2019-09-12 06:33:19 +05:30
|
|
|
from queue import Queue
|
2019-04-13 14:41:32 +05:30
|
|
|
import re
|
2019-09-14 21:22:07 +05:30
|
|
|
import glob
|
2019-09-12 06:33:19 +05:30
|
|
|
|
|
|
|
import h5py
|
2019-04-17 23:27:16 +05:30
|
|
|
import numpy as np
|
2019-09-12 06:33:19 +05:30
|
|
|
|
2019-05-20 23:24:57 +05:30
|
|
|
from . import util
|
2019-04-13 14:41:32 +05:30
|
|
|
|
|
|
|
# ------------------------------------------------------------------
|
|
|
|
class DADF5():
|
2019-09-12 06:33:19 +05:30
|
|
|
"""
|
|
|
|
Read and write to DADF5 files.
|
|
|
|
|
|
|
|
DADF5 files contain DAMASK results.
|
|
|
|
"""
|
2019-04-13 14:41:32 +05:30
|
|
|
|
|
|
|
# ------------------------------------------------------------------
|
2019-09-14 04:36:47 +05:30
|
|
|
def __init__(self,filename):
|
2019-09-12 06:33:19 +05:30
|
|
|
"""
|
|
|
|
Opens an existing DADF5 file.
|
|
|
|
|
|
|
|
Parameters
|
|
|
|
----------
|
|
|
|
filename : str
|
|
|
|
name of the DADF5 file to be openend.
|
2019-04-13 14:41:32 +05:30
|
|
|
|
2019-09-12 06:33:19 +05:30
|
|
|
"""
|
2019-04-13 14:41:32 +05:30
|
|
|
with h5py.File(filename,'r') as f:
|
|
|
|
|
2019-05-03 10:16:22 +05:30
|
|
|
if f.attrs['DADF5-major'] != 0 or f.attrs['DADF5-minor'] != 2:
|
2019-04-17 23:27:16 +05:30
|
|
|
raise TypeError('Unsupported DADF5 version {} '.format(f.attrs['DADF5-version']))
|
2019-04-13 14:41:32 +05:30
|
|
|
|
2019-05-03 10:16:22 +05:30
|
|
|
self.structured = 'grid' in f['geometry'].attrs.keys()
|
2019-04-13 14:41:32 +05:30
|
|
|
|
|
|
|
if self.structured:
|
2019-05-03 10:16:22 +05:30
|
|
|
self.grid = f['geometry'].attrs['grid']
|
|
|
|
self.size = f['geometry'].attrs['size']
|
2019-04-13 14:41:32 +05:30
|
|
|
|
|
|
|
r=re.compile('inc[0-9]+')
|
2019-09-16 08:34:52 +05:30
|
|
|
self.increments = [i for i in f.keys() if r.match(i)]
|
|
|
|
self.times = [round(f[i].attrs['time/s'],12) for i in self.increments]
|
|
|
|
|
2019-09-14 03:47:46 +05:30
|
|
|
self.Nmaterialpoints, self.Nconstituents = np.shape(f['mapping/cellResults/constituent'])
|
|
|
|
self.materialpoints = [m.decode() for m in np.unique(f['mapping/cellResults/materialpoint']['Name'])]
|
|
|
|
self.constituents = [c.decode() for c in np.unique(f['mapping/cellResults/constituent'] ['Name'])]
|
|
|
|
|
2019-09-16 03:32:16 +05:30
|
|
|
self.con_physics = []
|
2019-04-18 15:28:17 +05:30
|
|
|
for c in self.constituents:
|
2019-09-16 23:33:02 +05:30
|
|
|
self.con_physics += f['/'.join([self.increments[0],'constituent',c])].keys()
|
2019-09-16 03:32:16 +05:30
|
|
|
self.con_physics = list(set(self.con_physics)) # make unique
|
2019-05-16 03:02:23 +05:30
|
|
|
|
2019-09-16 03:32:16 +05:30
|
|
|
self.mat_physics = []
|
2019-05-16 03:02:23 +05:30
|
|
|
for m in self.materialpoints:
|
2019-09-16 23:33:02 +05:30
|
|
|
self.mat_physics += f['/'.join([self.increments[0],'materialpoint',m])].keys()
|
2019-09-16 03:32:16 +05:30
|
|
|
self.mat_physics = list(set(self.mat_physics)) # make unique
|
2019-09-14 03:47:46 +05:30
|
|
|
|
2019-09-16 23:33:02 +05:30
|
|
|
self.visible= {'increments': self.increments,
|
|
|
|
'constituents': self.constituents,
|
|
|
|
'materialpoints': self.materialpoints,
|
|
|
|
'constituent': range(self.Nconstituents), # ToDo: stupid naming
|
|
|
|
'con_physics': self.con_physics,
|
|
|
|
'mat_physics': self.mat_physics}
|
2019-09-14 07:06:06 +05:30
|
|
|
|
|
|
|
self.filename = filename
|
|
|
|
|
|
|
|
|
2019-09-16 04:30:19 +05:30
|
|
|
def __manage_visible(self,datasets,what,action):
|
2019-09-16 08:26:07 +05:30
|
|
|
"""Manages the visibility of the groups."""
