def get_DIC_image(dic_map, scaling_factor):
# Construct an array of Euler angles
grain_quats = np.empty((len(dic_map), 4))
# Transformation orientations from EBSD orientation reference frame
# to EBSD spatial reference frame
frame_transform = Quat.fromAxisAngle(np.array((1, 0, 0)), np.pi)
if dic_map.ebsdMap.crystalSym == 'hexagonal':
# Convert hex convention from y // a2 of EBSD map to x // a1 for DAMASK
hex_transform = Quat.fromAxisAngle(np.array([0, 0, 1]), -np.pi / 6)
for i, grain in enumerate(dic_map):
grain_quats[i] = (hex_transform * grain.ebsdGrain.refOri *
frame_transform).quatCoef
else:
for i, grain in enumerate(dic_map):
grain_quats[i] = (grain.ebsdGrain.refOri *
frame_transform).quatCoef
# Filter out -1 (grain boundary points) and -2 (too small grains)
# values in the grain image
grain_image = dic_map.grains
remove_boundary_points(grain_image)
remove_small_grain_points(grain_image)
remove_boundary_points(grain_image)
remove_boundary_points(grain_image, force_remove=True)
# scale down image if needed
if scaling_factor != 1:
grain_image = zoom(grain_image,
scaling_factor,
order=0,
prefilter=False,
mode='nearest')
# downstream expects grain numbering to start at 0 not 1
grain_image -= 1
DIC_image = {
'orientations': {
'type': 'quat',
'unit_cell_alignment': {
'x': 'a'
},
'quaternions': grain_quats,
'P': 1, # DefDAP uses P=+1 (e.g see `defdap.quat.Quat.__mul__`)
'quat_component_ordering': 'scalar-vector',
},
'grains': grain_image,
}
try:
DIC_image['scale'] = dic_map.scale
except ValueError:
pass
return DIC_image
def construct_beta_quat_array(
ebsd_map: ebsd.Map,
alpha_phase_id: int = 0,
variant_map: np.ndarray = None,
) -> np.ndarray:
"""Construct
Parameters
----------
ebsd_map:
EBSD map to assign the beta variants for.
alpha_phase_id
Index of the alpha phase in the EBSD map.
"""
if variant_map is None:
variant_map = construct_variant_map(ebsd_map, alpha_phase_id)
transformations = []
for sym in unq_hex_syms:
transformations.append(burg_trans * sym.conjugate)
trans_comps = Quat.extract_quat_comps(transformations)
trans_comps = trans_comps[:, variant_map[variant_map >= 0]]
quat_comps = Quat.extract_quat_comps(ebsd_map.quatArray[variant_map >= 0])
quat_comps_beta = np.empty_like(quat_comps)
# transformations[variant] * quat
quat_comps_beta[0, :] = (trans_comps[0, :] * quat_comps[0, :] -
trans_comps[1, :] * quat_comps[1, :] -
trans_comps[2, :] * quat_comps[2, :] -
trans_comps[3, :] * quat_comps[3, :])
quat_comps_beta[1, :] = (trans_comps[1, :] * quat_comps[0, :] +
trans_comps[0, :] * quat_comps[1, :] -
trans_comps[3, :] * quat_comps[2, :] +
trans_comps[2, :] * quat_comps[3, :])
quat_comps_beta[2, :] = (trans_comps[2, :] * quat_comps[0, :] +
trans_comps[0, :] * quat_comps[2, :] -
trans_comps[1, :] * quat_comps[3, :] +
trans_comps[3, :] * quat_comps[1, :])
quat_comps_beta[3, :] = (trans_comps[3, :] * quat_comps[0, :] +
trans_comps[0, :] * quat_comps[3, :] -
trans_comps[2, :] * quat_comps[1, :] +
trans_comps[1, :] * quat_comps[2, :])
# swap into positive hemisphere if required
quat_comps_beta[:, quat_comps_beta[0, :] < 0] *= -1
beta_quat_array = np.empty_like(ebsd_map.quatArray)
beta_quat_array[variant_map < 0] = Quat(1, 0, 0, 0)
for i, idx in enumerate(zip(*np.where(variant_map >= 0))):
beta_quat_array[idx] = Quat(quat_comps_beta[:, i])
return beta_quat_array
def get_EBSD_image(ebsd_map, scaling_factor):
# Construct an array of Euler angles
grain_quats = np.empty((len(ebsd_map), 4))
