DAMASK_EICMD/python/damask/_result.py

2000 lines
80 KiB
Python
Raw Normal View History

2020-06-26 15:15:06 +05:30
import multiprocessing as mp
2022-05-09 03:51:53 +05:30
from multiprocessing.synchronize import Lock
import re
2021-04-04 23:14:06 +05:30
import fnmatch
import os
2021-04-05 11:23:19 +05:30
import copy
2020-05-25 22:20:31 +05:30
import datetime
2022-02-02 04:32:03 +05:30
import xml.etree.ElementTree as ET # noqa
import xml.dom.minidom
2020-06-03 14:13:07 +05:30
from pathlib import Path
from functools import partial
from collections import defaultdict
2021-03-31 18:49:23 +05:30
from collections.abc import Iterable
2022-11-23 02:56:15 +05:30
from typing import Optional, Union, Callable, Any, Sequence, Literal, Dict, List, Tuple
import h5py
2019-04-17 23:27:16 +05:30
import numpy as np
import numpy.ma as ma
2020-06-28 15:10:19 +05:30
import damask
2020-03-11 11:20:13 +05:30
from . import VTK
2020-02-16 00:39:24 +05:30
from . import Orientation
2020-02-21 22:17:47 +05:30
from . import grid_filters
2020-03-13 00:22:33 +05:30
from . import mechanics
from . import tensor
2020-03-13 00:22:33 +05:30
from . import util
2022-05-09 03:51:53 +05:30
from ._typehints import FloatSequence, IntSequence
2022-05-02 14:48:35 +05:30
h5py3 = h5py.__version__[0] == '3'
2021-05-28 16:50:56 +05:30
chunk_size = 1024**2//8 # for compression in HDF5
prefix_inc = 'increment_'
2021-05-28 16:50:56 +05:30
2022-05-09 03:51:53 +05:30
def _read(dataset: h5py._hl.dataset.Dataset) -> np.ndarray:
2021-04-05 11:23:19 +05:30
"""Read a dataset and its metadata into a numpy.ndarray."""
2021-04-05 13:59:34 +05:30
metadata = {k:(v.decode() if not h5py3 and type(v) is bytes else v) for k,v in dataset.attrs.items()}
2022-05-09 03:51:53 +05:30
dtype = np.dtype(dataset.dtype,metadata=metadata) # type: ignore
return np.array(dataset,dtype=dtype)
2021-03-31 15:35:51 +05:30
2022-05-09 03:51:53 +05:30
def _match(requested,
existing: h5py._hl.base.KeysViewHDF5) -> List[Any]:
"""Find matches among two sets of labels."""
2021-04-03 14:38:22 +05:30
def flatten_list(list_of_lists):
return [e for e_ in list_of_lists for e in e_]
if requested is True:
requested = '*'
elif requested is False or requested is None:
requested = []
requested_ = requested if hasattr(requested,'__iter__') and not isinstance(requested,str) else \
[requested]
2021-04-04 23:14:06 +05:30
return sorted(set(flatten_list([fnmatch.filter(existing,r) for r in requested_])),
2021-04-03 14:38:22 +05:30
key=util.natural_sort)
def _empty_like(dataset: np.ma.core.MaskedArray,
N_materialpoints: int,
fill_float: float,
fill_int: int) -> np.ma.core.MaskedArray:
2021-04-05 11:23:19 +05:30
"""Create empty numpy.ma.MaskedArray."""
2021-04-04 16:55:16 +05:30
return ma.array(np.empty((N_materialpoints,)+dataset.shape[1:],dataset.dtype),
fill_value = fill_float if dataset.dtype in np.sctypes['float'] else fill_int,
mask = True)
2021-03-31 15:35:51 +05:30
class Result:
2019-09-20 01:02:15 +05:30
"""
Add data to and export data from a DADF5 file.
2019-09-16 08:49:14 +05:30
2021-04-24 22:42:44 +05:30
A DADF5 (DAMASK HDF5) file contains DAMASK results.