|
2019-09-14 22:08:04 +05:30
|
|
|
# allow True/False and string arguments
|
2019-09-16 04:30:19 +05:30
|
|
|
if datasets is True:
|
|
|
|
datasets = ['*']
|
|
|
|
elif datasets is False:
|
|
|
|
datasets = []
|
|
|
|
choice = [datasets] if isinstance(datasets,str) else datasets
|
|
|
|
|
|
|
|
valid = [e for e_ in [glob.fnmatch.filter(getattr(self,what) ,s) for s in choice] for e in e_]
|
|
|
|
existing = set(self.visible[what])
|
|
|
|
|
|
|
|
if action == 'set':
|
|
|
|
self.visible[what] = valid
|
|
|
|
elif action == 'add':
|
|
|
|
self.visible[what] = list(existing.union(valid))
|
|
|
|
elif action == 'del':
|
|
|
|
self.visible[what] = list(existing.difference_update(valid))
|
2019-09-16 08:49:14 +05:30
|
|
|
|
|
|
|
|
|
|
|
def __time_to_inc(self,start,end):
|
|
|
|
selected = []
|
|
|
|
for i,time in enumerate(self.times):
|
|
|
|
if start <= time < end:
|
|
|
|
selected.append(self.increments[i])
|
|
|
|
return selected
|
|
|
|
|
|
|
|
|
|
|
|
def set_by_time(self,start,end):
|
|
|
|
self.__manage_visible(self.__time_to_inc(start,end),'increments','set')
|
|
|
|
|
|
|
|
|
|
|
|
def add_by_time(self,start,end):
|
|
|
|
self.__manage_visible(self.__time_to_inc(start,end),'increments','add')
|
|
|
|
|
|
|
|
|
|
|
|
def del_by_time(self,start,end):
|
|
|
|
self.__manage_visible(self.__time_to_inc(start,end),'increments','del')
|
2019-09-16 04:30:19 +05:30
|
|
|
|
|
|
|
|
|
|
|
def iter_visible(self,what):
|
2019-09-16 08:26:07 +05:30
|
|
|
"""Iterates over visible items by setting each one visible."""
|
2019-09-16 04:30:19 +05:30
|
|
|
datasets = self.visible[what]
|
|
|
|
last_datasets = datasets.copy()
|
|
|
|
for dataset in datasets:
|
|
|
|
if last_datasets != self.visible[what]:
|
|
|
|
self.__manage_visible(datasets,what,'set')
|
2019-09-14 09:44:52 +05:30
|
|
|
raise Exception
|
2019-09-16 04:30:19 +05:30
|
|
|
self.__manage_visible(dataset,what,'set')
|
|
|
|
last_datasets = self.visible[what]
|
|
|
|
yield dataset
|
|
|
|
self.__manage_visible(datasets,what,'set')
|
2019-09-14 09:44:52 +05:30
|
|
|
|
2019-09-14 07:06:06 +05:30
|
|
|
|
2019-09-16 04:30:19 +05:30
|
|
|
def set_visible(self,what,datasets):
|
|
|
|
self.__manage_visible(datasets,what,'set')
|
2019-09-14 07:06:06 +05:30
|
|
|
|
|
|
|
|
2019-09-16 04:30:19 +05:30
|
|
|
def add_visible(self,what,datasets):
|
|
|
|
self.__manage_visible(datasets,what,'add')
|
2019-09-14 07:06:06 +05:30
|
|
|
|
|
|
|
|
2019-09-16 04:30:19 +05:30
|
|
|
def del_visible(self,what,datasets):
|
|
|
|
self.__manage_visible(datasets,what,'del')
|
2019-09-14 03:47:46 +05:30
|
|
|
|
2019-09-14 07:06:06 +05:30
|
|
|
|
2019-09-14 21:22:07 +05:30
|
|
|
def groups_with_datasets(self,datasets):
|
2019-09-12 09:50:14 +05:30
|
|
|
"""
|
|
|
|
Get groups that contain all requested datasets.
|
|
|
|
|
2019-09-14 21:22:07 +05:30
|
|
|
Only groups within inc?????/constituent/*_*/* inc?????/materialpoint/*_*/*
|
|
|
|
are considered as they contain the data.