# Transformation orientations from EBSD orientation reference frame
# to EBSD spatial reference frame
frame_transform = Quat.fromAxisAngle(np.array((1, 0, 0)), np.pi)
if ebsd_map.crystalSym == 'hexagonal':
# Convert hex convention from y // a2 of EBSD map to x // a1 for DAMASK
hex_transform = Quat.fromAxisAngle(np.array([0, 0, 1]), -np.pi / 6)
for i, grain in enumerate(ebsd_map):
grain_quats[i] = (hex_transform * grain.refOri *
frame_transform).quatCoef
else:
for i, grain in enumerate(ebsd_map):
grain_quats[i] = (grain.refOri * frame_transform).quatCoef
# Filter out -2 (too small grains) values in the grain image
grain_image = ebsd_map.grains
remove_small_grain_points(grain_image)
# scale down image if needed
if scaling_factor != 1:
grain_image = zoom(grain_image,
scaling_factor,
order=0,
prefilter=False,
mode='nearest')
# downstream expects grain numbering to start at 0 not 1
grain_image -= 1
EBSD_image = {
'orientations': {
'type': 'quat',
'unit_cell_alignment': {
'x': 'a'
},
'quaternions': grain_quats,
'P': 1, # DefDAP uses P=+1 (e.g see `defdap.quat.Quat.__mul__`)
'quat_component_ordering': 'scalar-vector',
},
'grains': grain_image,
'scale': ebsd_map.scale,
'phase_labels': [phase.name for phase in ebsd_map.phases],
'grain_phases': [grain.phaseID for grain in ebsd_map],
}
return EBSD_image
def plotGrainDataIPF(self,
direction,
mapData=None,
grainData=None,
grainIds=-1,
**kwargs):
"""
Plot IPF of grain reference (average) orientations with
points coloured by grain average values from map data.
Parameters
----------
mapData : numpy.ndarray
Array of map data to grain average. This must be cropped!
direction : numpy.ndarray
Vector of reference direction for the IPF.
plotColourBar : bool, optional
Set to False to exclude the colour bar from the plot.
vmin : float, optional
Minimum value of colour scale.
vmax : float, optional
Maximum value for colour scale.
cLabel : str, optional
Colour bar label text.
"""
# Set default plot parameters then update with any input
plotParams = {}
plotParams.update(kwargs)
if grainData is None:
if mapData is None:
raise ValueError("Either 'mapData' or 'grainData' must "
"be supplied.")
else:
grainData = self.calcGrainAv(mapData, grainIds=grainIds)
# Check that grains have been detected in the map
self.checkGrainsDetected()
if type(grainIds) is int and grainIds == -1:
grainIds = range(len(self))
if len(grainData) != len(grainIds):
raise Exception("Must be 1 value for each grain in grainData.")
grainOri = np.empty(len(grainIds), dtype=Quat)
for i, grainId in enumerate(grainIds):
grain = self[grainId]
grainOri[i] = grain.refOri
plot = Quat.plotIPF(grainOri,
direction,
self.crystalSym,
c=grainData,
**plotParams)
return plot
def calc_beta_oris_from_misori(
alpha_ori: Quat,
neighbour_oris: List[Quat],
burg_tol: float = 5) -> Tuple[List[List[Quat]], List[float]]:
"""Calculate the possible beta orientations for a given alpha
orientation using the misorientation relation to neighbour orientations.
Parameters
----------
alpha_ori
A quaternion representing the alpha orientation
neighbour_oris
Quaternions representing neighbour grain orientations
burg_tol
The threshold misorientation angle to determine neighbour relations
Returns
-------
list of lists of defdap.Quat.quat
Possible beta orientations, grouped by each neighbour. Any
neighbour with deviation greater than the tolerance is excluded.
list of float
Deviations from perfect Burgers transformation
"""
burg_tol *= np.pi / 180.