Its group/folder structure reflects the layout in material.yaml.
2021-04-24 22:42:44 +05:30
This class provides a customizable view on the DADF5 file.
Upon initialization, all attributes are visible.
Derived quantities are added to the file and existing data is
exported based on the current view.
Examples
--------
2021-04-24 22:42:44 +05:30
Open 'my_file.hdf5', which is assumed to contain deformation gradient 'F'
and first Piola-Kirchhoff stress 'P', add the Mises equivalent of the
Cauchy stress, and export it to VTK (file) and numpy.ndarray (memory).
>>> import damask
>>> r = damask.Result('my_file.hdf5')
2022-07-27 16:26:37 +05:30
>>> r.add_stress_Cauchy()
>>> r.add_equivalent_Mises('sigma')
2021-06-17 21:56:37 +05:30
>>> r.export_VTK()
>>> r_last = r.view(increments=-1)
>>> sigma_vM_last = r_last.get('sigma_vM')
2019-09-16 08:49:14 +05:30
2020-02-21 12:15:05 +05:30
"""
def __init__(self, fname: Union[str, Path]):
2020-02-21 12:15:05 +05:30
"""
New result view bound to a HDF5 file.
2019-09-16 08:49:14 +05:30
2020-02-21 12:15:05 +05:30
Parameters
----------
2020-11-19 18:15:40 +05:30
fname : str or pathlib.Path
Name of the DADF5 file to be opened.
2019-09-20 01:02:15 +05:30
2020-02-21 12:15:05 +05:30
"""
with h5py.File(fname,'r') as f:
self.version_major = f.attrs['DADF5_version_major']
self.version_minor = f.attrs['DADF5_version_minor']
if self.version_major != 0 or not 12 <= self.version_minor <= 14:
raise TypeError(f'unsupported DADF5 version "{self.version_major}.{self.version_minor}"')
if self.version_major == 0 and self.version_minor < 14:
self.export_simulation_setup = None # type: ignore
2020-03-22 20:43:35 +05:30
2021-04-14 10:36:24 +05:30
self.structured = 'cells' in f['geometry'].attrs.keys()
2020-03-22 20:43:35 +05:30
if self.structured:
2021-04-14 10:36:24 +05:30
self.cells = f['geometry'].attrs['cells']
2020-03-22 20:43:35 +05:30
self.size = f['geometry'].attrs['size']
self.origin = f['geometry'].attrs['origin']
2021-05-29 00:27:44 +05:30
else:
2022-05-09 03:51:53 +05:30
self.add_curl = self.add_divergence = self.add_gradient = None # type: ignore
r = re.compile(rf'{prefix_inc}([0-9]+)')
2022-04-22 21:56:52 +05:30
self.increments = sorted([i for i in f.keys() if r.match(i)],key=util.natural_sort)
2022-11-11 11:33:14 +05:30
self.times = np.around([f[i].attrs['t/s'] for i in self.increments],12)
2022-04-22 21:56:52 +05:30
if len(self.increments) == 0:
raise ValueError('incomplete DADF5 file')
self.N_materialpoints, self.N_constituents = np.shape(f['cell_to/phase'])
2021-03-25 23:52:59 +05:30
self.homogenization = f['cell_to/homogenization']['label'].astype('str')
self.homogenizations = sorted(np.unique(self.homogenization),key=util.natural_sort)
self.phase = f['cell_to/phase']['label'].astype('str')
self.phases = sorted(np.unique(self.phase),key=util.natural_sort)
2020-03-22 20:43:35 +05:30
self.fields: List[str] = []
for c in self.phases:
self.fields += f['/'.join([self.increments[0],'phase',c])].keys()
for m in self.homogenizations:
self.fields += f['/'.join([self.increments[0],'homogenization',m])].keys()
2021-04-03 14:38:22 +05:30
self.fields = sorted(set(self.fields),key=util.natural_sort) # make unique
2021-01-13 19:27:58 +05:30
self.visible = {'increments': self.increments,
'phases': self.phases,
'homogenizations': self.homogenizations,
'fields': self.fields,
2021-01-13 19:27:58 +05:30
}
self.fname = Path(fname).expanduser().absolute()
2021-12-17 15:07:45 +05:30
self._protected = True
def __copy__(self) -> "Result":
"""
Return deepcopy(self).