|
|
|
|
Single strings will be treated as list with one entry.
|
|
|
|
|
|
|
|
Wild card matching is allowed, but the number of arguments need to fit.
|
|
|
|
|
2019-09-12 09:50:14 +05:30
|
|
|
Parameters
|
|
|
|
----------
|
2019-09-14 21:22:07 +05:30
|
|
|
datasets : iterable or str or boolean
|
|
|
|
|
|
|
|
Examples
|
|
|
|
--------
|
|
|
|
datasets = False matches no group
|
|
|
|
datasets = True matches all groups
|
|
|
|
datasets = ['F','P'] matches a group with ['F','P','sigma']
|
2019-09-16 03:40:32 +05:30
|
|
|
datasets = ['*','P'] matches a group with ['F','P']
|
|
|
|
datasets = ['*'] does not match a group with ['F','P','sigma']
|
|
|
|
datasets = ['*','*'] does not match a group with ['F','P','sigma']
|
2019-09-14 21:22:07 +05:30
|
|
|
datasets = ['*','*','*'] matches a group with ['F','P','sigma']
|
|
|
|
|
|
|
|
"""
|
|
|
|
if datasets is False: return []
|
2019-09-16 03:40:32 +05:30
|
|
|
sets = [datasets] if isinstance(datasets,str) else datasets
|
2019-09-12 09:50:14 +05:30
|
|
|
|
2019-05-20 23:24:57 +05:30
|
|
|
groups = []
|
2019-09-14 21:22:07 +05:30
|
|
|
|
2019-05-20 23:24:57 +05:30
|
|
|
with h5py.File(self.filename,'r') as f:
|
2019-09-16 23:33:02 +05:30
|
|
|
for i in self.iter_visible('increments'):
|
|
|
|
for o,p in zip(['constituents','materialpoints'],['con_physics','mat_physics']):
|
|
|
|
for oo in self.iter_visible(o):
|
|
|
|
for pp in self.iter_visible(p):
|
|
|
|
group = '/'.join([i,o[:-1],oo,pp]) # o[:-1]: plural/singular issue
|
|
|
|
if sets is True:
|
|
|
|
groups.append(group)
|
|
|
|
else:
|
|
|
|
match = [e for e_ in [glob.fnmatch.filter(f[group].keys(),s) for s in sets] for e in e_]
|
|
|
|
if len(set(match)) == len(sets) : groups.append(group)
|
2019-05-20 23:24:57 +05:30
|
|
|
return groups
|
|
|
|
|
2019-05-16 03:02:23 +05:30
|
|
|
|
2019-09-16 08:26:07 +05:30
|
|
|
def list_data(self):
|
2019-09-16 23:33:02 +05:30
|
|
|
"""Gices information on all active datasets in the file."""
|
|
|
|
message = ''
|
2019-04-18 15:28:17 +05:30
|
|
|
with h5py.File(self.filename,'r') as f:
|
2019-09-16 23:33:02 +05:30
|
|
|
for i in self.iter_visible('increments'):
|
|
|
|
message+='\n{}\n'.format(i)
|
|
|
|
for o,p in zip(['constituents','materialpoints'],['con_physics','mat_physics']):
|
|
|
|
for oo in self.iter_visible(o):
|
|
|
|
message+=' {}\n'.format(oo)
|
|
|
|
for pp in self.iter_visible(p):
|
|
|
|
message+=' {}\n'.format(pp)
|
|
|
|
group = '/'.join([i,o[:-1],oo,pp]) # o[:-1]: plural/singular issue
|
|
|
|
for d in f[group].keys():
|
|
|
|
try:
|
|
|
|
message+=' {} ({})\n'.format(d,f['/'.join([group,d])].attrs['Description'].decode())
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
return message
|
|
|
|
|
2019-04-17 23:27:16 +05:30
|
|
|
|
2019-09-16 08:26:07 +05:30
|
|
|
def get_dataset_location(self,label):
|
|
|
|
"""Returns the location of all active datasets with given label."""
|
2019-04-17 23:27:16 +05:30
|
|
|
path = []
|
|
|
|
with h5py.File(self.filename,'r') as f:
|
2019-09-16 05:38:03 +05:30
|
|
|
for i in self.iter_visible('increments'):
|
2019-09-16 23:33:02 +05:30
|
|
|
for o,p in zip(['constituents','materialpoints'],['con_physics','mat_physics']):
|
|
|
|
for oo in self.iter_visible(o):
|
|
|
|
for pp in self.iter_visible(p):
|
|
|
|
try:
|
|
|
|
k = '/'.join([i,'constituent',oo,pp,label])
|
|
|
|
f[k]
|
|
|
|
path.append(k)
|
|
|
|
except KeyError as e:
|
|
|
|
print('unable to locate constituents dataset: '+ str(e))
|
2019-04-17 23:27:16 +05:30
|
|
|
return path
|
|
|
|
|
|
|
|
|
|
|
|
def read_dataset(self,path,c):
|
2019-04-18 15:28:17 +05:30
|
|
|
"""
|
2019-09-12 06:33:19 +05:30
|
|
|
Dataset for all points/cells.