# This needed to move further up calculation process
unq_cub_sym_comps = Quat.extract_quat_comps(unq_cub_syms)
alpha_ori_inv = alpha_ori.conjugate
beta_oris = []
beta_devs = []
for neighbour_ori in neighbour_oris:
min_misoris, min_cub_sym_idxs = calc_misori_of_variants(
alpha_ori_inv, neighbour_ori, unq_cub_sym_comps)
# find the hex symmetries (i, j) from give the minimum
# deviation from the burgers relation for the minimum store:
# the deviation, the hex symmetries (i, j) and the cubic
# symmetry if the deviation is over a threshold then set
# cubic symmetry to -1
min_misori_idx = np.unravel_index(np.argmin(min_misoris),
min_misoris.shape)
burg_dev = min_misoris[min_misori_idx]
if burg_dev < burg_tol:
beta_oris.append(
beta_oris_from_cub_sym(alpha_ori,
min_cub_sym_idxs[min_misori_idx],
int(min_misori_idx[0])))
beta_devs.append(burg_dev)
return beta_oris, beta_devs
def buildMisOriList(self, calcAxis=False):
quatCompsSym = Quat.calcSymEqvs(self.quatList, self.crystalSym)
if self.refOri is None:
self.refOri = Quat.calcAverageOri(quatCompsSym)
misOriArray, minQuatComps = Quat.calcMisOri(quatCompsSym, self.refOri)
self.averageMisOri = misOriArray.mean()
self.misOriList = list(misOriArray)
if calcAxis:
# Now for axis calulation
refOriInv = self.refOri.conjugate
misOriAxis = np.empty((3, minQuatComps.shape[1]))
Dq = np.empty((4, minQuatComps.shape[1]))
# refOriInv * minQuat for all points (* is quaternion product)
# change to minQuat * refOriInv
Dq[0, :] = (refOriInv[0] * minQuatComps[0, :] - refOriInv[1] * minQuatComps[1, :] -
refOriInv[2] * minQuatComps[2, :] - refOriInv[3] * minQuatComps[3, :])
Dq[1, :] = (refOriInv[1] * minQuatComps[0, :] + refOriInv[0] * minQuatComps[1, :] +
refOriInv[3] * minQuatComps[2, :] - refOriInv[2] * minQuatComps[3, :])
Dq[2, :] = (refOriInv[2] * minQuatComps[0, :] + refOriInv[0] * minQuatComps[2, :] +
refOriInv[1] * minQuatComps[3, :] - refOriInv[3] * minQuatComps[1, :])
Dq[3, :] = (refOriInv[3] * minQuatComps[0, :] + refOriInv[0] * minQuatComps[3, :] +
refOriInv[2] * minQuatComps[1, :] - refOriInv[1] * minQuatComps[2, :])
Dq[:, Dq[0] < 0] = -Dq[:, Dq[0] < 0]
# numpy broadcasting taking care of different array sizes
misOriAxis[:, :] = (2 * Dq[1:4, :] * np.arccos(Dq[0, :])) / np.sqrt(1 - np.power(Dq[0, :], 2))
# hack it back into a list. Need to change self.*List to be arrays, it was a bad decision to
# make them lists in the beginning
self.misOriAxisList = []
for row in misOriAxis.transpose():
self.misOriAxisList.append(row)
def transformData(self):
"""
Rotate map by 180 degrees and transform quats
"""
print("\rTransforming EBSD data...", end="")
self.eulerAngleArray = self.eulerAngleArray[:, ::-1, ::-1]
self.bandContrastArray = self.bandContrastArray[::-1, ::-1]
self.phaseArray = self.phaseArray[::-1, ::-1]
self.buildQuatArray()
transformQuat = Quat.fromAxisAngle(np.array([0, 0, 1]), np.pi)
for i in range(self.xDim):
for j in range(self.yDim):
self.quatArray[j, i] = self.quatArray[j, i] * transformQuat
print("\rDone ", end="")
def buildQuatArray(self):
"""
Build quaternion array
"""
print("\rBuilding quaternion array...", end="")
self.checkDataLoaded()
if self.quatArray is None:
# create the array of quat objects
self.quatArray = Quat.createManyQuats(self.eulerAngleArray)
print("\rDone ", end="")
return
def beta_oris() -> List[Quat]:
"""The 6 possible beta orientations for the `ori_single_valid` fixture."""
return [
Quat(0.11460994, 0.97057328, 0.10987647, -0.18105038),
Quat(0.71894262, 0.52597293, -0.12611599, 0.43654181),
Quat(0.4812518, 0.86403135, -0.00937589, 0.14750805),
Quat(0.23608204, -0.12857975, 0.96217918, 0.04408791),
Quat(0.39243229, -0.10727847, -0.59866151, -0.68999466),
Quat(0.62851433, -0.23585823, 0.36351768, -0.64590675)
]
def test_ori_tol(beta_oris: List[Quat]):
ori_tol = 5.
possible_beta_oris = [
[beta_oris[0] * Quat.fromAxisAngle(np.array([1, 0, 0]),
1.01 * ori_tol * np.pi / 180)],
[beta_oris[1]],
[beta_oris[1]],
[beta_oris[1], beta_oris[3], beta_oris[4]]
]
variant_count = recon.count_beta_variants(
beta_oris, possible_beta_oris, ori_tol
)
expected_variant_count = [0, 3, 0, 1, 1, 0]
assert all(np.equal(variant_count, expected_variant_count))
def plotIPFMap(self, direction, **kwargs):
# Set default plot parameters then update with any input
plotParams = {}
plotParams.update(kwargs)
# calculate IPF colours
IPFcolours = Quat.calcIPFcolours(
self.quatArray.flatten(),
direction,
self.crystalSym
)
# reshape back to map shape array
IPFcolours = np.reshape(IPFcolours, (self.yDim, self.xDim, 3))
plot = MapPlot.create(self, IPFcolours, **plotParams)
return plot
def calc_misori_of_variants(
alpha_ori_inv: Quat, neighbour_ori: Quat,
unq_cub_sym_comps: np.ndarray) -> Tuple[np.ndarray, np.ndarray]:
"""Calculate possible symmetry variants between two orientations.
