def makemap(self):
"""
Generate a map fit versus position - no args?
"""
for m in self.ubisread:
g = grain(m)
self.scandata=self.allscandata.copy()
self.compute_gv(g)
h=np.dot(g.ubi,np.transpose(self.gv))
hint=np.floor(h+0.5).astype(np.int) # rounds down
diff=h-hint
drlv=np.sqrt(np.sum(diff*diff,0))
indices = np.compress(drlv < 0.05,
list(range(self.scandata.shape[0])))
print(indices.shape,"hello")
self.scandata = self.allscandata[indices]
npts = 10
for i in range(npts):
for j in range(npts):
x=(i-npts*0.5)*1000/npts
y=(j-npts*0.5)*1000/npts
g.translation[0]=x
g.translation[1]=y
self.compute_gv(g)
for tol in [ 0.1]:
mat = g.ubi.copy()
npks = closest.score_and_refine(mat, self.gv, tol)
# 2nd time with refined
npks = closest.score_and_refine(mat, self.gv, tol)
h=np.dot(mat,np.transpose(self.gv))
hint=np.floor(h+0.5).astype(np.int)
# rounds down
diff=h-hint
drlv=np.sqrt(np.sum(diff*diff,0))
tthscore = np.sum(np.sum(hint*diff) *
np.sum(hint*diff) /
np.sum(h*h))
print(x,y, tol, "%5d"%(npks),sum(drlv)/drlv.shape[0],\
tthscore/drlv.shape[0],indexing.ubitocellpars(mat), end=' ')
print()
sys.stdout.flush()
print()
print()
print()
def refine(self, ubi, quiet=True):
"""
Fit the matrix without changing the peak assignments
"""
mat = ubi.copy()
# print "In refine",self.tolerance, self.gv.shape
# First time fits the mat
self.npks, self.avg_drlv2 = closest.score_and_refine(
mat, self.gv, self.tolerance)
# apply symmetry to mat:
if self.latticesymmetry is not triclinic:
cp = xfab.tools.ubi_to_cell(mat)
U = xfab.tools.ubi_to_u(mat)
mat = xfab.tools.u_to_ubi(U, self.latticesymmetry(cp))
# Second time updates the score with the new mat
self.npks, self.avg_drlv2 = closest.score_and_refine(
mat, self.gv, self.tolerance)
# apply symmetry to mat:
if self.latticesymmetry is not triclinic:
cp = xfab.tools.ubi_to_cell(mat)
U = xfab.tools.ubi_to_u(mat)
mat = xfab.tools.u_to_ubi(U, self.latticesymmetry(cp))
if not quiet:
import math
try:
print("%-8d %.6f" % (self.npks, math.sqrt(self.avg_drlv2)))
except:
print(self.npks, self.avg_drlv2, mat, self.gv.shape,
self.tolerance)
# raise
#print self.tolerance
# self.npks, self.avg_drlv2 = closest.score_and_refine(mat, self.gv,
# self.tolerance)
#tm = indexing.refine(ubi,self.gv,self.tolerance,quiet=quiet)
#print ubi, tm,ubi-tm,mat-tm
return mat
def compute_gv(self, thisgrain, update_columns=False):
"""
Makes self.gv refer be g-vectors computed for this grain in this scan
"""
peaks_xyz = thisgrain.peaks_xyz
om = thisgrain.om
try:
sign = self.parameterobj.parameters['omegasign']
except:
sign = 1.0
# translation should match grain translation here...
self.tth, self.eta = transform.compute_tth_eta_from_xyz(
peaks_xyz, omega=om * sign, **self.parameterobj.parameters)
gv = transform.compute_g_vectors(
self.tth, self.eta, om * sign,
float(self.parameterobj.parameters['wavelength']),
self.parameterobj.parameters['wedge'],
self.parameterobj.parameters['chi'])
# update tth_per_grain and eta_per_grain
if update_columns:
name = thisgrain.name.split(":")[1]
numpy.put(self.scandata[name].tth_per_grain, thisgrain.ind,
self.tth)
numpy.put(self.scandata[name].eta_per_grain, thisgrain.ind,
self.eta)
if not self.OMEGA_FLOAT:
self.gv = gv.T
if self.OMEGA_FLOAT:
mat = thisgrain.ubi.copy()
junk = closest.score_and_refine(mat, gv.T, self.tolerance)
hklf = numpy.dot(mat, gv)
hkli = numpy.floor(hklf + 0.5)
gcalc = numpy.dot(numpy.linalg.inv(mat), hkli)
tth, [eta1,
eta2], [omega1, omega2] = transform.uncompute_g_vectors(
gcalc, float(self.parameterobj.parameters['wavelength']),
self.parameterobj.parameters['wedge'],
self.parameterobj.parameters['chi'])
e1e = numpy.abs(eta1 - self.eta)
e2e = numpy.abs(eta2 - self.eta)
try:
eta_err = numpy.array([e1e, e2e])
except:
print e1e.shape, e2e.shape, e1e
raise
best_fitting = numpy.argmin(eta_err, axis=0)
# These are always 1 or zero
# pick the right omega (confuddled by take here)
omega_calc = best_fitting * omega2 + (1 - best_fitting) * omega1
# Take a weighted average within the omega error of the observed
omerr = (om * sign - omega_calc)
# print omerr[0:5]
omega_calc = om * sign - numpy.clip(omerr, -self.slop, self.slop)
# print omega_calc[0], om[0]
# Now recompute with improved omegas... (tth, eta do not change much)
#self.tth, self.eta = transform.compute_tth_eta(
# numpy.array([x, y]),
# omega = omega_calc,
# **self.parameterobj.parameters)
self.tth, self.eta = transform.compute_tth_eta_from_xyz(
peaks_xyz, omega=om * sign, **self.parameterobj.parameters)
gv = transform.compute_g_vectors(
self.tth, self.eta, omega_calc,
float(self.parameterobj.parameters['wavelength']),
self.parameterobj.parameters['wedge'],
self.parameterobj.parameters['chi'])
self.gv = gv.T
return