def makeTable(G, trunc=False):
'''
'''
data = readData(G)
groupMask, counts, mask, extraName = getGroupBounds(G, 3)
maxOrder = max(getGroupBounds(G, 3)[1].values())
order = 2
table = np.zeros((11, 11))
# Load the reference
table[0, 0] = data['k']['ref']
table[1:, 0] = data['dens']['ref']
for homog in range(5):
# Get the non-energy truncated results
table[0, homog + 1] = data['k']['klt_combine'][homog][maxOrder - 1]
table[1:, homog + 1] = data['dens']['klt_combine'][homog][maxOrder - 1]
# Get the energy truncated results
table[0, homog + 6] = data['k']['klt_combine'][homog][order]
table[1:, homog + 6] = data['dens']['klt_combine'][homog][order]
table = table[:, [0, 6, 7, 8, 9, 10]] if trunc else table[:,
[0, 2, 3, 4, 5]]
label = ['$k_{\\text{eff}}$', 'Cell {}']
print '\nData table for Truncation={}'.format(trunc)
for i, row in enumerate(table):
head = label[0] if i == 0 else label[1].format(i)
print head + '&' + ' & '.join(['{: 4.3f}'.format(r)
for r in row]) + '\\\\'
if i == 0:
print '\\hline'
def readData(G):
'''
Reads data from the data directory that has been make via Unotran
Returns the % relative error for both k and pin-cell fission density
'''
groupMask, counts, mask, extraName = getGroupBounds(G, 3)
maxOrder = max(getGroupBounds(G, 3)[1].values())
data = {}
for d in ['dens', 'k']:
# Load the reference
try:
ref = np.load('reference/ref_{}_{}_{}.npy'.format(d, 'full', G))
if d == 'dens':
# Normalize the reference
ref /= np.linalg.norm(ref)
# Get the density for each pin
ref = np.sum(ref.reshape(10, -1), axis=1)
ref /= np.mean(ref)
except IOError:
raise IOError('Missing reference for {} with {} groups'.format(
d, G))
data[d] = {}
# Add the reference to the data
data[d]['ref'] = ref
for basis in [
'dlp', 'klt_full', 'klt_combine', 'klt_uo2', 'klt_mox',
'klt_pins_full'
]:
data[d][basis] = {}
for homog in range(5):
data[d][basis][homog] = []
for o in range(maxOrder):
# Load the data
try:
sol = np.load(
'data/dgm_{}_{}_{}_g{}_c{}_o{}_h{}.npy'.format(
d, basis, 'full', G, 3, o, homog))
if d == 'dens':
sol /= np.linalg.norm(sol)
sol = np.sum(sol.reshape(10, -1), axis=1)
sol /= np.mean(sol)
except IOError as e:
print('Missing data/dgm_{}_{}_{}_g{}_c{}_o{}_h{}.npy'.
format(d, basis, 'full', G, 3, o, homog))
sol = np.zeros(ref.shape)
# Compute the error
err = (sol - ref) / ref * 100
data[d][basis][homog].append(err)
return data
def plot(A, basisType, contig, bName):
colors = ['b', 'g', 'm']
plt.clf()
G = A.shape[1]
groupMask, counts, mask, extraName = getGroupBounds(G, contig)
bounds = [0]
for i, a in enumerate(A):
ming = 0
for CG, nGroups in counts.items():
maxg = ming + nGroups
plt.plot(range(ming, maxg), a[ming:maxg], c=colors[i], label='order {}'.format(i))
bounds.append(maxg)
ming += nGroups
bounds = np.array(bounds)
plt.vlines(bounds[1:-1] - 0.5, -1, 1)
plt.xlim([0, G - 1])
plt.ylim([-1, 1])
plt.xlabel('Energy group')
plt.ylabel('Normalized basis')
handles, labels = plt.gca().get_legend_handles_labels()
by_label = OrderedDict(zip(labels, handles))
legend = plt.legend(by_label.values(), by_label.keys(), loc='upper center', ncol=3, fancybox=True, framealpha=0.0, bbox_to_anchor=(0.5, 1.1))
plt.savefig('{}.png'.format(bName), transparent=True, additional_artists=[legend])
return
def makeBasis(G, contig):
# Get the coarse group bounds
groupMask, counts, mask, extraName = getGroupBounds(G, contig)
eName = ('_' + extraName) if extraName is not None else ''
print 'building {0}{2}_{1}g'.format('dlp', G, eName)
print counts
# Initialize the basis lists
basis = np.zeros((G, G))
# Compute the basis for each coarse group
for group, order in counts.items():
# Get the DLP basis for the given order
A = DLP(order)