Calculate all possible sym variants for misorientation between two
orientations undergoing a Burgers type transformation. Then calculate
the misorientation to the nearest cubic symmetry, this is the deviation
to a perfect Burgers transformation.
Parameters
----------
alpha_ori_inv
Inverse of first orientation
neighbour_ori
Second orientation
unq_cub_sym_comps
Components of the unique cubic symmetries
Returns
-------
min_misoris : np.ndarray
The minimum misorientation for each of the possible beta variants - shape (12, 12)
min_cub_sym_idx : np.ndarray
The minimum cubic symmetry index for each of the possible variants - shape (12, 12)
"""
# calculate all possible S^B_m (eqn 11. from [1]) from the
# measured misorientation from 2 neighbour alpha grains
# for each S^B_m calculate the 'closest' cubic symmetry
# (from reduced subset) and the deviation from this symmetry
# Vectorised calculation of:
# hex_sym[j].inv * ((neighbour_ori * alpha_ori_inv) * hex_sym[i])
# labelled: d = h2.inv * (c * h1)
hex_sym_comps = Quat.extract_quat_comps(hex_syms)
c = (neighbour_ori * alpha_ori_inv).quatCoef
h1 = np.repeat(hex_sym_comps, 12, axis=1) # outer loop
h2 = np.tile(hex_sym_comps, (1, 12)) # inner loop
d = np.zeros_like(h1)
c_dot_h1 = c[1] * h1[1] + c[2] * h1[2] + c[3] * h1[3]
c_dot_h2 = c[1] * h2[1] + c[2] * h2[2] + c[3] * h2[3]
h1_dot_h2 = h1[1] * h2[1] + h1[2] * h2[2] + h1[3] * h2[3]
d[0] = (c[0] * h1[0] * h2[0] - h2[0] * c_dot_h1 + c[0] * h1_dot_h2 +
h1[0] * c_dot_h2 + h2[1] * (c[2] * h1[3] - c[3] * h1[2]) + h2[2] *
(c[3] * h1[1] - c[1] * h1[3]) + h2[3] *
(c[1] * h1[2] - c[2] * h1[1]))
d[1] = (c[0] * h2[0] * h1[1] + h1[0] * h2[0] * c[1] -
c[0] * h1[0] * h2[1] + c_dot_h1 * h2[1] + c_dot_h2 * h1[1] -
h1_dot_h2 * c[1] + h2[0] * (c[2] * h1[3] - c[3] * h1[2]) + c[0] *
(h1[2] * h2[3] - h1[3] * h2[2]) + h1[0] *
(c[2] * h2[3] - c[3] * h2[2]))
d[2] = (c[0] * h2[0] * h1[2] + h1[0] * h2[0] * c[2] -
c[0] * h1[0] * h2[2] + c_dot_h1 * h2[2] + c_dot_h2 * h1[2] -
h1_dot_h2 * c[2] + h2[0] * (c[3] * h1[1] - c[1] * h1[3]) + c[0] *
(h1[3] * h2[1] - h1[1] * h2[3]) + h1[0] *
(c[3] * h2[1] - c[1] * h2[3]))
d[3] = (c[0] * h2[0] * h1[3] + h1[0] * h2[0] * c[3] -
c[0] * h1[0] * h2[3] + c_dot_h1 * h2[3] + c_dot_h2 * h1[3] -
h1_dot_h2 * c[3] + h2[0] * (c[1] * h1[2] - c[2] * h1[1]) + c[0] *
(h1[1] * h2[2] - h1[2] * h2[1]) + h1[0] *
(c[1] * h2[2] - c[2] * h2[1]))
# Vectorised calculation of:
# burg_trans * (d * burg_trans.inv)
# labelled: beta_vars = b * (c * b.inv)
b = burg_trans.quatCoef
beta_vars = np.zeros_like(h1)
b_dot_b = b[1] * b[1] + b[2] * b[2] + b[3] * b[3]
b_dot_d = b[1] * d[1] + b[2] * d[2] + b[3] * d[3]
beta_vars[0] = d[0] * (b[0] * b[0] + b_dot_b)
beta_vars[1] = (d[1] * (b[0] * b[0] - b_dot_b) + 2 * b_dot_d * b[1] +
2 * b[0] * (b[2] * d[3] - b[3] * d[2]))
beta_vars[2] = (d[2] * (b[0] * b[0] - b_dot_b) + 2 * b_dot_d * b[2] +
2 * b[0] * (b[3] * d[1] - b[1] * d[3]))
beta_vars[3] = (d[3] * (b[0] * b[0] - b_dot_b) + 2 * b_dot_d * b[3] +
2 * b[0] * (b[1] * d[2] - b[2] * d[1]))
# calculate misorientation to each of the cubic symmetries
misoris = np.einsum("ij,ik->jk", beta_vars, unq_cub_sym_comps)
misoris = np.abs(misoris)
misoris[misoris > 1] = 1.
