Geometry library

The loadable library is accessible from Python and Ruby. It provides access to geometry functions for the construction of paths and surfaces compatible with those available in the Eilmer flow solver.

This is the reference manual for the Python flavour of the library which sits in the gdtk.geom package. For example, to construct a simple linear path element from within your Python script and then evaluate the midpoint on that line, you might try the following:

from gdtk.geom.vector3 import Vector3
from gdtk.geom.path import *

a = Vector3(0.0, 2.0)
b = Vector3(2.0, 0.0)
line_ab = Line(p0=a, p1=b)
print("line_ab=", line_ab)
c = line_ab(0.5)

If you have not yet read the Geometry Package User Guide, this is a good time to do so.

1. Installing the library

The geometry library for Python3 is part of a larger gas-dynamics toolkit and general getting started notes can be found at https://gdtk.uqcloud.net/docs/getting-started/prerequisites . There, you will see how to get a copy of the source code, and a list of what other software you will need to build and install the tool kit, and a collection of environment variables that need to be set.

To install the library, move to the gas source directory and use the make utility.

cd dgd/src/gas
make install

Even though this part of the package is a pure Python library, the rest of the loadable library, including gas models, will be built and installed with this command. So that the Python interpreter can find the installed library, set your environment variables with something like:

export DGD=$HOME/dgdinst
export PYTHONPATH=${PYTHONPATH}:${DGD}/lib

2. Geometric elements

These functions are the Python equivalent of the Lua functions found in the Geometry User Guide.

2.1. Vector3

This class defines geometric vector objects with three Cartesian components: x, y and z. The constructor accepts values for these components in a number of ways.

from gdtk.geom.vector3 import Vector3
p0 = Vector3(x=1.0, y=2.0, z=3.0)          # named arguments
p1 = Vector3(1.0, 2.0, 3.0)                # positional arguments x, y, z
p2 = Vector3([1.0, 2.0, 3.0])              # list of numbers
p3 = Vector3({'x':1.0, 'y':2.0, 'z':3.0})  # dictionary
p4 = Vector3(p3)                           # another Vector3 object

You need to specify at least the x component. The y and z components will default to values of 0.0.

2.1.1. Vector3 expressions

A number of methods have been defined so that you can write arithmetic expressions that involve Vector3 objects. To see the embedded doc strings, you can use the Python help function.

from gdtk.geom.vector3 import Vector3
help(Vector3)

Sample expressions include:

from gdtk.geom.vector3 import Vector3
p0 = Vector3(x=1.0, y=2.0, z=3.0)
p1 = Vector3(1.0, 2.0, 3.0)
p2 = +p0       # positive copy --> Vector3(x=1.0, y=2.0, z=3.0)
p2 = -p0       # negative copy --> Vector3(x=-1.0, y=-2.0, z=-3.0)
p2 = p0 + p1   # addition --> Vector3(x=2.0, y=4.0, z=6.0)
p2 = p0 - p1   # subtraction --> Vector3(x=0.0, y=0.0, z=0.0)
p2 += p1       # augmented addition --> Vector3(x=1.0, y=2.0, z=3.0)
p2 -= p1       # augmented subtraction --> Vector3(x=0.0, y=0.0, z=0.0)
p2 = Vector3(p0)
p2 = p0 * 3.0  # scaling --> Vector3(x=3.0, y=6.0, z=9.0)
p2 = 3.0 * p0  # scaling --> Vector3(x=3.0, y=6.0, z=9.0)
p2 = p0 / 3.0  # scaling --> Vector3(x=0.333333, y=0.666666, z=1.0)
p2 *= 3.0      # scaling --> Vector3(x=1.0, y=2.0, z=3.0)
p2 /= 3.0      # scaling --> Vector3(x=0.333333, y=0.666666, z=1.0)
p2.normalize() # scales to unit magnitude --> Vector3(x=0.267261, y=0.534522, z=0.801783)
a = abs(p2)    # magnitude --> 1.0
b = p1.dot(p0) # dot product --> 14.0
p2 = p0.unit() # new unit vector --> Vector3(x=0.267261, y=0.534522, z=0.801783)

There are also a pair of transformations, so that you change change into and out of a local coordinate system.

from gdtk.geom.vector3 import Vector3
p0 = Vector3(x=1.0, y=2.0, z=3.0)
c = Vector3(0.0, 1.0, 2.0)
n = Vector3(-1.0, 0.0, 0.0)
t1 = Vector3(0.0, -1.0, 0.0)
t2 = Vector3(0.0, 0.0, -1.0)
p1 = p0.transform_to_local_frame(n, t1, t2, c)  # --> Vector3(x=-1.0, y=-1.0, z=-1.0)
p2 = p1.transform_to_global_frame(n, t1, t2, c) # --> Vector3(x=1.0, y=2.0, z=3.0)

Remember that the Python assignment operator binds names to objects.

