def _coord_extract_new(self,path_to_folder,abs_path,tgpost,ioflg):
full_path = misc_utils.check_paths(path_to_folder,'0_log_monitors',
'.')
if not abs_path:
x_coord_file = os.path.abspath(os.path.join(full_path,'CHK_COORD_XND.dat'))
y_coord_file = os.path.abspath(os.path.join(full_path,'CHK_COORD_YND.dat'))
z_coord_file = os.path.abspath(os.path.join(full_path,'CHK_COORD_ZND.dat'))
else:
x_coord_file = os.path.join(full_path,'CHK_COORD_XND.dat')
y_coord_file = os.path.join(full_path,'CHK_COORD_YND.dat')
z_coord_file = os.path.join(full_path,'CHK_COORD_ZND.dat')
#===================================================================
#Extracting XND from the .dat file
file=open(x_coord_file,'rb')
x_coord=np.loadtxt(file,comments='#',usecols=1)
if tgpost and not ioflg:
XND = x_coord[:-1]
else:
index = int(self.metaDF['NCL1_tg']) + 1
if (tgpost and ioflg):
XND = x_coord[:index]
XND -= XND[0]
elif self.metaDF['iCase'] == 5:
index = int(self.metaDF['NCL1_io']) + 1
XND = x_coord[:index]
else:
XND = x_coord[index:]
file.close()
#===========================================================
#Extracting YCC from the .dat file
file=open(y_coord_file,'rb')
y_coord=np.loadtxt(file,usecols=1,skiprows=1)
index = int(self.metaDF['NCL2']) + 1
YCC=y_coord[index:]
YND = y_coord[:index]
y_size = YCC.size
file.close()
#============================================================
file=open(z_coord_file,'rb')
ZND=np.loadtxt(file,comments='#',usecols=1)
#============================================================
XCC, ZCC = self._coord_interp(XND,ZND)
z_size = ZCC.size
x_size = XCC.size
y_size = YCC.size
file.close()
CoordDF = cd.datastruct({'x':XCC,'y':YCC,'z':ZCC})
Coord_ND_DF = cd.datastruct({'x':XND,'y':YND,'z':ZND})
NCL = [x_size, y_size, z_size]
return CoordDF, Coord_ND_DF, NCL
def _quad_extract(self,
h_list,
path_to_folder='.',
time0=None,
abs_path=True):
times = misc_utils.time_extract(path_to_folder, abs_path)
if time0 is not None:
times = list(filter(lambda x: x > time0, times))
if cp.rcParams['TEST']:
times.sort()
times = times[-3:]
self._meta_data = self._module._meta_class(path_to_folder, abs_path)
self.NCL = self._meta_data.NCL
try:
self._avg_data = self._module._avg_io_class(
max(times), self._meta_data, path_to_folder, time0, abs_path)
except Exception:
times.remove(max(times))
self._avg_data = self._module._avg_io_class(
max(times), self._meta_data, path_to_folder, time0)
i = 1
for timing in times:
fluct_data = self._module._fluct_io_class(
timing,
self._avg_data,
time0=time0,
path_to_folder=path_to_folder,
abs_path=abs_path)
fluct_uv, quadrant_array = self._quadrant_extract(
fluct_data.fluctDF, timing, self.NCL)
coe3 = (i - 1) / i
coe2 = 1 / i
if i == 1:
quad_anal_array = self._quad_calc(self._avg_data, fluct_uv,
quadrant_array, self.NCL,
h_list, timing)
else:
local_quad_anal_array = self._quad_calc(
self._avg_data, fluct_uv, quadrant_array, self.NCL, h_list,
timing)
assert local_quad_anal_array.shape == quad_anal_array.shape, "shape of previous array (%d,%d) " % quad_anal_array.shape\
+ " and current array (%d,%d) must be the same" % local_quad_anal_array.shape
quad_anal_array = quad_anal_array * coe3 + local_quad_anal_array * coe2
gc.collect()
i += 1
index = [[], []]
for h in h_list:
index[0].extend([h] * 4)
index[1] = [1, 2, 3, 4] * len(h_list)
self.shape = self._avg_data.shape
self.QuadAnalDF = cd.datastruct(quad_anal_array, index=index)
def _POD_extract(self,
comp,
path_to_folder='.',
method='svd',
low_memory=True,
abs_path=True,
time0=None,
subdomain=None,
nsnapshots=100,
nmodes=10):
max_time = misc_utils.max_time_calc(path_to_folder, abs_path)
self.avg_data = self._module._avg_class(max_time,