misoris = 2 * np.arccos(misoris)
# find the cubic symmetry with minimum misorientation for each of
# the beta misorientation variants
min_cub_sym_idx = np.argmin(misoris, axis=1)
min_misoris = misoris[np.arange(144), min_cub_sym_idx]
# reshape to 12 x 12 for each of the hex sym multiplications
min_cub_sym_idx = min_cub_sym_idx.reshape((12, 12))
min_misoris = min_misoris.reshape((12, 12))
return min_misoris, min_cub_sym_idx
def test_calc(self):
mock_map = Mock(spec=ebsd.Map)
mock_map.quatArray = np.array([
Quat([0.31666724, -0.91588522, 0.13405465, -0.20713635]),
Quat([0.17612928, -0.20214505, -0.21599713, -0.93886159]),
Quat([0.24967011, -0.224283, -0.26045348, -0.90527673]),
Quat([0.68070418, -0.53435491, 0.23619103, -0.44195072]),
Quat([0.7293098, -0.34952765, -0.07857021, -0.58289309]),
Quat([0.00169551, 0.14777021, 0.12010977, 0.98169992]),
Quat([0.47365417, -0.20536756, -0.20468846, -0.83161201]),
Quat([0.78776179, -0.27972705, 0.09209514, -0.54101999]),
Quat([0.33727506, -0.13311405, -0.13924545, -0.92148624]),
Quat([0.41137644, -0.85164404, 0.21420115, -0.24411007]),
Quat([0.26092708, -0.27640969, -0.29049792, -0.87813763]),
Quat([0.51237664, -0.62480323, -0.16679851, -0.56503925])
]).reshape((3, 4))
mock_map.shape = mock_map.quatArray.shape
variant_map = np.array([[0, 3, -1, 1],
[2, 2, 4, 4],
[5, -1, 1, 2]])
beta_quat_array = recon.construct_beta_quat_array(
mock_map, variant_map=variant_map
)
expected_comps = [
np.array([0.3586687, -0.42249045, -0.28570384, 0.78181321]),
np.array([0.04396812, -0.20834367, -0.45309011, 0.86566105]),
np.array([1., 0., 0., 0.]),
np.array([0.13489744, -0.93982694, 0.30093181, -0.08926393]),
np.array([0.66997125, -0.66745906, 0.24828395, 0.2097427]),
np.array([0.46111146, -0.72030688, -0.25269131, -0.4524172]),
np.array([0.31807292, 0.2294237, -0.791435, 0.46885503]),
np.array([0.1535714, -0.26761921, -0.83850223, 0.44912114]),
np.array([0.51219503, 0.6609771, -0.31826009, -0.44662741]),
np.array([1., 0., 0., 0.]),
np.array([0.26763438, -0.77098237, -0.48040495, -0.32119948]),
np.array([0.56407948, -0.75208451, -0.06296879, 0.33498979])
]
assert all([np.allclose(quat.quatCoef, row) for quat, row
in zip(beta_quat_array.flat, expected_comps)])
def ori_quat_list_valid() -> List[Quat]:
"""A list of sample quaternion representing the orientations of grains."""
return [
Quat(0.22484510, 0.45464871, -0.70807342, 0.49129550),
Quat(0.36520321, 0.25903472, -0.40342268, 0.79798357)
]
def ori_single_valid_3() -> Quat:
"""A single sample quaternion representing the orientation of a grain."""
return Quat(0.8730071, -0.41360125, -0.02295757, -0.25742097)
def ori_single_valid_2() -> Quat:
"""A single sample quaternion representing the orientation of a grain."""
return Quat(0.11939881, -0.36445855, -0.67237386, -0.63310922)
def ori_single_valid() -> Quat:
"""A single sample quaternion representing the orientation of a grain."""