|
2019-04-18 15:28:17 +05:30
|
|
|
|
|
|
|
If more than one path is given, the dataset is composed of the individual contributions
|
|
|
|
"""
|
2019-04-17 23:27:16 +05:30
|
|
|
with h5py.File(self.filename,'r') as f:
|
2019-04-18 15:28:17 +05:30
|
|
|
shape = (self.Nmaterialpoints,) + np.shape(f[path[0]])[1:]
|
2019-05-16 13:01:13 +05:30
|
|
|
if len(shape) == 1: shape = shape +(1,)
|
2019-04-17 23:27:16 +05:30
|
|
|
dataset = np.full(shape,np.nan)
|
2019-04-18 15:28:17 +05:30
|
|
|
for pa in path:
|
|
|
|
label = pa.split('/')[2]
|
2019-09-14 04:31:30 +05:30
|
|
|
|
|
|
|
p = np.where(f['mapping/cellResults/constituent'][:,c]['Name'] == str.encode(label))[0]
|
|
|
|
if len(p)>0:
|
2019-05-16 03:57:06 +05:30
|
|
|
u = (f['mapping/cellResults/constituent'][p,c]['Position'])
|
2019-05-16 13:01:13 +05:30
|
|
|
a = np.array(f[pa])
|
|
|
|
if len(a.shape) == 1:
|
|
|
|
a=a.reshape([a.shape[0],1])
|
2019-05-16 15:14:03 +05:30
|
|
|
dataset[p,:] = a[u,:]
|
2019-09-14 04:31:30 +05:30
|
|
|
|
|
|
|
p = np.where(f['mapping/cellResults/materialpoint']['Name'] == str.encode(label))[0]
|
|
|
|
if len(p)>0:
|
2019-05-16 03:57:06 +05:30
|
|
|
u = (f['mapping/cellResults/materialpoint'][p.tolist()]['Position'])
|
2019-05-16 13:01:13 +05:30
|
|
|
a = np.array(f[pa])
|
|
|
|
if len(a.shape) == 1:
|
|
|
|
a=a.reshape([a.shape[0],1])
|
|
|
|
dataset[p,:] = a[u,:]
|
2019-05-16 03:57:06 +05:30
|
|
|
|
2019-04-17 23:27:16 +05:30
|
|
|
return dataset
|
2019-09-13 18:32:42 +05:30
|
|
|
|
|
|
|
|
2019-05-23 12:24:20 +05:30
|
|
|
def add_Cauchy(self,P='P',F='F'):
|
2019-09-12 06:27:24 +05:30
|
|
|
"""
|
|
|
|
Adds Cauchy stress calculated from 1st Piola-Kirchhoff stress and deformation gradient.
|
|
|
|
|
2019-09-14 07:06:06 +05:30
|
|
|
Resulting tensor is symmetrized as the Cauchy stress should be symmetric.
|
2019-09-12 06:27:24 +05:30
|
|
|
"""
|
2019-05-20 23:24:57 +05:30
|
|
|
def Cauchy(F,P):
|
2019-09-14 07:06:06 +05:30
|
|
|
sigma = np.einsum('i,ijk,ilk->ijl',1.0/np.linalg.det(F['data']),P['data'],F['data'])
|
2019-09-14 07:48:41 +05:30
|
|
|
sigma = (sigma + np.einsum('ikj',sigma))*0.5 # enforce symmetry
|
2019-09-13 18:32:42 +05:30
|
|
|
return {
|
2019-09-14 07:06:06 +05:30
|
|
|
'data' : sigma,
|
2019-09-13 18:32:42 +05:30
|
|
|
'label' : 'sigma',
|
|
|
|
'meta' : {
|
|
|
|
'Unit' : P['meta']['Unit'],
|
|
|
|
'Description' : 'Cauchy stress calculated from {} ({}) '.format(P['label'],P['meta']['Description'])+\
|
2019-09-16 03:40:32 +05:30
|
|
|
'and deformation gradient {} ({})'.format(F['label'],F['meta']['Description']),
|
2019-09-13 18:32:42 +05:30
|
|
|
'Creator' : 'dadf5.py:add_Cauchy vXXXXX'
|
|
|
|
}
|
|
|
|
}
|
2019-05-20 23:24:57 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
requested = [{'label':F,'arg':'F'},
|
|
|
|
{'label':P,'arg':'P'} ]
|
2019-05-20 23:24:57 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
self.__add_generic_pointwise(Cauchy,requested)
|
2019-05-20 23:24:57 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
|
|
|
|
def add_Mises(self,x):
|
2019-09-14 07:48:41 +05:30
|
|
|
"""Adds the equivalent Mises stress or strain of a tensor."""