path_to_folder=path_to_folder,
abs_path=abs_path,
time0=time0)
if subdomain is not None:
self._coorddata.create_subdomain(**subdomain)
self._meta_data = self.avg_data._meta_data
times = misc_utils.time_extract(path_to_folder, abs_path)
if time0:
times = list(filter(lambda x: x > time0, times))
times = times[-nsnapshots:]
if rcParams['TEST']:
times = times[-7:]
nmodes = 7
if nmodes > len(times):
nmodes = len(times)
if method.lower() == "svd":
PODmodes, self._eig_values = self._performSVD(
comp, path_to_folder, abs_path, self.avg_data, times, nmodes,
subdomain)
elif method.lower() == "snapshots":
PODmodes, self._eig_values = self._performSnapShots(
comp, path_to_folder, abs_path, low_memory, self.avg_data,
times, nmodes, subdomain)
else:
msg = f"Method selected ({method}) is not valid"
raise ValueError(msg)
index = [[None] * len(comp), list(comp)]
self.POD_modesDF = cd.datastruct(PODmodes, index=index)
def _budget_extract(self, PhyTime):
laminar = self._laminar_extract(PhyTime)
turbulent = self._turbulent_extract(PhyTime)
inertia = self._inertia_extract(PhyTime)
array_concat = [laminar, turbulent, inertia]
budget_array = np.stack(array_concat, axis=0)
budget_index = ['laminar', 'turbulent', 'non-homogeneous']
phystring_index = [PhyTime] * 3
budgetDF = cd.datastruct(budget_array,
index=[phystring_index, budget_index])
return budgetDF
def _quad_extract(self,
h_list,
path_to_folder='.',
time0=None,
abs_path=True):
times = misc_utils.time_extract(path_to_folder, abs_path)
if time0 is not None:
times = list(filter(lambda x: x > time0, times))
if cp.rcParams['TEST']:
times.sort()
times = times[-3:]
meta_data = self._module._meta_class(path_to_folder, abs_path)
NCL = meta_data.NCL
avg_data = self._module._avg_tg_base_class(
times,
meta_data=meta_data,
path_to_folder=path_to_folder,
time0=time0,
abs_path=abs_path)
for timing in times:
fluct_data = self._module._fluct_tg_class(
timing,
avg_data,
path_to_folder=path_to_folder,
abs_path=abs_path)
fluct_uv, quadrant_array = self._quadrant_extract(
fluct_data.fluctDF, timing, NCL)
if 'quad_anal_array' not in locals():
quad_anal_array = self._quad_calc(avg_data, fluct_uv,
quadrant_array, NCL, h_list,
timing)
else:
local_quad_anal_array = self._quad_calc(
avg_data, fluct_uv, quadrant_array, NCL, h_list, timing)
quad_anal_array = np.vstack(
[quad_anal_array, local_quad_anal_array])
gc.collect()
index = [[], []]
for h in h_list:
index[0].extend([h] * 4)
index[1] = [1, 2, 3, 4] * len(h_list)
shape = avg_data.shape
QuadAnalDF = cd.datastruct(quad_anal_array, index=index)
return meta_data, NCL, avg_data, QuadAnalDF, shape
def _budget_extract(self, PhyTime, comp):
advection = self._advection_extract(PhyTime, comp)
pressure_grad = self._pressure_grad(PhyTime, comp)
viscous = self._viscous_extract(PhyTime, comp)
Reynolds_stress = self._turb_transport(PhyTime, comp)
array_concat = [advection, pressure_grad, viscous, Reynolds_stress]
budget_array = np.stack(array_concat, axis=0)
budget_index = [
'advection', 'pressure gradient', 'viscous stresses',
'reynolds stresses'
]
phystring_index = [PhyTime] * 4
budgetDF = cd.datastruct(budget_array,
index=[phystring_index, budget_index])
return budgetDF
def _fluctDF_calc(self, inst_data, avg_data):
avg_times = avg_data.get_times()
inst_times = list(set([x[0] for x in inst_data.InstDF.index]))
u_comp = [x[1] for x in avg_data.flow_AVGDF.index]
indices = inst_data.InstDF.index
fluct = np.zeros((len(indices), *inst_data.shape))
if len(avg_times) == 1:
j = 0
for time in inst_times:
avg_index = avg_data._return_index(time)
for comp in u_comp:
avg_values = avg_data.flow_AVGDF[None, comp]
inst_values = inst_data.InstDF[time, comp]