return Quat(0.22484510, 0.45464871, -0.70807342, 0.49129550)
def plotOriSpread(self, direction=np.array([0, 0, 1]), **kwargs):
plotParams = {'marker': '.'}
plotParams.update(kwargs)
return Quat.plotIPF(self.quatList, direction, self.crystalSym,
**plotParams)
import numpy as np
from defdap.quat import Quat
hex_syms = Quat.symEqv("hexagonal")
# subset of hexagonal symmetries that give unique orientations when the
# Burgers transformation is applied
unq_hex_syms = [
hex_syms[0], hex_syms[5], hex_syms[4], hex_syms[2], hex_syms[10],
hex_syms[11]
]
cubic_syms = Quat.symEqv("cubic")
# subset of cubic symmetries that give unique orientations when the
# Burgers transformation is applied
unq_cub_syms = [
cubic_syms[0], cubic_syms[7], cubic_syms[9], cubic_syms[1], cubic_syms[22],
cubic_syms[16], cubic_syms[12], cubic_syms[15], cubic_syms[4],
cubic_syms[8], cubic_syms[21], cubic_syms[20]
]
# HCP -> BCC
burg_eulers = np.array([135, 90, 354.74]) * np.pi / 180
burg_trans = Quat.fromEulerAngles(*burg_eulers).conjugate
def calc_beta_oris_from_boundary_misori(
grain: ebsd.Grain,
neighbour_network: nx.Graph,
quat_array: np.ndarray,
alpha_phase_id: int,
burg_tol: float = 5
) -> Tuple[List[List[Quat]], List[float], List[Quat]]:
"""Calculate the possible beta orientations for pairs of alpha and
neighbour orientations using the misorientation relation to neighbour
orientations.
Parameters
----------
grain
The grain currently being reconstructed
neighbour_network
A neighbour network mapping grain boundary connectivity
quat_array
Array of quaternions, representing the orientations of the pixels of the EBSD map
burg_tol :
The threshold misorientation angle to determine neighbour relations
Returns
-------
list of lists of defdap.Quat.quat
Possible beta orientations, grouped by each neighbour. Any
neighbour with deviation greater than the tolerance is excluded.
list of float
Deviations from perfect Burgers transformation
list of Quat
Alpha orientations
"""
# This needed to move further up calculation process
unq_cub_sym_comps = Quat.extract_quat_comps(unq_cub_syms)
beta_oris = []
beta_devs = []
alpha_oris = []
neighbour_grains = neighbour_network.neighbors(grain)
neighbour_grains = [
grain for grain in neighbour_grains if grain.phaseID == alpha_phase_id
]
for neighbour_grain in neighbour_grains:
bseg = neighbour_network[grain][neighbour_grain]['boundary']
# check sense of bseg
if grain is bseg.grain1:
ipoint = 0
else:
ipoint = 1
for boundary_point_pair in bseg.boundaryPointPairsX:
point = boundary_point_pair[ipoint]
alpha_ori = quat_array[point[1], point[0]]
point = boundary_point_pair[ipoint - 1]
neighbour_ori = quat_array[point[1], point[0]]
min_misoris, min_cub_sym_idxs = calc_misori_of_variants(
alpha_ori.conjugate, neighbour_ori, unq_cub_sym_comps)
# find the hex symmetries (i, j) from give the minimum
# deviation from the burgers relation for the minimum store:
# the deviation, the hex symmetries (i, j) and the cubic
# symmetry if the deviation is over a threshold then set
# cubic symmetry to -1
min_misori_idx = np.unravel_index(np.argmin(min_misoris),
min_misoris.shape)
burg_dev = min_misoris[min_misori_idx]
if burg_dev < burg_tol / 180 * np.pi:
beta_oris.append(
beta_oris_from_cub_sym(alpha_ori,
min_cub_sym_idxs[min_misori_idx],
int(min_misori_idx[0])))
beta_devs.append(burg_dev)
alpha_oris.append(alpha_ori)
return beta_oris, beta_devs, alpha_oris
def findBoundaries(self, boundDef=10):
"""
Find grain boundaries
:param boundDef: critical misorientation
:type boundDef: float
"""
self.buildQuatArray()
print("\rFinding boundaries...", end="")
