len(lines):
line=lines[i].strip()
ifnotlineorline.startswith("#"):
i+=1continue# Check section headers (case-sensitive)ifline=="Vertices":
i+=1vertex_count=int(lines[i].strip())
i+=1for_inrange(vertex_count):
parts=lines[i].strip().split()
# take only the first 3 coordinatesvertices.append([float(v) forvinparts[:3]])
i+=1continueelifline=="Tetrahedra":
i+=1tet_count=int(lines[i].strip())
i+=1for_inrange(tet_count):
parts=lines[i].strip().split()
# take only the first 4 indices, converting them to int.tetrahedra.append([int(idx) foridxinparts[:4]])
i+=1continuei+=1vertices=np.array(vertices)
tetrahedra=np.array(tetrahedra, dtype=np.int32) -1# 0-based indexingreturnvertices, tetrahedradefcreate_discrete3dmesh(vertices, tetrahedra):
returnDiscrete3DMesh(
vertices, tetrahedra, np.ones((tetrahedra.shape[0])), False, 0
)
defcreate_tetrameshdata(vertices, tetrahedra):
returnTetraMeshData(vertices, tetrahedra)
defload_tetrameshdata(mesh_filepath):
returnTetraMeshData.from_file(mesh_filepath)
if__name__=="__main__":
ROOT=Path(__file__).parentmesh_file=ROOT/"demos"/"resources"/"stanford_bunny.mesh"vertices, tetrahedra=load_mesh(mesh_file)
# Save tetramesh data in advancetetra=TetraMeshData(vertices, tetrahedra)
tetra_file=ROOT/"temp_tetra.rsm"tetra.save(tetra_file)
# === Measure time ===loop=5# create Discrete3DMeshresult=timeit(
"create_discrete3dmesh(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create Discrete3DMesh: {elapsed_time}")
# create TetraMeshDataresult=timeit(
"create_tetrameshdata(vertices, tetrahedra)", globals=globals(), number=loop
)
elapsed_time=timedelta(seconds=result/loop)
print(f"Create TetraMeshData: {elapsed_time}")
# load TetraMeshData from fileresult=timeit("load_tetrameshdata(tetra_file)", globals=globals(), number=loop)
elapsed_time=timedelta(seconds=result/loop)
print(f"Load TetraMeshData from file: {elapsed_time}")

Class structure

The structure of the TetraMeshData I implemented is as follows.

classDiagram
class KDTree3DCore {
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
}
class TetraMeshData {
+__init__(object vertices, object tetrahedra, bint tolerant=True)
+__getstate__()
+__setstate__(state)
+__reduce__()
+vertices
+tetrahedra
+Point3D vertex(int index)
+ndarray tetrahedron(int index)
+Point3D barycenter(int index)
+double volume(int index)
+double volume_total()
+BoundingBox3D bounding_box(AffineMatrix3D to_world)
+bint is_contained(Point3D point)
+void save(object file)
+void load(object file)
+classmethod from_file(cls, file)
}
KDTree3DCore <|-- TetraMeshData

Loading

Details of unit test

I also implemented a unit test for TestMeshData in raysect/primitive/mesh/tests/test_tetra_mesh.py.
Below is a table showing the correspondence between the methods of the TestTetraMeshData class in unit tests and the methods tested in the TestMeshData class.

Test Method Name in TestTetraMeshData	Tested Method/Function in TetraMeshData
test_initialization()	__init__()
test_invalid_tetrahedron_indices()	__init__()
test_vertex_method()	vertex()
test_invalid_vertex_index()	vertex()
test_barycenter()	barycenter()
test_compute_volume()	volume(), volume_total()
test_is_contained()	is_contained()
test_bounding_box()	bounding_box()
test_pickle_state ()	__getstate__(), __setstate__(), save(), load()

---

I would appreciate it if you would review it and make any suggestions and comments.