Creating packings and filling volumes defined by boundary representation or constructive solid geometry.

For examples, see

yade.pack.SpherePack_toSimulation(self, rot=Matrix3(1, 0, 0, 0, 1, 0, 0, 0, 1), **kw)[source]

Append spheres directly to the simulation. In addition calling O.bodies.append, this method also appropriately sets periodic cell information of the simulation.

>>> from yade import pack; from math import *
>>> sp=pack.SpherePack()


Create random periodic packing with 20 spheres:

>>> sp.makeCloud((0,0,0),(5,5,5),rMean=.5,rRelFuzz=.5,periodic=True,num=20)
20


Virgin simulation is aperiodic:

>>> O.reset()
>>> O.periodic
False


Add generated packing to the simulation, rotated by 45° along +z

>>> sp.toSimulation(rot=Quaternion((0,0,1),pi/4),color=(0,0,1))
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19]


Periodic properties are transferred to the simulation correctly, including rotation (this could be avoided by explicitly passing “hSize=O.cell.hSize” as an argument):

>>> O.periodic
True
>>> O.cell.refSize
Vector3(5,5,5)
>>> O.cell.hSize
Matrix3(3.53553,-3.53553,0, 3.53553,3.53553,0, 0,0,5)


The current state (even if rotated) is taken as mechanically undeformed, i.e. with identity transformation:

>>> O.cell.trsf
Matrix3(1,0,0, 0,1,0, 0,0,1)

Parameters: rot (Quaternion/Matrix3) – rotation of the packing, which will be applied on spheres and will be used to set Cell.trsf as well. **kw – passed to utils.sphere list of body ids added (like O.bodies.append)
yade.pack.filterSpherePack(predicate, spherePack, returnSpherePack=None, **kw)[source]

Using given SpherePack instance, return spheres that satisfy predicate. It returns either a pack.SpherePack (if returnSpherePack) or a list. The packing will be recentered to match the predicate and warning is given if the predicate is larger than the packing.

yade.pack.gtsSurface2Facets(surf, **kw)[source]

Construct facets from given GTS surface. **kw is passed to utils.facet.

yade.pack.gtsSurfaceBestFitOBB(surf)[source]

Return (Vector3 center, Vector3 halfSize, Quaternion orientation) describing best-fit oriented bounding box (OBB) for the given surface. See cloudBestFitOBB for details.

yade.pack.hexaNet(radius, cornerCoord=[0, 0, 0], xLength=1.0, yLength=0.5, mos=0.08, a=0.04, b=0.04, startAtCorner=True, isSymmetric=False, **kw)[source]

Definition of the particles for a hexagonal wire net in the x-y-plane for the WireMatPM.

Parameters: radius – radius of the particle cornerCoord – coordinates of the lower left corner of the net xLenght – net length in x-direction yLenght – net length in y-direction mos – mesh opening size (horizontal distance between the double twists) a – length of double-twist b – height of single wire section startAtCorner – if true the generation starts with a double-twist at the lower left corner isSymmetric – defines if the net is symmetric with respect to the y-axis set of spheres which defines the net (net) and exact dimensions of the net (lx,ly).

Note

This packing works for the WireMatPM only. The particles at the corner are always generated first. For examples on how to use this packing see examples/WireMatPM. In order to create the proper interactions for the net the interaction radius has to be adapted in the simulation.

class yade.pack.inConvexPolyhedron(inherits Predicate)[source]
aabb()[source]
center()[source]
dim()[source]
class yade.pack.inGtsSurface_py(inherits Predicate)[source]

This class was re-implemented in c++, but should stay here to serve as reference for implementing Predicates in pure python code. C++ allows us to play dirty tricks in GTS which are not accessible through pygts itself; the performance penalty of pygts comes from fact that if constructs and destructs bb tree for the surface at every invocation of gts.Point().is_inside(). That is cached in the c++ code, provided that the surface is not manipulated with during lifetime of the object (user’s responsibility).

Predicate for GTS surfaces. Constructed using an already existing surfaces, which must be closed.