syms = Quat.symEqv(self.crystalSym)
numSyms = len(syms)
# array to store quat components of initial and symmetric equivalents
quatComps = np.empty((numSyms, 4, self.yDim, self.xDim))
# populate with initial quat components
for i, row in enumerate(self.quatArray):
for j, quat in enumerate(row):
quatComps[0, :, i, j] = quat.quatCoef
# loop of over symmetries and apply to initial quat components
# (excluding first symmetry as this is the identity transformation)
for i, sym in enumerate(syms[1:], start=1):
# sym[i] * quat for all points (* is quaternion product)
quatComps[i, 0] = (quatComps[0, 0] * sym[0] - quatComps[0, 1] * sym[1] -
quatComps[0, 2] * sym[2] - quatComps[0, 3] * sym[3])
quatComps[i, 1] = (quatComps[0, 0] * sym[1] + quatComps[0, 1] * sym[0] -
quatComps[0, 2] * sym[3] + quatComps[0, 3] * sym[2])
quatComps[i, 2] = (quatComps[0, 0] * sym[2] + quatComps[0, 2] * sym[0] -
quatComps[0, 3] * sym[1] + quatComps[0, 1] * sym[3])
quatComps[i, 3] = (quatComps[0, 0] * sym[3] + quatComps[0, 3] * sym[0] -
quatComps[0, 1] * sym[2] + quatComps[0, 2] * sym[1])
# swap into positve hemisphere if required
quatComps[i, :, quatComps[i, 0] < 0] *= -1
# Arrays to store neigbour misorientation in positive x and y direction
misOrix = np.zeros((numSyms, self.yDim, self.xDim))
misOriy = np.zeros((numSyms, self.yDim, self.xDim))
# loop over symmetries calculating misorientation to initial
for i in range(numSyms):
for j in range(self.xDim - 1):
misOrix[i, :, j] = abs(np.einsum("ij,ij->j", quatComps[0, :, :, j], quatComps[i, :, :, j + 1]))
for j in range(self.yDim - 1):
misOriy[i, j, :] = abs(np.einsum("ij,ij->j", quatComps[0, :, j, :], quatComps[i, :, j + 1, :]))
misOrix[misOrix > 1] = 1
misOriy[misOriy > 1] = 1
# find min misorientation (max here as misorientaion is cos of this)
misOrix = np.max(misOrix, axis=0)
misOriy = np.max(misOriy, axis=0)
# convert to misorientation in degrees
misOrix = 360 * np.arccos(misOrix) / np.pi
misOriy = 360 * np.arccos(misOriy) / np.pi
# set boundary locations where misOrix or misOriy are greater than set value
self.boundaries = np.zeros((self.yDim, self.xDim), dtype=int)
for i in range(self.xDim):
for j in range(self.yDim):
if (misOrix[j, i] > boundDef) or (misOriy[j, i] > boundDef):
self.boundaries[j, i] = -1
print("\rDone ", end="")
return
def calcNye(self):
"""
Calculates Nye tensor and related GND density for the EBSD map.
Stores result in self.Nye and self.GND.
"""
self.buildQuatArray()
print("\rFinding boundaries...", end="")
syms = Quat.symEqv(self.crystalSym)
numSyms = len(syms)
# array to store quat components of initial and symmetric equivalents
quatComps = np.empty((numSyms, 4, self.yDim, self.xDim))
# populate with initial quat components
for i, row in enumerate(self.quatArray):
for j, quat in enumerate(row):
quatComps[0, :, i, j] = quat.quatCoef
# loop of over symmetries and apply to initial quat components
# (excluding first symmetry as this is the identity transformation)
for i, sym in enumerate(syms[1:], start=1):
# sym[i] * quat for all points (* is quaternion product)
quatComps[i, 0] = (quatComps[0, 0] * sym[0] - quatComps[0, 1] * sym[1] -
quatComps[0, 2] * sym[2] - quatComps[0, 3] * sym[3])
quatComps[i, 1] = (quatComps[0, 0] * sym[1] + quatComps[0, 1] * sym[0] -
quatComps[0, 2] * sym[3] + quatComps[0, 3] * sym[2])
quatComps[i, 2] = (quatComps[0, 0] * sym[2] + quatComps[0, 2] * sym[0] -
quatComps[0, 3] * sym[1] + quatComps[0, 1] * sym[3])
quatComps[i, 3] = (quatComps[0, 0] * sym[3] + quatComps[0, 3] * sym[0] -
quatComps[0, 1] * sym[2] + quatComps[0, 2] * sym[1])
# swap into positve hemisphere if required
quatComps[i, :, quatComps[i, 0] < 0] *= -1