for i in range(inst_data.shape[0]):
for k in range(inst_data.shape[2]):
fluct[j, i, :,
k] = inst_values[i, :,
k] - avg_values[:,
avg_index]
j += 1
elif all(time in avg_times for time in inst_times):
for j, index in enumerate(indices):
avg_index = avg_data._return_index(index[0])
avg_values = avg_data.flow_AVGDF[None, index[1]]
inst_values = inst_data.InstDF[index]
for i in range(inst_data.shape[0]):
for k in range(inst_data.shape[2]):
fluct[j, i, :,
k] = inst_values[i, :, k] - avg_values[:,
avg_index]
else:
raise ValueError(
"avg_data must either be length 1 or the same length as inst_data"
)
# DF_shape = (len(indices),np.prod(inst_data.shape))
return cd.datastruct(
fluct, index=inst_data.InstDF.index) #.data(inst_data.shape)
def _budget_extract(self, PhyTime, comp1, comp2):
production = self._production_extract(PhyTime, comp1, comp2)
advection = self._advection_extract(PhyTime, comp1, comp2)
turb_transport = self._turb_transport(PhyTime, comp1, comp2)
pressure_diffusion = self._pressure_diffusion(PhyTime, comp1, comp2)
pressure_strain = self._pressure_strain(PhyTime, comp1, comp2)
viscous_diff = self._viscous_diff(PhyTime, comp1, comp2)
dissipation = self._dissipation_extract(PhyTime, comp1, comp2)
array_concat = [production,advection,turb_transport,pressure_diffusion,\
pressure_strain,viscous_diff,dissipation]
budget_array = np.stack(array_concat, axis=0)
budget_index = ['production','advection','turbulent transport','pressure diffusion',\
'pressure strain','viscous diffusion','dissipation']
phystring_index = [PhyTime] * 7
budgetDF = cd.datastruct(budget_array,
index=[phystring_index, budget_index])
return budgetDF
def _create_uniform(self, PhyTime):
advection = self.budgetDF[PhyTime, 'advection']
centre_index = int(0.5 * advection.shape[0])
advection_centre = advection[centre_index]
uniform_bf = self.budgetDF[PhyTime,
'pressure gradient'] + advection_centre
non_uniform_bf = advection - advection_centre
times = self.budgetDF.outer_index
for time in times:
key1 = (time, 'advection')
key2 = (time, 'pressure gradient')
del self.budgetDF[key1]
del self.budgetDF[key2]
cal_dict = {
(PhyTime, 'uniform'): uniform_bf,
(PhyTime, 'non-uniform'): non_uniform_bf
}
index = [[PhyTime] * 2, ['uniform', 'non-uniform']]
dstruct = cd.datastruct(np.array([uniform_bf, non_uniform_bf]),
index=index)
self.budgetDF.concat(dstruct)
def _coord_extract_old(self,path_to_folder,abs_path,tgpost,ioflg):
""" Function to extract the coordinates from their .dat file """
full_path = misc_utils.check_paths(path_to_folder,'0_log_monitors',
'.')
if not abs_path:
x_coord_file = os.path.abspath(os.path.join(full_path,'CHK_COORD_XND.dat'))
y_coord_file = os.path.abspath(os.path.join(full_path,'CHK_COORD_YND.dat'))
z_coord_file = os.path.abspath(os.path.join(full_path,'CHK_COORD_ZND.dat'))
else:
x_coord_file = os.path.join(full_path,'CHK_COORD_XND.dat')
y_coord_file = os.path.join(full_path,'CHK_COORD_YND.dat')
z_coord_file = os.path.join(full_path,'CHK_COORD_ZND.dat')
#===================================================================
#Extracting XND from the .dat file
file=open(x_coord_file,'rb')
x_coord=np.loadtxt(file,comments='#')
x_size= int(x_coord[0])
x_coord=np.delete(x_coord,0)
if tgpost and not ioflg:
XND = x_coord[:-1]
else:
for i in range(x_size):
if x_coord[i] == 0.0:
index=i
break
if tgpost and ioflg:
XND = x_coord[:index+1]
XND -= XND[0]
else:
XND = x_coord[index+1:]
file.close()
#===========================================================
#Extracting YCC from the .dat file
file=open(y_coord_file,'rb')
y_coord=np.loadtxt(file,comments='#',usecols=1)
y_size = int(y_coord[0])