Create deep copy.
"""
2021-04-05 11:23:19 +05:30
return copy.deepcopy(self)
copy = __copy__
def __repr__(self) -> str:
"""
Return repr(self).
Give short human-readable summary.
"""
with h5py.File(self.fname,'r') as f:
header = [f'Created by {f.attrs["creator"]}',
f' on {f.attrs["created"]}',
f' executing "{f.attrs["call"]}"']
2021-01-13 19:27:58 +05:30
visible_increments = self.visible['increments']
2020-04-21 14:47:15 +05:30
first = self.view(increments=visible_increments[0:1]).list_data()
2020-04-21 14:47:15 +05:30
last = [] if len(visible_increments) < 2 else \
self.view(increments=visible_increments[-1:]).list_data()
2020-04-21 14:47:15 +05:30
in_between = [] if len(visible_increments) < 3 else \
[f'\n{inc}\n ...' for inc in visible_increments[1:-1]]
2020-04-21 14:47:15 +05:30
return util.srepr([util.deemph(header)] + first + in_between + last)
def _manage_view(self,
action: Literal['set', 'add', 'del'],
2022-11-23 02:56:15 +05:30
increments: Union[None, int, Sequence[int], str, Sequence[str], bool] = None,
times: Union[None, float, Sequence[float], str, Sequence[str], bool] = None,
phases: Union[None, str, Sequence[str], bool] = None,
homogenizations: Union[None, str, Sequence[str], bool] = None,
fields: Union[None, str, Sequence[str], bool] = None) -> "Result":
2020-02-21 12:15:05 +05:30
"""
Manages the visibility of the groups.
2020-02-21 12:15:05 +05:30
Parameters
----------
2020-03-03 04:17:29 +05:30
action : str
2020-11-23 23:52:48 +05:30
Select from 'set', 'add', and 'del'.
2020-03-03 04:17:29 +05:30
Returns
-------
view : damask.Result
Modified or new view on the DADF5 file.
2020-02-21 12:15:05 +05:30
"""
if increments is not None and times is not None:
2022-03-09 19:48:18 +05:30
raise ValueError('"increments" and "times" are mutually exclusive')
2020-02-21 12:15:05 +05:30
2021-04-05 11:23:19 +05:30
dup = self.copy()
for what,datasets in zip(['increments','times','phases','homogenizations','fields'],
[ increments, times, phases, homogenizations, fields ]):
if datasets is None:
continue
# allow True/False and string arguments
elif datasets is True:
datasets = '*'
elif datasets is False:
datasets = []
2022-05-09 03:51:53 +05:30
choice = [datasets] if not hasattr(datasets,'__iter__') or isinstance(datasets,str) else list(datasets) # type: ignore
if what == 'increments':
choice = [c if isinstance(c,str) and c.startswith(prefix_inc) else
self.increments[c] if isinstance(c,int) and c<0 else
f'{prefix_inc}{c}' for c in choice]
elif what == 'times':
2022-11-11 11:33:14 +05:30
atol = 1e-2 * np.min(np.diff(self.times))
what = 'increments'
if choice == ['*']:
choice = self.increments
else:
2022-11-11 11:33:14 +05:30
iterator = np.array(choice).astype(float)
choice = []
for c in iterator:
2022-11-11 11:33:14 +05:30
idx = np.searchsorted(self.times,c,side='left')
if idx<len(self.times) and np.isclose(c,self.times[idx],rtol=0,atol=atol):
choice.append(self.increments[idx])
2022-11-11 11:33:14 +05:30
elif idx>0 and np.isclose(c,self.times[idx-1],rtol=0,atol=atol):
choice.append(self.increments[idx-1])
valid = _match(choice,getattr(self,what))
existing = set(self.visible[what])
if action == 'set':
dup.visible[what] = sorted(set(valid), key=util.natural_sort)
elif action == 'add':
2022-11-11 11:33:14 +05:30
dup.visible[what] = sorted(existing.union(valid), key=util.natural_sort)
elif action == 'del':
2022-11-11 11:33:14 +05:30
dup.visible[what] = sorted(existing.difference(valid), key=util.natural_sort)
2021-04-05 11:23:19 +05:30
return dup
2020-02-21 12:15:05 +05:30
def increments_in_range(self,
2022-11-23 02:56:15 +05:30
start: Union[None, str, int] = None,
end: Union[None, str, int] = None) -> Sequence[int]:
2020-11-19 18:15:40 +05:30
"""
Get all increments within a given range.