Note

Padding is optionally supported by testing 6 points along the axes in the pad distance. This must be enabled in the ctor by saying doSlowPad=True. If it is not enabled and pad is not zero, warning is issued.

aabb()[source]
center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade.pack.inHalfSpace(inherits Predicate)[source]

Predicate returning True any points, with infinite bounding box.

aabb()[source]
center()[source]
dim()[source]
class yade.pack.inSpace(inherits Predicate)[source]

Predicate returning True for any points, with infinite bounding box.

aabb()[source]
center()[source]
dim()[source]
yade.pack.randomDensePack(predicate, radius, material=-1, dim=None, cropLayers=0, rRelFuzz=0.0, spheresInCell=0, memoizeDb=None, useOBB=False, memoDbg=False, color=None, returnSpherePack=None, seed=-1)[source]

Generator of random dense packing with given geometry properties, using TriaxialTest (aperiodic) or PeriIsoCompressor (periodic). The periodicity depens on whether the spheresInCell parameter is given.

O.switchScene() magic is used to have clean simulation for TriaxialTest without deleting the original simulation. This function therefore should never run in parallel with some code accessing your simulation.

yade.pack.randomPeriPack(radius, initSize, rRelFuzz=0.0, memoizeDb=None, noPrint=False, seed=-1)[source]

Generate periodic dense packing.

A cell of initSize is stuffed with as many spheres as possible, then we run periodic compression with PeriIsoCompressor, just like with randomDensePack.

Parameters: radius – mean sphere radius rRelFuzz – relative fuzz of sphere radius (equal distribution); see the same param for randomDensePack. initSize – initial size of the periodic cell. SpherePack object, which also contains periodicity information.
yade.pack.regularHexa(predicate, radius, gap, **kw)[source]

Return set of spheres in regular hexagonal grid, clipped inside solid given by predicate. Created spheres will have given radius and will be separated by gap space.

yade.pack.regularOrtho(predicate, radius, gap, **kw)[source]

Return set of spheres in regular orthogonal grid, clipped inside solid given by predicate. Created spheres will have given radius and will be separated by gap space.

yade.pack.revolutionSurfaceMeridians(sects, angles, origin=Vector3(0, 0, 0), orientation=Quaternion((1, 0, 0), 0))[source]

Revolution surface given sequences of 2d points and sequence of corresponding angles, returning sequences of 3d points representing meridian sections of the revolution surface. The 2d sections are turned around z-axis, but they can be transformed using the origin and orientation arguments to give arbitrary orientation.

yade.pack.sweptPolylines2gtsSurface(pts, threshold=0, capStart=False, capEnd=False)[source]

Create swept suface (as GTS triangulation) given same-length sequences of points (as 3-tuples).

If threshold is given (>0), then

• degenerate faces (with edges shorter than threshold) will not be created
• gts.Surface().cleanup(threshold) will be called before returning, which merges vertices mutually closer than threshold. In case your pts are closed (last point concident with the first one) this will the surface strip of triangles. If you additionally have capStart==True and capEnd==True, the surface will be closed.

Note

capStart and capEnd make the most naive polygon triangulation (diagonals) and will perhaps fail for non-convex sections.

Warning

the algorithm connects points sequentially; if two polylines are mutually rotated or have inverse sense, the algorithm will not detect it and connect them regardless in their given order.

Creation, manipulation, IO for generic sphere packings.

class yade._packSpheres.SpherePack

Set of spheres represented as centers and radii. This class is returned by pack.randomDensePack, pack.randomPeriPack and others. The object supports iteration over spheres, as in

>>> sp=SpherePack()

>>> for sphere in sp: print sphere[0],sphere[1]   ## same, but without unpacking the tuple automatically

>>> for i in range(0,len(sp)): print sp[i][0], sp[i][1]   ## same, but accessing spheres by index


Special constructors

Construct from list of [(c1,r1),(c2,r2),…]. To convert two same-length lists of centers and radii, construct with zip(centers,radii).

__init__((object)arg1[, (list)list]) → None :

Empty constructor, optionally taking list [ ((cx,cy,cz),r), … ] for initial data.

aabb((SpherePack)arg1) → tuple :

Get axis-aligned bounding box coordinates, as 2 3-tuples.

add((SpherePack)arg1, (Vector3)arg2, (float)arg3) → None :

appliedPsdScaling

A factor between 0 and 1, uniformly applied on all sizes of of the PSD.

cellFill((SpherePack)arg1, (Vector3)arg2) → None :

Repeat the packing (if periodic) so that the results has dim() >= given size. The packing retains periodicity, but changes cellSize. Raises exception for non-periodic packing.

cellRepeat((SpherePack)arg1, (Vector3i)arg2) → None :