for i in range(y_coord.size):
if y_coord[i] == 1.0:
index=i
break
# YCC=np.delete(y_coord,np.arange(index+1))
YND = y_coord[:index+1]
YCC = y_coord[(index+1):]
y_size = YCC.size
file.close()
#============================================================
file=open(z_coord_file,'rb')
z_coord=np.loadtxt(file,comments='#')
z_size = int(z_coord[0])
ZND=np.delete(z_coord,0)
#============================================================
XCC, ZCC = self._coord_interp(XND,ZND)
z_size = ZCC.size
x_size = XCC.size
y_size = YCC.size
file.close()
CoordDF = cd.datastruct({'x':XCC,'y':YCC,'z':ZCC})
Coord_ND_DF = cd.datastruct({'x':XND,'y':YND,'z':ZND})
NCL = [x_size, y_size, z_size]
return CoordDF, Coord_ND_DF, NCL
def _extract_fluct(self,
x,
y,
path_to_folder=None,
time0=None,
gridsize=200,
y_mode='half-channel',
use_ini=True,
xy_inner=True,
tgpost=False,
abs_path=True):
times = misc_utils.time_extract(path_to_folder, abs_path)
if time0 is not None:
times = list(filter(lambda x: x > time0, times))
if cp.rcParams['TEST']:
times.sort()
times = times[-5:]
self._meta_data = self._module._meta_class(path_to_folder, abs_path)
try:
self.avg_data = self._module._avg_io_class(max(times),
self._meta_data,
path_to_folder, time0,
abs_path)
except Exception:
times.remove(max(times))
self.avg_data = self._module._avg_io_class(max(times),
self._meta_data,
path_to_folder, time0)
if xy_inner:
if len(x) != len(y):
msg = "length of x coordinate array must be same"+\
" as the y coord array. Lengths provided %d (x),"%len(x)+\
" %d (y)"%len(y)
raise ValueError(msg)
x_coord_list = x
y_coord_list = y
else:
x_coord_list = []
y_coord_list = []
for x_val in x:
for y_val in y:
x_coord_list.append(x_val)
y_coord_list.append(y_val)
x_index = indexing.coord_index_calc(self.avg_data.CoordDF, 'x',
x_coord_list)
self._x_loc_norm = x_coord_list if not use_ini else [0] * len(
y_coord_list)
y_index = indexing.y_coord_index_norm(self.avg_data, y_coord_list,
self._x_loc_norm, y_mode)
y_index = np.diag(np.array(y_index))
u_prime_array = [[] for _ in range(len(y_index))]
v_prime_array = [[] for _ in range(len(y_index))]
for time in times:
fluct_data = self._module._fluct_io_class(time, self.avg_data,
path_to_folder, abs_path)
u_prime_data = fluct_data.fluctDF[time, 'u']
v_prime_data = fluct_data.fluctDF[time, 'v']
for i in range(len(y_index)):
u_prime_array[i].extend(u_prime_data[:, y_index[i],
x_index[i]])
v_prime_array[i].extend(v_prime_data[:, y_index[i],
x_index[i]])
if cp.rcParams['SymmetryAVG'] and self.metaDF['iCase'] == 1:
y_size = u_prime_data.shape[1]
u_prime_array[i].extend(u_prime_data[:, -1 - y_index[i],
x_index[i]])
v_prime_array[i].extend(
-1 * v_prime_data[:, -1 - y_index[i], x_index[i]])
# del fluct_data#; del u_prime_data; del v_prime_data
gc.collect()
pdf_array = [[] for _ in range(len(y_index))]
u_array = [[] for _ in range(len(y_index))]
v_array = [[] for _ in range(len(y_index))]
estimator = seaborn._statistics.KDE(gridsize=gridsize)
for i, y in enumerate(y_index):
pdf_array[i], (u_array[i],
v_array[i]) = estimator(np.array(u_prime_array[i]),
np.array(v_prime_array[i]))
# ax = seaborn.kdeplot(u_prime_array[i],v_prime_array[i],gridsize=gridsize)
# for artist in ax.get_children():
# if isinstance(artist,mpl.contour.QuadContourSet):
index = list(zip(x_coord_list, y_coord_list))
pdf_array = np.array(pdf_array)
u_array = np.array(u_array)
v_array = np.array(v_array)
self._y_mode = y_mode
self.pdf_arrayDF = cd.datastruct(pdf_array, index=index)
self.u_arrayDF = cd.datastruct(u_array, index=index)
self.v_arrayDF = cd.datastruct(v_array, index=index)
def _autocov_extract(self,
comp1,
comp2,
path_to_folder='.',
time0=None,
abs_path=True,