2020-11-19 18:15:40 +05:30
Parameters
----------
2022-04-02 03:34:50 +05:30
start : int or str, optional
Start increment. Defaults to first.
end : int or str, optional
End increment. Defaults to last.
2020-11-19 18:15:40 +05:30
Returns
-------
increments : list of ints
Increment number of all increments within the given bounds.
2021-04-24 18:17:52 +05:30
2020-11-19 18:15:40 +05:30
"""
s,e = map(lambda x: int(x.split(prefix_inc)[-1] if isinstance(x,str) and x.startswith(prefix_inc) else x),
2022-04-02 03:34:50 +05:30
(self.incs[ 0] if start is None else start,
self.incs[-1] if end is None else end))
return [i for i in self.incs if s <= i <= e]
def times_in_range(self,
2022-11-23 02:56:15 +05:30
start: Optional[float] = None,
end: Optional[float] = None) -> Sequence[float]:
2020-11-19 18:15:40 +05:30
"""
Get times of all increments within a given time range.
2020-11-19 18:15:40 +05:30
Parameters
----------
2022-04-22 01:26:17 +05:30
start : float, optional
Time of start increment. Defaults to time of first.
2022-04-22 01:26:17 +05:30
end : float, optional
Time of end increment. Defaults to time of last.
2020-11-19 18:15:40 +05:30
Returns
-------
times : list of float
2022-04-22 01:26:17 +05:30
Time of each increment within the given bounds.
2021-04-24 18:17:52 +05:30
2020-11-19 18:15:40 +05:30
"""
2022-04-22 01:26:17 +05:30
s,e = (self.times[ 0] if start is None else start,
self.times[-1] if end is None else end)
return [t for t in self.times if s <= t <= e]
2022-03-05 00:30:33 +05:30
def view(self,*,
2022-11-23 02:56:15 +05:30
increments: Union[None, int, Sequence[int], str, Sequence[str], bool] = None,
times: Union[None, float, Sequence[float], str, Sequence[str], bool] = None,
phases: Union[None, str, Sequence[str], bool] = None,
homogenizations: Union[None, str, Sequence[str], bool] = None,
fields: Union[None, str, Sequence[str], bool] = None,
protected: Optional[bool] = None) -> "Result":
2020-02-21 12:15:05 +05:30
"""
2021-01-13 19:27:58 +05:30
Set view.
Wildcard matching with '?' and '*' is supported.
True is equivalent to '*', False is equivalent to [].
2020-02-21 12:15:05 +05:30
Parameters
----------
increments: (list of) int, (list of) str, or bool, optional.
Number(s) of increments to select.
times: (list of) float, (list of) str, or bool, optional.
Simulation time(s) of increments to select.
phases: (list of) str, or bool, optional.
Name(s) of phases to select.
homogenizations: (list of) str, or bool, optional.
Name(s) of homogenizations to select.
fields: (list of) str, or bool, optional.
Name(s) of fields to select.
2021-12-17 15:01:41 +05:30
protected: bool, optional.
Protection status of existing data.
Returns
-------
view : damask.Result
2021-04-24 22:42:44 +05:30
View with only the selected attributes being visible.