# Get the mask for the currect group
m = groupMask == group
#plt.plot(A[np.argsort(mask[m]), :3])
# plt.show()
# Slice into the basis with the current group
basis[np.ix_(m, m)] = A[np.argsort(mask[m])]
testBasis(basis)
# Save the basis to file
bName = 'basis/{0}{2}_{1}g'.format('dlp', G, eName)
np.savetxt(bName, basis)
plotBasis(G, bName)
def makePlots(G, pType='phi'):
'''
'''
groupMask, counts, mask, extraName = getGroupBounds(G, 3)
maxOrder = max(getGroupBounds(G, 3)[1].values())
lowOrder = 2
fm, cm, mm, bounds = setGeometry('full')
dx = []
for i, c in enumerate(cm):
if i == len(cm) - 1:
break
for x in np.linspace(c, cm[i + 1], fm[i] + 1)[:-1]:
dx.append(x)
dx.append(cm[-1])
dx = np.array(dx)
x = 0.5 * (dx[1:] + dx[:-1])
labels = ['Case 1', 'Case 2', 'Case 3', 'Case 4']
ls = [':', '-', '-.', '--']
for order in [lowOrder, maxOrder - 1]:
for group in (range(G) if pType == 'phi' else [0]):
ref = np.load('reference/ref_{}_{}_{}.npy'.format(
pType, 'full', G))
if pType == 'phi':
ref = ref[0, group]
ref /= np.linalg.norm(ref)
plt.plot(x, ref, label='Reference')
for homog in range(1, 5):
sol = np.load('data/dgm_{}_{}_{}_g{}_c{}_o{}_h{}.npy'.format(
pType, 'klt_combine', 'full', G, 3, order, homog))
if pType == 'phi':
sol = sol[0, group]
sol /= np.linalg.norm(sol)
plt.plot(x, sol, ls=ls[homog - 1], label=labels[homog - 1])
plt.legend(fancybox=True, framealpha=0.0)
plt.xlabel('length [cm]')
plt.ylabel('normalized scalar flux' if pType ==
'phi' else 'normalized fission density')
plt.xlim([0.0, 12.8])
plt.savefig('plots/{}_g{}_o{}.png'.format(pType, group, order),
transparent=True)
plt.clf()
def getInfo(task):
Gs = [44, 238, 1968]
item = 0
for G in Gs:
makeSnapPlot = True
for contig in [3]:
# makePlot(G)
for basisType in ['uo2', 'mox', 'pins_full', 'combine', 'full', 'pins_core1', 'pins_core2', 'assay_12', 'assay_13', 'assay_all', 'core1', 'core2']:
if makeSnapPlot:
makePlot(G)
makeSnapePlot = False
if item == task:
x = getGroupBounds(G, contig)[1]
print x, np.cumsum(x.values())
return G, basisType, contig
else:
item += 1
def setGroup(G, contig, order=None, basisType='dlp'):
groupBounds, counts, mask, bName = getGroupBounds(G, contig)
pydgm.control.xs_name = 'XS/{}gXS.anlxs'.format(G).ljust(256)
print 'using basis basis/{}_{}_{}g'.format(basisType, bName, G)
pydgm.control.dgm_basis_name = 'basis/{}_{}_{}g'.format(
basisType, bName, G).ljust(256)
if order is not None:
print basisType, G, counts.values(), order, [
min(d - 1, order) for d in counts.values()
]
pydgm.control.truncation_map = [
min(d - 1, order) for d in counts.values()
]
if len(groupBounds) > 0:
pydgm.control.energy_group_map = np.array(groupBounds) + 1
def getInfo(task):
Gs = [44, 238, 1968]
geos = ['full']
contigs = [3]
item = 0
for i, G in enumerate(Gs):
for homogOption in [0, 1, 2, 3, 4]:
for contig in contigs:
maxOrder = max(getGroupBounds(G, contig)[1].values())
for geo in geos:
# Select the basis based on the geometry
if geo == 0:
basisTypes = ['dlp']
elif geo == 'full':
basisTypes = [
'dlp', 'klt_full', 'klt_combine', 'klt_uo2',
'klt_mox', 'klt_pins_full'
]
elif geo == 'core1':
basisTypes = [
'dlp', 'klt_assay_12', 'klt_assay_all',
'klt_core1', 'klt_pins_core1'
]
elif geo == 'core2':
basisTypes = [
'dlp', 'klt_assay_13', 'klt_assay_all',
'klt_core2', 'klt_pins_core2'
]
for j, basisType in enumerate(basisTypes):
for order in range(maxOrder):
if item == task:
return geo, G, basisType, order, contig, homogOption
else:
item += 1
def makeBasis(G, basisType, contig, flat=False):
if basisType == 'full':
name = 'klt_full'
data = getSnapshots(G, 'full')
elif basisType == 'mox':
name = 'klt_mox'
data = getSnapshots(G, 'mox')