|
2019-09-14 21:22:07 +05:30
|
|
|
def Mises(x):
|
2019-05-20 23:24:57 +05:30
|
|
|
|
2019-09-14 21:22:07 +05:30
|
|
|
if x['meta']['Unit'] == b'Pa': #ToDo: Should we use this? Then add_Cauchy and add_strain_tensors also should perform sanity checks
|
2019-09-13 18:32:42 +05:30
|
|
|
factor = 3.0/2.0
|
2019-09-14 21:22:07 +05:30
|
|
|
t = 'stress'
|
|
|
|
elif x['meta']['Unit'] == b'1':
|
2019-09-13 18:32:42 +05:30
|
|
|
factor = 2.0/3.0
|
2019-09-14 21:22:07 +05:30
|
|
|
t = 'strain'
|
2019-09-13 18:32:42 +05:30
|
|
|
else:
|
2019-09-14 21:22:07 +05:30
|
|
|
print(x['meta']['Unit'])
|
|
|
|
raise ValueError
|
2019-09-13 18:32:42 +05:30
|
|
|
|
|
|
|
d = x['data']
|
|
|
|
dev = d - np.einsum('ijk,i->ijk',np.broadcast_to(np.eye(3),[d.shape[0],3,3]),np.trace(d,axis1=1,axis2=2)/3.0)
|
2019-09-14 07:48:41 +05:30
|
|
|
#dev_sym = (dev + np.einsum('ikj',dev))*0.5 # ToDo: this is not needed (only if the input is not symmetric, but then the whole concept breaks down)
|
2019-09-13 18:32:42 +05:30
|
|
|
|
|
|
|
return {
|
2019-09-14 07:48:41 +05:30
|
|
|
'data' : np.sqrt(np.einsum('ijk->i',dev**2)*factor),
|
2019-09-14 21:22:07 +05:30
|
|
|
'label' : '{}_vM'.format(x['label']),
|
2019-09-13 18:32:42 +05:30
|
|
|
'meta' : {
|
|
|
|
'Unit' : x['meta']['Unit'],
|
2019-09-14 21:22:07 +05:30
|
|
|
'Description' : 'Mises equivalent {} of {} ({})'.format(t,x['label'],x['meta']['Description']),
|
2019-09-13 18:32:42 +05:30
|
|
|
'Creator' : 'dadf5.py:add_Mises_stress vXXXXX'
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
requested = [{'label':x,'arg':'x'}]
|
2019-05-23 20:52:57 +05:30
|
|
|
|
2019-09-14 21:22:07 +05:30
|
|
|
self.__add_generic_pointwise(Mises,requested)
|
2019-09-13 18:32:42 +05:30
|
|
|
|
|
|
|
|
2019-05-23 12:24:20 +05:30
|
|
|
def add_norm(self,x,ord=None):
|
2019-09-12 07:24:26 +05:30
|
|
|
"""
|
2019-09-14 21:22:07 +05:30
|
|
|
Adds norm of vector or tensor.
|
2019-09-14 04:36:47 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
See numpy.linalg.norm manual for details.
|
2019-09-12 07:24:26 +05:30
|
|
|
"""
|
2019-09-13 18:32:42 +05:30
|
|
|
def norm(x,ord):
|
|
|
|
|
2019-09-14 09:44:52 +05:30
|
|
|
o = ord
|
2019-09-14 21:22:07 +05:30
|
|
|
if len(x['data'].shape) == 2:
|
2019-09-13 18:32:42 +05:30
|
|
|
axis = 1
|
|
|
|
t = 'vector'
|
2019-09-14 09:44:52 +05:30
|
|
|
if o is None: o = 2
|
2019-09-13 18:32:42 +05:30
|
|
|
elif len(x['data'].shape) == 3:
|
|
|
|
axis = (1,2)
|
|
|
|
t = 'tensor'
|
2019-09-14 09:44:52 +05:30
|
|
|
if o is None: o = 'fro'
|
2019-09-13 18:32:42 +05:30
|
|
|
else:
|
|
|
|
raise ValueError
|
|
|
|
|
|
|
|
return {
|
2019-09-14 09:44:52 +05:30
|
|
|
'data' : np.linalg.norm(x['data'],ord=o,axis=axis,keepdims=True),
|
2019-09-14 21:22:07 +05:30
|
|
|
'label' : '|{}|_{}'.format(x['label'],o),
|
2019-09-13 18:32:42 +05:30
|
|
|
'meta' : {
|
|
|
|
'Unit' : x['meta']['Unit'],
|
2019-09-14 09:44:52 +05:30
|
|
|
'Description' : '{}-Norm of {} {} ({})'.format(ord,t,x['label'],x['meta']['Description']),
|
2019-09-13 18:32:42 +05:30
|
|
|
'Creator' : 'dadf5.py:add_norm vXXXXX'
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
requested = [{'label':x,'arg':'x'}]
|
2019-09-14 21:22:07 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
self.__add_generic_pointwise(norm,requested,{'ord':ord})
|
2019-09-14 21:22:07 +05:30
|
|
|
|
|
|
|
|
|
|
|
def add_absolute(self,x):
|
|
|
|
"""Adds absolute value."""