Repeat the packing given number of times in each dimension. Periodicity is retained, cellSize changes. Raises exception for non-periodic packing.

cellSize

Size of periodic cell; is Vector3(0,0,0) if not periodic. (Change this property only if you know what you’re doing).

center((SpherePack)arg1) → Vector3 :

Return coordinates of the bounding box center.

dim((SpherePack)arg1) → Vector3 :

Return dimensions of the packing in terms of aabb(), as a 3-tuple.

fromList((SpherePack)arg1, (list)arg2) → None :

Make packing from given list, same format as for constructor. Discards current data.

fromList( (SpherePack)arg1, (object)centers, (object)radii) -> None :
Make packing from given list, same format as for constructor. Discards current data.
fromSimulation((SpherePack)arg1) → None :

Make packing corresponding to the current simulation. Discards current data.

getClumps((SpherePack)arg1) → tuple :

Return lists of sphere ids sorted by clumps they belong to. The return value is (standalones,[clump1,clump2,…]), where each item is list of id’s of spheres.

hasClumps((SpherePack)arg1) → bool :

Whether this object contains clumps.

isPeriodic

was the packing generated in periodic boundaries?

load((SpherePack)arg1, (str)fileName) → None :

makeCloud((SpherePack)arg1[, (Vector3)minCorner=Vector3(0, 0, 0)[, (Vector3)maxCorner=Vector3(0, 0, 0)[, (float)rMean=-1[, (float)rRelFuzz=0[, (int)num=-1[, (bool)periodic=False[, (float)porosity=0.65[, (object)psdSizes=[][, (object)psdCumm=[][, (bool)distributeMass=False[, (int)seed=-1[, (Matrix3)hSize=Matrix3(0, 0, 0, 0, 0, 0, 0, 0, 0)]]]]]]]]]]]]) → int :

Create a random cloud of particles enclosed in a parallelepiped. The resulting packing is a gas-like state with no contacts between particles initially. Usually used as a first step before reaching a dense packing.

Parameters: minCorner (Vector3) – lower corner of an axis-aligned box maxCorner (Vector3) – upper corner of an axis-aligned box hSize (Matrix3) – base vectors of a generalized box (arbitrary parallelepiped, typically Cell::hSize), superseeds minCorner and maxCorner if defined. For periodic boundaries only. rMean (float) – mean radius or spheres rRelFuzz (float) – dispersion of radius relative to rMean num (int) – number of spheres to be generated. If negative (default), generate as many as possible with stochastic sizes, ending after a fixed number of tries to place the sphere in space, else generate exactly num spheres with deterministic size distribution. periodic (bool) – whether the packing to be generated should be periodic porosity (float) – initial guess for the iterative generation procedure (if num>1). The algorithm will be retrying until the number of generated spheres is num. The first iteration tries with the provided porosity, but next iterations increase it if necessary (hence an initialy high porosity can speed-up the algorithm). If psdSizes is not defined, rRelFuzz ($$z$$) and num ($$N$$) are used so that the porosity given ($$\rho$$) is approximately achieved at the end of generation, $$r_m=\sqrt[3]{\frac{V(1-\rho)}{\frac{4}{3}\pi(1+z^2)N}}$$. The default is $$\rho$$=0.5. The optimal value depends on rRelFuzz or psdSizes. psdSizes – sieve sizes (particle diameters) when particle size distribution (PSD) is specified. psdCumm – cummulative fractions of particle sizes given by psdSizes; must be the same length as psdSizes and should be non-decreasing. distributeMass (bool) – if True, given distribution will be used to distribute sphere’s mass rather than radius of them. seed – number used to initialize the random number generator. number of created spheres, which can be lower than num depending on the method used.

Note

• Works in 2D if minCorner[k]=maxCorner[k] for one coordinate.

• If num is defined, then sizes generation is deterministic, giving the best fit of target distribution. It enables spheres placement in descending size order, thus giving lower porosity than the random generation.

• By default (with distributeMass==False), the distribution is applied to particle count (i.e. particle count percent passing). The typical geomechanics sense of “particle size distribution” is the distribution of mass fraction (i.e. mass percent passing); this can be achieved with distributeMass=True.