elif basisType == 'uo2':
name = 'klt_uo2'
data = getSnapshots(G, 'uo2-1')
elif basisType == 'pins_full':
name = 'klt_pins_full'
data = np.concatenate((getSnapshots(G, 'uo2-1'), getSnapshots(G, 'mox')), axis=1)
elif basisType == 'pins_core1':
name = 'klt_pins_core1'
data = np.concatenate((getSnapshots(G, 'uo2-1'), getSnapshots(G, 'uo2-2'), getSnapshots(G, 'uo2-Gd')), axis=1)
elif basisType == 'pins_core2':
name = 'klt_pins_core2'
data = np.concatenate((getSnapshots(G, 'uo2-1'), getSnapshots(G, 'uo2-2'), getSnapshots(G, 'mox')), axis=1)
elif basisType == 'combine':
name = 'klt_combine'
data = np.concatenate((getSnapshots(G, 'uo2-1'), getSnapshots(G, 'mox'), getSnapshots(G, 'junction')), axis=1)
elif basisType == 'core1':
name = 'klt_core1'
data = getSnapshots(G, 'core1')
elif basisType == 'core2':
name = 'klt_core2'
data = getSnapshots(G, 'core2')
elif basisType == 'assay_12':
name = 'klt_assay_12'
data = np.concatenate((getSnapshots(G, 'assay1'), getSnapshots(G, 'assay2')), axis=1)
elif basisType == 'assay_13':
name = 'klt_assay_13'
data = np.concatenate((getSnapshots(G, 'assay1'), getSnapshots(G, 'assay3')), axis=1)
elif basisType == 'assay_all':
name = 'klt_assay_all'
data = np.concatenate((getSnapshots(G, 'assay1'), getSnapshots(G, 'assay2'), getSnapshots(G, 'assay3')), axis=1)
else:
raise NotImplementedError('The type: {} has not been implemented'.format(basisType))
# Get the coarse group bounds
groupMask, counts, mask, extraName = getGroupBounds(G, contig)
eName = ('_' + extraName) if extraName is not None else ''
# Initialize the basis lists
basis = np.zeros((G, G))
# Compute the basis for each coarse group
for group, order in counts.items():
# Get the mask for the currect group
m = groupMask == group
# Get the DLP basis for the given order
A = KLT(data[m], flat)
# Slice into the basis with the current group
basis[np.ix_(m, m)] = A
bName = 'basis/{0}{2}_{1}g'.format(name, G, eName)
# Save the basis to file
np.savetxt(bName, basis)
plotBasis(G, basisType, contig, bName)
def run():
solver = 'eigen'
geo = 'full'
G = 44
basisType = 'dlp'
contig = 3
order = 12
path = 'data'
# Solve the DGM Problem
setVariables(G, geo, solver, True, contig, basisType, order)
# Setup the variable containers
pydgm.dgmsolver.initialize_dgmsolver()
# Give some dummy phi
pydgm.state.phi = np.arange(np.product(pydgm.state.phi.shape)).reshape(
pydgm.state.phi.shape) + 1
# Get the right shapes from the inputs
nL, nG, nC = pydgm.state.phi.shape
# Compute the flux momenst
pydgm.dgmsolver.compute_flux_moments()
# Get the flux moments and shape
m_phi = pydgm.state.mg_phi[0].T
nCG = m_phi.shape[1]
# Compute the test cross sections
pydgm.dgmsolver.compute_xs_moments()
sig_t_test = pydgm.state.mg_sig_t
# Get the group bounds
groupBounds, counts, mask, bName = getGroupBounds(G, contig)
nO = np.max(counts.values())
bounds = [0]
for o in range(np.max(counts.keys()) + 1):
bounds.append(bounds[o] + counts[o])
# Load the basis
P = np.loadtxt('basis/{}_{}_{}g'.format(basisType, bName, G))
# Get the fine group flux
phi = pydgm.state.phi[0, :, :]
nMat = np.max(pydgm.mesh.mmap)
# Create the XS containers
ex_sig_t = np.zeros((nMat, nCG, nO))
m_sig_t = np.zeros((nC, nCG, nO))
# Loop over the spatial cells
for m in range(nMat):
# Get the fine XS
sig_t = pydgm.material.sig_t[:, m]
# Loop over the energy group
for cg in range(nCG):
P_G = P.T[bounds[cg]:bounds[cg + 1]]
for gp in range(bounds[cg], bounds[cg + 1]):
for o, p in enumerate(P_G):
ex_sig_t[m, cg, o] += p[gp] * sig_t[gp] * P_G[0, gp]
print ex_sig_t.shape
print m_phi.shape
for c in range(nC):
pass
exit()
m_sig_t[j] = m_sig_t.dot(m_phi) / m_phi[0]
print m_phi
print m_sig_t.T
print sig_t_test