|
|
|
|
def absolute(x):
|
|
|
|
|
|
|
|
return {
|
|
|
|
'data' : np.abs(x['data']),
|
|
|
|
'label' : '|{}|'.format(x['label']),
|
|
|
|
'meta' : {
|
|
|
|
'Unit' : x['meta']['Unit'],
|
|
|
|
'Description' : 'Absolute value of {} ({})'.format(x['label'],x['meta']['Description']),
|
|
|
|
'Creator' : 'dadf5.py:add_abs vXXXXX'
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
requested = [{'label':x,'arg':'x'}]
|
|
|
|
|
|
|
|
self.__add_generic_pointwise(absolute,requested)
|
2019-09-13 18:32:42 +05:30
|
|
|
|
|
|
|
|
|
|
|
def add_determinant(self,x):
|
|
|
|
"""Adds the determinant component of a tensor."""
|
|
|
|
def determinant(x):
|
|
|
|
|
|
|
|
return {
|
|
|
|
'data' : np.linalg.det(x['data']),
|
|
|
|
'label' : 'det({})'.format(x['label']),
|
|
|
|
'meta' : {
|
|
|
|
'Unit' : x['meta']['Unit'],
|
|
|
|
'Description' : 'Determinant of tensor {} ({})'.format(x['label'],x['meta']['Description']),
|
|
|
|
'Creator' : 'dadf5.py:add_determinant vXXXXX'
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
requested = [{'label':x,'arg':'x'}]
|
2019-05-23 12:24:20 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
self.__add_generic_pointwise(determinant,requested)
|
2019-07-16 02:25:14 +05:30
|
|
|
|
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
def add_spherical(self,x):
|
2019-09-12 06:33:19 +05:30
|
|
|
"""Adds the spherical component of a tensor."""
|
2019-09-13 18:32:42 +05:30
|
|
|
def spherical(x):
|
|
|
|
|
|
|
|
if not np.all(np.array(x['data'].shape[1:]) == np.array([3,3])):
|
|
|
|
raise ValueError
|
2019-07-16 02:25:14 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
return {
|
|
|
|
'data' : np.trace(x['data'],axis1=1,axis2=2)/3.0,
|
|
|
|
'label' : 'sph({})'.format(x['label']),
|
|
|
|
'meta' : {
|
|
|
|
'Unit' : x['meta']['Unit'],
|
|
|
|
'Description' : 'Spherical component of tensor {} ({})'.format(x['label'],x['meta']['Description']),
|
|
|
|
'Creator' : 'dadf5.py:add_spherical vXXXXX'
|
|
|
|
}
|
|
|
|
}
|
2019-07-16 02:25:14 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
requested = [{'label':x,'arg':'x'}]
|
|
|
|
|
|
|
|
self.__add_generic_pointwise(spherical,requested)
|
|
|
|
|
|
|
|
|
|
|
|
def add_deviator(self,x):
|
2019-09-12 06:33:19 +05:30
|
|
|
"""Adds the deviator of a tensor."""
|
2019-09-13 18:32:42 +05:30
|
|
|
def deviator(x):
|
|
|
|
d = x['data']
|
2019-09-14 07:48:41 +05:30
|
|
|
|
|
|
|
if not np.all(np.array(d.shape[1:]) == np.array([3,3])):
|
|
|
|
raise ValueError
|
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
return {
|
|
|
|
'data' : d - np.einsum('ijk,i->ijk',np.broadcast_to(np.eye(3),[d.shape[0],3,3]),np.trace(d,axis1=1,axis2=2)/3.0),
|
|
|
|
'label' : 'dev({})'.format(x['label']),
|
|
|
|
'meta' : {
|
|
|
|
'Unit' : x['meta']['Unit'],
|
|
|
|
'Description' : 'Deviator of tensor {} ({})'.format(x['label'],x['meta']['Description']),
|
|
|
|
'Creator' : 'dadf5.py:add_deviator vXXXXX'
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
requested = [{'label':x,'arg':'x'}]
|
|
|
|
|
|
|
|
self.__add_generic_pointwise(deviator,requested)
|
2019-09-15 00:30:23 +05:30
|
|
|
|
|
|
|
|
|
|
|
def add_calculation(self,formula,label,unit='n/a',description=None,vectorized=True):
|
|
|
|
"""
|
|
|
|
General formula.