• Sphere radius distribution can be specified using one of the following ways:

1. rMean, rRelFuzz and num gives uniform radius distribution in $$rMean×(1 \pm rRelFuzz)$$. Less than num spheres can be generated if it is too high.
2. rRelFuzz, num and (optional) porosity, which estimates mean radius so that porosity is attained at the end. rMean must be less than 0 (default). porosity is only an initial guess for the generation algorithm, which will retry with higher porosity until the prescibed num is obtained.
3. psdSizes and psdCumm, two arrays specifying points of the particle size distribution function. As many spheres as possible are generated.
4. psdSizes, psdCumm, num, and (optional) porosity, like above but if num is not obtained, psdSizes will be scaled down uniformly, until num is obtained (see appliedPsdScaling).
makeClumpCloud((SpherePack)arg1, (Vector3)minCorner, (Vector3)maxCorner, (object)clumps[, (bool)periodic=False[, (int)num=-1[, (int)seed=-1]]]) → int :

Create a random loose packing of clumps the same way makeCloud does with spheres. The parameters minCorner, maxCorner, periodic, num and seed are the same as in makeCloud. The parameter clumps is a list containing all the different clumps to be appended as SpherePack objects. Here is an exemple that shows how to create a cloud made of 10 identical clumps :

clp = SpherePack([((0,0,0), 1e-2), ((1e-2,0,0), 1e-2)]) # The clump we want a cloud of
sp = SpherePack()
sp.makeClumpCloud((0,0,0), (1,1,1), [clp], num=10, seed=42)
sp.toSimulation() # All the particles in the cloud are now appended to O.bodies

psd((SpherePack)arg1[, (int)bins=50[, (bool)mass=True]]) → tuple :

Return particle size distribution of the packing.

Parameters: bins (int) – number of bins between minimum and maximum diameter mass – Compute relative mass rather than relative particle count for each bin. Corresponds to distributeMass parameter for makeCloud. tuple of (cumm,edges), where cumm are cummulative fractions for respective diameters and edges are those diameter values. Dimension of both arrays is equal to bins+1.
relDensity((SpherePack)arg1) → float :

Relative packing density, measured as sum of spheres’ volumes / aabb volume. (Sphere overlaps are ignored.)

rotate((SpherePack)arg1, (Vector3)axis, (float)angle) → None :

Rotate all spheres around packing center (in terms of aabb()), given axis and angle of the rotation.

save((SpherePack)arg1, (str)fileName) → None :

Save packing to external text file (will be overwritten).

scale((SpherePack)arg1, (float)arg2) → None :

Scale the packing around its center (in terms of aabb()) by given factor (may be negative).

toList((SpherePack)arg1) → list :

Return packing data as python list.

toSimulation(rot=Matrix3(1, 0, 0, 0, 1, 0, 0, 0, 1), **kw)

Append spheres directly to the simulation. In addition calling O.bodies.append, this method also appropriately sets periodic cell information of the simulation.

>>> from yade import pack; from math import *
>>> sp=pack.SpherePack()


Create random periodic packing with 20 spheres:

>>> sp.makeCloud((0,0,0),(5,5,5),rMean=.5,rRelFuzz=.5,periodic=True,num=20)
20


Virgin simulation is aperiodic:

>>> O.reset()
>>> O.periodic
False


Add generated packing to the simulation, rotated by 45° along +z

>>> sp.toSimulation(rot=Quaternion((0,0,1),pi/4),color=(0,0,1))
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19]


Periodic properties are transferred to the simulation correctly, including rotation (this could be avoided by explicitly passing “hSize=O.cell.hSize” as an argument):

>>> O.periodic
True
>>> O.cell.refSize
Vector3(5,5,5)
>>> O.cell.hSize
Matrix3(3.53553,-3.53553,0, 3.53553,3.53553,0, 0,0,5)


The current state (even if rotated) is taken as mechanically undeformed, i.e. with identity transformation:

>>> O.cell.trsf
Matrix3(1,0,0, 0,1,0, 0,0,1)

Parameters: rot (Quaternion/Matrix3) – rotation of the packing, which will be applied on spheres and will be used to set Cell.trsf as well. **kw – passed to utils.sphere list of body ids added (like O.bodies.append)
translate((SpherePack)arg1, (Vector3)arg2) → None :

Translate all spheres by given vector.

class yade._packSpheres.SpherePackIterator
__init__((object)arg1, (SpherePackIterator)arg2) → None
next()

__next__( (SpherePackIterator)arg1) -> tuple

Spatial predicates for volumes (defined analytically or by triangulation).

class yade._packPredicates.Predicate
aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.PredicateBoolean(inherits Predicate)

Boolean operation on 2 predicates (abstract class)

A
B
__init__()

Raises an exception This class cannot be instantiated from Python

aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.PredicateDifference(inherits PredicateBoolean → Predicate)

Difference (conjunction with negative predicate) of 2 predicates. A point has to be inside the first and outside the second predicate. Can be constructed using the - operator on predicates: pred1 - pred2.