max_x_sep=None,
max_z_sep=None):
times = misc_utils.time_extract(path_to_folder, abs_path)
if time0 is not None:
times = list(filter(lambda x: x > time0, times))
if cp.rcParams['TEST']:
times.sort()
times = times[-3:]
self._meta_data = self._module.CHAPSim_meta(path_to_folder)
self.comp = (comp1, comp2)
self.NCL = self._meta_data.NCL
self._avg_data = self._module._avg_tg_base_class(
times,
meta_data=self._meta_data,
path_to_folder=path_to_folder,
time0=time0,
abs_path=abs_path)
if max_z_sep is None:
max_z_sep = int(self.NCL[2] / 2)
elif max_z_sep > self.NCL[2]:
raise ValueError(
"\033[1;32 Variable max_z_sep must be less than half NCL3 in readdata file\n"
)
if max_x_sep is None:
max_x_sep = int(self._NCL[0] / 2)
elif max_x_sep > self.NCL[0]:
raise ValueError(
"\033[1;32 Variable max_x_sep must be less than half NCL3 in readdata file\n"
)
# self.shape_x = (max_x_sep,self.NCL[1],len(times))
# self.shape_z = (max_z_sep,self.NCL[1],len(times))
shape_x = (max_x_sep, self.NCL[1], len(times))
shape_z = (max_z_sep, self.NCL[1], len(times))
for timing in times:
fluct_data = self._module._fluct_tg_class(
timing,
self._avg_data,
path_to_folder=path_to_folder,
abs_path=abs_path)
if 'R_z' not in locals():
R_z, R_x = self._autocov_calc(fluct_data, comp1, comp2, timing,
max_x_sep, max_z_sep)
else:
local_R_z, local_R_x = self._autocov_calc(
fluct_data, comp1, comp2, timing, max_x_sep, max_z_sep)
R_z = np.vstack([R_z, local_R_z])
R_x = np.vstack([R_x, local_R_x])
gc.collect()
R_z = R_z.T.reshape(shape_z)
R_x = R_x.T.reshape(shape_x)
self.autocorrDF = cd.datastruct({'x': R_x, 'z': R_z})
def _autocov_extract(self,
comp1,
comp2,
path_to_folder=".",
time0=None,
abs_path=True,
max_x_sep=None,
max_z_sep=None):
times = misc_utils.time_extract(path_to_folder, abs_path)
if time0 is not None:
times = list(filter(lambda x: x > time0, times))
if cp.rcParams['TEST']:
times.sort()
times = times[-5:]
self._meta_data = self._module._meta_class(path_to_folder)
self.comp = (comp1, comp2)
try:
self._avg_data = self._module._avg_io_class(
max(times), self._meta_data, path_to_folder, time0, abs_path)
except Exception:
times.remove(max(times))
self._avg_data = self._module._avg_io_class(
max(times), self._meta_data, path_to_folder, time0)
if max_z_sep is None:
max_z_sep = int(self.NCL[2] * 0.5)
elif max_z_sep > self.NCL[2]:
raise ValueError(
"Variable max_z_sep must be less than half NCL3 in readdata file\n"
)
if max_x_sep is None:
max_x_sep = int(self.NCL[0] * 0.5)
elif max_x_sep > self.NCL[0]:
raise ValueError(
"Variable max_x_sep must be less than half NCL3 in readdata file\n"
)
# self.shape_x = (max_x_sep,self.NCL[1],self.NCL[0]-max_x_sep)
# self.shape_z = (max_z_sep,self.NCL[1],self.NCL[0])
for i, timing in enumerate(times):
fluct_data = self._module._fluct_io_class(
timing,
self._avg_data,
time0=time0,
path_to_folder=path_to_folder,
abs_path=abs_path)
coe3 = i / (i + 1)
coe2 = 1 / (i + 1)
if i == 0:
R_x, R_z = self._autocov_calc(fluct_data, comp1, comp2, timing,
max_x_sep, max_z_sep)
else:
local_R_x, local_R_z = self._autocov_calc(
fluct_data, comp1, comp2, timing, max_x_sep, max_z_sep)
if R_x.shape != local_R_x.shape or R_z.shape != local_R_z.shape:
msg = "There is a problem. the shapes of the local and averaged array are different"
raise ValueError(msg)
R_x = R_x * coe3 + local_R_x * coe2
R_z = R_z * coe3 + local_R_z * coe2
if cp.rcParams['SymmetryAVG'] and self.metaDF['iCase'] == 1:
vy_count = comp1.count('v') + comp2.count('v')
R_x = 0.5 * (R_x + R_x[:, ::-1] * (-1)**vy_count)
R_z = 0.5 * (R_z + R_z[:, ::-1] * (-1)**vy_count)
self.autocorrDF = cd.datastruct({
'x': R_x,
'z': R_z
}) #.data([shape_x,shape_z])