|
|
|
|
|
|
|
|
Works currently only for vectorized expressions
|
|
|
|
|
|
|
|
"""
|
|
|
|
if vectorized is not True:
|
|
|
|
raise NotImplementedError
|
|
|
|
|
|
|
|
def calculation(**kwargs):
|
|
|
|
|
|
|
|
formula = kwargs['formula']
|
|
|
|
for d in re.findall(r'#(.*?)#',formula):
|
|
|
|
formula = re.sub('#{}#'.format(d),"kwargs['{}']['data']".format(d),formula)
|
|
|
|
|
|
|
|
return {
|
|
|
|
'data' : eval(formula),
|
|
|
|
'label' : kwargs['label'],
|
|
|
|
'meta' : {
|
|
|
|
'Unit' : kwargs['unit'],
|
|
|
|
'Description' : '{}'.format(kwargs['description']),
|
|
|
|
'Creator' : 'dadf5.py:add_calculation vXXXXX'
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
requested = [{'label':d,'arg':d} for d in re.findall(r'#(.*?)#',formula)] # datasets used in the formula
|
|
|
|
pass_through = {'formula':formula,'label':label,'unit':unit,'description':description}
|
|
|
|
|
|
|
|
self.__add_generic_pointwise(calculation,requested,pass_through)
|
2019-07-16 02:25:14 +05:30
|
|
|
|
|
|
|
|
2019-09-14 07:48:41 +05:30
|
|
|
def add_strain_tensor(self,t,ord,defgrad='F'): #ToDo: Use t and ord
|
2019-09-14 23:23:33 +05:30
|
|
|
"""
|
|
|
|
Adds the a strain tensor.
|
|
|
|
|
|
|
|
Albrecht Bertram: Elasticity and Plasticity of Large Deformations An Introduction (3rd Edition, 2012), p. 102.
|
|
|
|
"""
|
2019-09-13 18:32:42 +05:30
|
|
|
def strain_tensor(defgrad,t,ord):
|
2019-09-14 23:23:33 +05:30
|
|
|
|
|
|
|
operator = {
|
|
|
|
'V#ln': lambda V: np.log(V),
|
|
|
|
'U#ln': lambda U: np.log(U),
|
|
|
|
'V#Biot': lambda V: np.broadcast_to(np.ones(3),[V.shape[0],3]) - 1.0/V,
|
|
|
|
'U#Biot': lambda U: U - np.broadcast_to(np.ones(3),[U.shape[0],3]),
|
|
|
|
'V#Green':lambda V: np.broadcast_to(np.ones(3),[V.shape[0],3]) - 1.0/V**2,
|
|
|
|
'U#Biot': lambda U: U**2 - np.broadcast_to(np.ones(3),[U.shape[0],3]),
|
|
|
|
}
|
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
(U,S,Vh) = np.linalg.svd(defgrad['data']) # singular value decomposition
|
2019-09-14 07:48:41 +05:30
|
|
|
R_inv = np.einsum('ikj',np.matmul(U,Vh)) # inverse rotation of polar decomposition
|
2019-09-13 18:32:42 +05:30
|
|
|
U = np.matmul(R_inv,defgrad['data']) # F = RU
|
|
|
|
(D,V) = np.linalg.eigh((U+np.einsum('ikj',U))*.5) # eigen decomposition (of symmetric(ed) matrix)
|
2019-07-16 02:25:14 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
neg = np.where(D < 0.0) # find negative eigenvalues ...
|
|
|
|
D[neg[0],neg[1]] = D[neg[0],neg[1]]* -1 # ... flip value ...