from gdtk.geom.vector3 import Vector3
p0 = Vector3(x=1.0, y=2.0, z=3.0)
p1 = Vector3(1.0, 2.0, 3.0)
p2 = p1        # assignment binds new name p2 to same object as p1
p2             # --> Vector3(x=1.0, y=2.0, z=3.0)
p1.normalize() # change object details
p2             # --> Vector3(x=0.267261, y=0.534522, z=0.801783)
p1 = p0        # change binding for name p1
p1             # --> Vector3(x=1.0, y=2.0, z=3.0)
p2             # --> Vector3(x=0.267261, y=0.534522, z=0.801783)

2.1.2. Other functions

The other functions in module eilmer.geom.vector3 include:

approxEqualVectors(a, b, rel_tol=0.01, abs_tol=1e-05)

Returns True if all components if vectors a and b are close.

cross(a, b)

Returns the Vector3 cross product of vector a with vector b.

dot(a, b)

Returns dot product of vector a with vector b.

hexahedron_properties(p0, p1, p2, p3, p4, p5, p6, p7)

Returns centroid and volume for the hexahedron defined by the 8 vertices.

hexahedron_volume(p0, p1, p2, p3, p4, p5, p6, p7)

Returns volume for the hexahedron defined by the 8 vertices.

quad_area(p0, p1, p2, p3)

Returns area for quadrilateral defined by the 4 corner points.

quad_centroid(p0, p1, p2, p3)

Returns centroid of quadrilateral.

quad_normal(p0, p1, p2, p3)

Returns unit normal for quadrilateral.

quad_properties(p0, p1, p2, p3)

Returns centroid, quadrilateral-defining unit vectors, and area.

tetrahedron_properties(p0, p1, p2, p3)

Returns centroid and volume of tetrahedron defined by 4 points.

unit(a)

Returns a new unit vector.

wedge_properties(p0, p1, p2, p3, p4, p5)

Returns centroid and volume for wedge defined by 6 points.

2.2. Path elements

The module gdtk.geom.path includes classes for making Path objects. A Path object may be called to evaluate a point on the path at parameter t, where the parametric range is 0.0 to 1.0.

The constructors for Path objects include:

Line(p0, p1)

Defines a straight line from point p0 (t=0) to point p1 (t=1).

Bezier(B)

Defines a Bezier curve from the sequence of points B.

NURBS(P, w, U, p)

Defines a NURBS from control points P, weights w, knot vector U and degree p.

Arc(a, b, c)

Defines a circular arc from point a to point b about centre c.

ArcLengthParameterizedPath(underlying_path, n=1000)

Derives path from underlying_path that has a uniformly-distributed set of points with parameter t.

Polyline(segments, closed=False, tolerance=1e-10)

Builds a single path from a sequence of Path objects. Setting closed=True will connect the ends with a straight-line segment if the original end points are further apart than tolerance.

Spline(points, closed=False, tolerance=1e-10)

Builds a spline of Bezier segments through the sequence of points. Setting closed=True will connect the ends with an extra segment if the original end points are further apart than tolerance.

2.3. ParametricSurface elements

The module gdtk.geom.surface includes classes for making ParametricSurface objects. These objects may be called two parameters r, and s to evaluate a point on the surface. Presently, only one class of ParametricSurface is implemented in the Python module.

CoonsPatch(north=None, east=None, south=None, west=None, p00=None, p10=None, p11=None, p01=None)

Define a surface using the method of interpolation described in S.A. Coons "Surfaces for Computer Aided Design of Space Forms" MAC TR-41 Contract No. AF-33 (6000-42859) MIT June 1967. The surface may be defined either by 4 Path objects as edges (named north, east, south, west) or by 4 corner points (named p00, p10, p11, p01). If defined by corner points, straight-line paths will be used for the 4 edges.

2.4. Cluster functions

The module gdtk.geom.cluster includes classes for constructing various ClusterFunction objects. These objects have a distribute_parameter_values(nv) method that returns a sequence of nv values spread over the parameter range 0.0 to 1.0, inclusive.

2.4.1. Linear

from gdtk.geom.cluster import *
cf = LinearFunction()
values = cf.distribute_parameter_values(11)

will result in values being array([0. , 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1. ]).

2.4.2. Roberts

If you want to cluster values toward either (or both) ends of the range, there is RobertsFunction(end0, end1, beta) where:

end0

Set True to cluster values toward t=0.

end1

Set True to cluster values toward t=1.

beta

The clustering parameter is larger than 1.0, and clustering increases in strength as beta approaches 1.0.

3. Grid elements

3.1. StructuredGrid

The constructor for a structured-grid object accepts all of its arguments via keywords. For example:

from gdtk.geom.vector3 import Vector3
from gdtk.geom.surface import CoonsPatch
from gdtk.geom.sgrid import StructuredGrid
from gdtk.geom.cluster import RobertsFunction

my_patch = CoonsPatch(p01=Vector3(0.1, 0.9), p11=Vector3(1.0, 1.0),
                      p00=Vector3(0.0, 0.0), p10=Vector3(0.9, 0.1))

cf_y = RobertsFunction(True, False, 1.1)
g = StructuredGrid(psurf=my_patch, niv=11, njv=5,
                   cf_list=[None, cf_y, None, cf_y])

g.write_to_vtk_file("my_grid.vtk")
sgrid example