Examples
--------
Get a view that shows only results from the initial configuration:
>>> import damask
>>> r = damask.Result('my_file.hdf5')
2022-06-06 03:09:06 +05:30
>>> r_first = r.view(increments=0)
2021-04-24 22:42:44 +05:30
Get a view that shows all results between simulation times of 10 to 40:
>>> import damask
>>> r = damask.Result('my_file.hdf5')
>>> r_t10to40 = r.view(times=r.times_in_range(10.0,40.0))
2020-02-21 12:15:05 +05:30
"""
dup = self._manage_view('set',increments,times,phases,homogenizations,fields)
2021-12-17 15:01:41 +05:30
if protected is not None:
2021-12-17 18:58:12 +05:30
if not protected:
2021-12-17 15:07:45 +05:30
print(util.warn('Warning: Modification of existing datasets allowed!'))
dup._protected = protected
2021-12-17 15:01:41 +05:30
return dup
2022-03-05 00:30:33 +05:30
def view_more(self,*,
2022-11-23 02:56:15 +05:30
increments: Union[None, int, Sequence[int], str, Sequence[str], bool] = None,
times: Union[None, float, Sequence[float], str, Sequence[str], bool] = None,
phases: Union[None, str, Sequence[str], bool] = None,
homogenizations: Union[None, str, Sequence[str], bool] = None,
fields: Union[None, str, Sequence[str], bool] = None) -> "Result":
2020-02-21 12:15:05 +05:30
"""
2021-01-13 19:27:58 +05:30
Add to view.
Wildcard matching with '?' and '*' is supported.
True is equivalent to '*', False is equivalent to [].
2020-02-21 12:15:05 +05:30
Parameters
----------
increments: (list of) int, (list of) str, or bool, optional.
Number(s) of increments to select.
times: (list of) float, (list of) str, or bool, optional.
Simulation time(s) of increments to select.
phases: (list of) str, or bool, optional.
Name(s) of phases to select.
homogenizations: (list of) str, or bool, optional.
Name(s) of homogenizations to select.
fields: (list of) str, or bool, optional.
Name(s) of fields to select.
Returns
-------
modified_view : damask.Result
2021-04-24 22:42:44 +05:30
View with additional visible attributes.
Examples
--------
Get a view that shows only results from first and last increment:
>>> import damask
>>> r_empty = damask.Result('my_file.hdf5').view(increments=False)
>>> r_first = r_empty.view_more(increments=0)
>>> r_first_and_last = r.first.view_more(increments=-1)
2020-02-21 12:15:05 +05:30
"""
return self._manage_view('add',increments,times,phases,homogenizations,fields)
2022-03-05 00:30:33 +05:30
def view_less(self,*,
2022-11-23 02:56:15 +05:30
increments: Union[None, int, Sequence[int], str, Sequence[str], bool] = None,
times: Union[None, float, Sequence[float], str, Sequence[str], bool] = None,
phases: Union[None, str, Sequence[str], bool] = None,
homogenizations: Union[None, str, Sequence[str], bool] = None,
fields: Union[None, str, Sequence[str], bool] = None) -> "Result":
2020-02-21 12:15:05 +05:30
"""
2021-04-24 22:42:44 +05:30
Remove from view.
Wildcard matching with '?' and '*' is supported.
True is equivalent to '*', False is equivalent to [].
2020-02-21 12:15:05 +05:30
Parameters
----------
increments: (list of) int, (list of) str, or bool, optional.
Number(s) of increments to select.
times: (list of) float, (list of) str, or bool, optional.
Simulation time(s) of increments to select.
phases: (list of) str, or bool, optional.
Name(s) of phases to select.
homogenizations: (list of) str, or bool, optional.
Name(s) of homogenizations to select.
fields: (list of) str, or bool, optional.
Name(s) of fields to select.
2019-10-19 16:40:46 +05:30
Returns
-------
modified_view : damask.Result
2021-04-24 22:42:44 +05:30
View with fewer visible attributes.