|
|
|
|
V[neg[0],:,neg[1]] = V[neg[0],:,neg[1]]* -1 # ... and vector
|
|
|
|
|
2019-09-14 23:23:33 +05:30
|
|
|
d = operator['V#ln'](D)
|
2019-09-14 07:48:41 +05:30
|
|
|
a = np.matmul(V,np.einsum('ij,ikj->ijk',d,V))
|
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
return {
|
|
|
|
'data' : a,
|
2019-09-14 07:48:41 +05:30
|
|
|
'label' : 'ln(V)({})'.format(defgrad['label']),
|
2019-09-13 18:32:42 +05:30
|
|
|
'meta' : {
|
|
|
|
'Unit' : defgrad['meta']['Unit'],
|
2019-09-14 07:48:41 +05:30
|
|
|
'Description' : 'Strain tensor ln(V){} ({})'.format(defgrad['label'],defgrad['meta']['Description']),
|
2019-09-13 18:32:42 +05:30
|
|
|
'Creator' : 'dadf5.py:add_deviator vXXXXX'
|
|
|
|
}
|
|
|
|
}
|
2019-05-20 23:24:57 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
requested = [{'label':defgrad,'arg':'defgrad'}]
|
2019-05-20 23:24:57 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
self.__add_generic_pointwise(strain_tensor,requested,{'t':t,'ord':ord})
|
2019-05-20 23:24:57 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
|
|
|
|
def __add_generic_pointwise(self,func,datasets_requested,extra_args={}):
|
2019-09-12 06:27:24 +05:30
|
|
|
"""
|
2019-09-12 06:33:19 +05:30
|
|
|
General function to add pointwise data.
|
2019-09-14 04:36:47 +05:30
|
|
|
|
|
|
|
Parameters
|
|
|
|
----------
|
|
|
|
func : function
|
2019-09-14 09:44:52 +05:30
|
|
|
Function that calculates a new dataset from one or more datasets per HDF5 group.
|
2019-09-14 04:36:47 +05:30
|
|
|
datasets_requested : list of dictionaries
|
2019-09-14 09:44:52 +05:30
|
|
|
Details of the datasets to be used: label (in HDF5 file) and arg (argument to which the data is parsed in func).
|
|
|
|
extra_args : dictionary, optional
|
|
|
|
Any extra arguments parsed to func.
|
2019-09-13 18:32:42 +05:30
|
|
|
|
2019-09-14 04:36:47 +05:30
|
|
|
"""
|
2019-09-12 06:27:24 +05:30
|
|
|
def job(args):
|
2019-09-14 04:36:47 +05:30
|
|
|
"""Call function with input data + extra arguments, returns results + group."""
|
2019-09-13 18:32:42 +05:30
|
|
|
args['results'].put({**args['func'](**args['in']),'group':args['group']})
|
2019-09-12 06:27:24 +05:30
|
|
|
|
2019-09-13 18:32:42 +05:30
|
|
|
|
|
|
|
N_threads = 1 # ToDo: should be a parameter
|
|
|
|
|
|
|
|
results = Queue(N_threads)
|
|
|
|
pool = util.ThreadPool(N_threads)
|
|
|
|
N_added = N_threads + 1
|
2019-09-12 06:27:24 +05:30
|
|
|
|
|
|
|
todo = []
|
2019-09-13 18:32:42 +05:30
|
|
|
# ToDo: It would be more memory efficient to read only from file when required, i.e. do to it in pool.add_task
|
2019-09-14 21:22:07 +05:30
|
|
|
for group in self.groups_with_datasets([d['label'] for d in datasets_requested]):
|
2019-09-12 06:27:24 +05:30
|
|
|
with h5py.File(self.filename,'r') as f:
|
2019-09-13 18:32:42 +05:30
|
|
|
datasets_in = {}
|
|
|
|
for d in datasets_requested:
|
|
|
|
loc = f[group+'/'+d['label']]
|
|
|
|
data = loc[()]
|
|
|
|
meta = {k:loc.attrs[k] for k in loc.attrs.keys()}
|
|
|
|
datasets_in[d['arg']] = {'data': data, 'meta' : meta, 'label' : d['label']}
|
|
|
|
|
|
|
|
todo.append({'in':{**datasets_in,**extra_args},'func':func,'group':group,'results':results})
|
|
|
|
|
|
|
|
pool.map(job, todo[:N_added]) # initialize
|
|
|
|
|
|
|
|
N_not_calculated = len(todo)
|
|
|
|
while N_not_calculated > 0:
|
|
|
|
result = results.get()
|
2019-09-14 04:31:30 +05:30
|
|
|
with h5py.File(self.filename,'a') as f: # write to file
|
2019-09-13 18:32:42 +05:30
|
|
|
dataset_out = f[result['group']].create_dataset(result['label'],data=result['data'])
|
|
|
|
for k in result['meta'].keys():
|
|
|
|
dataset_out.attrs[k] = result['meta'][k]
|
|
|
|
N_not_calculated-=1
|
|
|
|
|
|
|
|
if N_added < len(todo): # add more jobs
|
|
|
|
pool.add_task(job,todo[N_added])
|
|
|
|
N_added +=1
|
2019-09-12 06:27:24 +05:30
|
|
|
|
|
|
|
pool.wait_completion()
|