A
B
__init__((object)arg1, (object)arg2, (object)arg3) → None
aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.PredicateIntersection(inherits PredicateBoolean → Predicate)

Intersection (conjunction) of 2 predicates. A point has to be inside both predicates. Can be constructed using the & operator on predicates: pred1 & pred2.

A
B
__init__((object)arg1, (object)arg2, (object)arg3) → None
aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.PredicateSymmetricDifference(inherits PredicateBoolean → Predicate)

SymmetricDifference (exclusive disjunction) of 2 predicates. A point has to be in exactly one predicate of the two. Can be constructed using the ^ operator on predicates: pred1 ^ pred2.

A
B
__init__((object)arg1, (object)arg2, (object)arg3) → None
aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.PredicateUnion(inherits PredicateBoolean → Predicate)

Union (non-exclusive disjunction) of 2 predicates. A point has to be inside any of the two predicates to be inside. Can be constructed using the | operator on predicates: pred1 | pred2.

A
B
__init__((object)arg1, (object)arg2, (object)arg3) → None
aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.inAlignedBox(inherits Predicate)

Axis-aligned box predicate

__init__((object)arg1, (Vector3)minAABB, (Vector3)maxAABB) → None :

Ctor taking minumum and maximum points of the box (as 3-tuples).

aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.inCylinder(inherits Predicate)

Cylinder predicate

__init__((object)arg1, (Vector3)centerBottom, (Vector3)centerTop, (float)radius) → None :

Ctor taking centers of the lateral walls (as 3-tuples) and radius.

aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.inEllipsoid(inherits Predicate)

Ellipsoid predicate

__init__((object)arg1, (Vector3)centerPoint, (Vector3)abc) → None :

Ctor taking center of the ellipsoid (3-tuple) and its 3 radii (3-tuple).

aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.inGtsSurface(inherits Predicate)

GTS surface predicate

__init__((object)arg1, (object)surface[, (bool)noPad]) → None :

Ctor taking a gts.Surface() instance, which must not be modified during instance lifetime. The optional noPad can disable padding (if set to True), which speeds up calls several times. Note: padding checks inclusion of 6 points along +- cardinal directions in the pad distance from given point, which is not exact.

aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
surf

The associated gts.Surface object.

class yade._packPredicates.inHyperboloid(inherits Predicate)

Hyperboloid predicate

__init__((object)arg1, (Vector3)centerBottom, (Vector3)centerTop, (float)radius, (float)skirt) → None :

Ctor taking centers of the lateral walls (as 3-tuples), radius at bases and skirt (middle radius).

aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.inParallelepiped(inherits Predicate)

Parallelepiped predicate

__init__((object)arg1, (Vector3)o, (Vector3)a, (Vector3)b, (Vector3)c) → None :

Ctor taking four points: o (for origin) and then a, b, c which define endpoints of 3 respective edges from o.

aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.inSphere(inherits Predicate)

Sphere predicate.

__init__((object)arg1, (Vector3)center, (float)radius) → None :

Ctor taking center (as a 3-tuple) and radius

aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3
class yade._packPredicates.notInNotch(inherits Predicate)

Outside of infinite, rectangle-shaped notch predicate

__init__((object)arg1, (Vector3)centerPoint, (Vector3)edge, (Vector3)normal, (float)aperture) → None :

Ctor taking point in the symmetry plane, vector pointing along the edge, plane normal and aperture size. The side inside the notch is edge×normal. Normal is made perpendicular to the edge. All vectors are normalized at construction time.

aabb((Predicate)arg1) → tuple

aabb( (Predicate)arg1) -> None

center((Predicate)arg1) → Vector3
dim((Predicate)arg1) → Vector3

Computation of oriented bounding box for cloud of points.

yade._packObb.cloudBestFitOBB((tuple)arg1) → tuple

Return (Vector3 center, Vector3 halfSize, Quaternion orientation) of best-fit oriented bounding-box for given tuple of points (uses brute-force velome minimization, do not use for very large clouds).