Examples
--------
2021-04-24 22:42:44 +05:30
Get a view that omits the undeformed configuration:
>>> import damask
>>> r_all = damask.Result('my_file.hdf5')
>>> r_deformed = r_all.view_less(increments=0)
2020-02-21 12:15:05 +05:30
"""
return self._manage_view('del',increments,times,phases,homogenizations,fields)
def rename(self,
name_src: str,
name_dst: str):
"""
Rename/move datasets (within the same group/folder).
This operation is discouraged because the history of the
2021-04-25 21:04:41 +05:30
data becomes untraceable and data integrity is not ensured.
2020-06-25 01:04:51 +05:30
2020-06-02 01:43:01 +05:30
Parameters
----------
name_src : str
Name of the datasets to be renamed.
name_dst : str
New name of the datasets.
Examples
--------
Rename datasets containing the deformation gradient from 'F' to 'def_grad':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
2021-12-17 15:01:41 +05:30
>>> r_unprotected = r.view(protected=False)
>>> r_unprotected.rename('F','def_grad')
"""
2021-12-17 15:07:45 +05:30
if self._protected:
raise PermissionError('rename datasets')
2021-04-04 23:14:06 +05:30
with h5py.File(self.fname,'a') as f:
for inc in self.visible['increments']:
2021-04-05 11:23:19 +05:30
for ty in ['phase','homogenization']:
for label in self.visible[ty+'s']:
for field in _match(self.visible['fields'],f['/'.join([inc,ty,label])].keys()):
path_src = '/'.join([inc,ty,label,field,name_src])
path_dst = '/'.join([inc,ty,label,field,name_dst])
if path_src in f.keys():
f[path_dst] = f[path_src]
f[path_dst].attrs['renamed'] = f'original name: {name_src}' if h5py3 else \
f'original name: {name_src}'.encode()
del f[path_src]
def remove(self, name: str):
"""
Remove/delete datasets.
This operation is discouraged because the history of the
2021-04-25 21:04:41 +05:30
data becomes untraceable and data integrity is not ensured.
Parameters
----------
name : str
Name of the datasets to be deleted.
Examples
--------
Delete the deformation gradient 'F':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
2021-12-17 15:01:41 +05:30
>>> r_unprotected = r.view(protected=False)
>>> r_unprotected.remove('F')
"""
2021-12-17 15:07:45 +05:30
if self._protected:
raise PermissionError('delete datasets')
with h5py.File(self.fname,'a') as f:
for inc in self.visible['increments']:
for ty in ['phase','homogenization']:
for label in self.visible[ty+'s']:
for field in _match(self.visible['fields'],f['/'.join([inc,ty,label])].keys()):
path = '/'.join([inc,ty,label,field,name])
if path in f.keys(): del f[path]
2021-04-04 23:14:06 +05:30
def list_data(self) -> List[str]:
"""
Collect information on all active datasets in the file.
Returns
-------
data : list of str
Line-formatted information about active datasets.
"""
msg = []
2020-02-21 12:15:05 +05:30
with h5py.File(self.fname,'r') as f:
2021-04-04 15:57:39 +05:30
for inc in self.visible['increments']:
msg += [f'\n{inc} ({self.times[self.increments.index(inc)]} s)']
2021-04-05 11:23:19 +05:30
for ty in ['phase','homogenization']:
msg += [f' {ty}']
2021-04-05 11:23:19 +05:30
for label in self.visible[ty+'s']:
msg += [f' {label}']
for field in _match(self.visible['fields'],f['/'.join([inc,ty,label])].keys()):
msg += [f' {field}']
2021-04-05 11:23:19 +05:30
for d in f['/'.join([inc,ty,label,field])].keys():
dataset = f['/'.join([inc,ty,label,field,d])]
unit = dataset.attrs["unit"] if h5py3 else \
dataset.attrs["unit"].decode()
2021-04-14 10:36:24 +05:30
description = dataset.attrs['description'] if h5py3 else \
dataset.attrs['description'].decode()
msg += [f' {d} / {unit}: {description}']
2020-02-21 12:15:05 +05:30
2021-04-05 11:23:19 +05:30
return msg
2020-02-21 12:15:05 +05:30
def enable_user_function(self,
func: Callable):
2020-07-03 10:59:31 +05:30
globals()[func.__name__]=func
print(f'Function {func.__name__} enabled in add_calculation.')