# Arrays to store neigbour misorientation in positive x and y direction
misOrix = np.zeros((numSyms, self.yDim, self.xDim))
misOriy = np.zeros((numSyms, self.yDim, self.xDim))
# loop over symmetries calculating misorientation to initial
for i in range(numSyms):
for j in range(self.xDim - 1):
misOrix[i, :, j] = abs(np.einsum("ij,ij->j", quatComps[0, :, :, j], quatComps[i, :, :, j + 1]))
for j in range(self.yDim - 1):
misOriy[i, j, :] = abs(np.einsum("ij,ij->j", quatComps[0, :, j, :], quatComps[i, :, j + 1, :]))
misOrix[misOrix > 1] = 1
misOriy[misOriy > 1] = 1
# find min misorientation (max here as misorientaion is cos of this)
argmisOrix = np.argmax(misOrix, axis=0)
argmisOriy = np.argmax(misOriy, axis=0)
misOrix = np.max(misOrix, axis=0)
misOriy = np.max(misOriy, axis=0)
# convert to misorientation in degrees
misOrix = 360 * np.arccos(misOrix) / np.pi
misOriy = 360 * np.arccos(misOriy) / np.pi
# calculate relative elastic distortion tensors at each point in the two directions
betaderx = np.zeros((3, 3, self.yDim, self.xDim))
betadery = betaderx
for i in range(self.xDim - 1):
for j in range(self.yDim - 1):
q0x = Quat(quatComps[0, 0, j, i], quatComps[0, 1, j, i],
quatComps[0, 2, j, i], quatComps[0, 3, j, i])
qix = Quat(quatComps[argmisOrix[j, i], 0, j, i + 1],
quatComps[argmisOrix[j, i], 1, j, i + 1],
quatComps[argmisOrix[j, i], 2, j, i + 1],
quatComps[argmisOrix[j, i], 3, j, i + 1])
misoquatx = qix.conjugate * q0x
# change stepsize to meters
betaderx[:, :, j, i] = (Quat.rotMatrix(misoquatx) - np.eye(3)) / self.stepSize / 1e-6
q0y = Quat(quatComps[0, 0, j, i], quatComps[0, 1, j, i],
quatComps[0, 2, j, i], quatComps[0, 3, j, i])
qiy = Quat(quatComps[argmisOriy[j, i], 0, j + 1, i],
quatComps[argmisOriy[j, i], 1, j + 1, i],
quatComps[argmisOriy[j, i], 2, j + 1, i],
quatComps[argmisOriy[j, i], 3, j + 1, i])
misoquaty = qiy.conjugate * q0y
# change stepsize to meters
betadery[:, :, j, i] = (Quat.rotMatrix(misoquaty) - np.eye(3)) / self.stepSize / 1e-6
# Calculate the Nye Tensor
alpha = np.empty((3, 3, self.yDim, self.xDim))
bavg = 1.4e-10 # Burgers vector
alpha[0, 2] = (betadery[0, 0] - betaderx[0, 1]) / bavg
alpha[1, 2] = (betadery[1, 0] - betaderx[1, 1]) / bavg
alpha[2, 2] = (betadery[2, 0] - betaderx[2, 1]) / bavg
alpha[:, 1] = betaderx[:, 2] / bavg
alpha[:, 0] = -1 * betadery[:, 2] / bavg
# Calculate 3 possible L1 norms of Nye tensor for total
# disloction density
alpha_total3 = np.empty((self.yDim, self.xDim))
alpha_total5 = np.empty((self.yDim, self.xDim))
alpha_total9 = np.empty((self.yDim, self.xDim))
alpha_total3 = 30 / 10. *(
abs(alpha[0, 2]) + abs(alpha[1, 2]) +
abs(alpha[2, 2])
)
alpha_total5 = 30 / 14. * (
abs(alpha[0, 2]) + abs(alpha[1, 2]) + abs(alpha[2, 2]) +
abs(alpha[1, 0]) + abs(alpha[0, 1])
)
alpha_total9 = 30 / 20. * (
abs(alpha[0, 2]) + abs(alpha[1, 2]) + abs(alpha[2, 2]) +
abs(alpha[0, 0]) + abs(alpha[1, 0]) + abs(alpha[2, 0]) +
abs(alpha[0, 1]) + abs(alpha[1, 1]) + abs(alpha[2, 1])
)
alpha_total3[abs(alpha_total3) < 1] = 1e12
alpha_total5[abs(alpha_total3) < 1] = 1e12
alpha_total9[abs(alpha_total3) < 1] = 1e12
# choose from the different alpha_totals according to preference;
# see Ruggles GND density paper
self.GND = alpha_total9
self.Nye = alpha
def calcAverageOri(self):
quatCompsSym = Quat.calcSymEqvs(self.quatList, self.crystalSym)
self.refOri = Quat.calcAverageOri(quatCompsSym)
def plotRefOri(self, direction=np.array([0, 0, 1]), **kwargs):
plotParams = {'marker': '+'}
plotParams.update(kwargs)
return Quat.plotIPF([self.refOri], direction, self.crystalSym,
**plotParams)