@property
def simulation_setup_files(self):
"""Simulation setup files used to generate the Result object."""
files = []
with h5py.File(self.fname,'r') as f_in:
f_in['setup'].visititems(lambda name,obj: files.append(name) if isinstance(obj,h5py.Dataset) else None)
return files
2022-04-02 03:34:50 +05:30
@property
def incs(self):
return [int(i.split(prefix_inc)[-1]) for i in self.increments]
2022-04-02 03:34:50 +05:30
2020-07-31 20:20:01 +05:30
@property
def coordinates0_point(self) -> np.ndarray:
"""Initial/undeformed cell center coordinates."""
2020-02-21 12:15:05 +05:30
if self.structured:
return grid_filters.coordinates0_point(self.cells,self.size,self.origin).reshape(-1,3,order='F')
2020-02-21 12:15:05 +05:30
else:
2020-02-21 22:17:47 +05:30
with h5py.File(self.fname,'r') as f:
2021-04-26 03:52:37 +05:30
return f['geometry/x_p'][()]
2020-02-21 12:15:05 +05:30
2020-07-31 20:20:01 +05:30
@property
def coordinates0_node(self) -> np.ndarray:
"""Initial/undeformed nodal coordinates."""
2020-04-22 11:10:02 +05:30
if self.structured:
return grid_filters.coordinates0_node(self.cells,self.size,self.origin).reshape(-1,3,order='F')
2020-04-22 11:10:02 +05:30
else:
with h5py.File(self.fname,'r') as f:
return f['geometry/x_n'][()]
@property
def geometry0(self) -> VTK:
"""Initial/undeformed geometry."""
if self.structured:
return VTK.from_image_data(self.cells,self.size,self.origin)
else:
with h5py.File(self.fname,'r') as f:
return VTK.from_unstructured_grid(f['/geometry/x_n'][()],
f['/geometry/T_c'][()]-1,
f['/geometry/T_c'].attrs['VTK_TYPE'] if h5py3 else \
f['/geometry/T_c'].attrs['VTK_TYPE'].decode())
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-02 14:48:35 +05:30
def _add_absolute(x: Dict[str, Any]) -> Dict[str, Any]:
return {
'data': np.abs(x['data']),
2020-06-25 01:04:51 +05:30
'label': f'|{x["label"]}|',
'meta': {
2021-03-25 23:52:59 +05:30
'unit': x['meta']['unit'],
'description': f"absolute value of {x['label']} ({x['meta']['description']})",
'creator': 'add_absolute'
}
}
def add_absolute(self, x: str):
2020-02-21 12:15:05 +05:30
"""
Add absolute value.
Parameters
----------
x : str
Name of scalar, vector, or tensor dataset to take absolute value of.
2020-02-21 12:15:05 +05:30
"""
self._add_generic_pointwise(self._add_absolute,{'x':x})
2020-02-21 12:15:05 +05:30
@staticmethod
def _add_calculation(**kwargs) -> Dict[str, Any]:
formula = kwargs['formula']
for d in re.findall(r'#(.*?)#',formula):
2020-06-25 01:04:51 +05:30
formula = formula.replace(f'#{d}#',f"kwargs['{d}']['data']")
2021-08-17 01:01:17 +05:30
data = eval(formula)
if not hasattr(data,'shape') or data.shape[0] != kwargs[d]['data'].shape[0]:
raise ValueError('"{}" results in invalid shape'.format(kwargs['formula']))
2020-02-21 12:15:05 +05:30
return {
2021-08-17 01:01:17 +05:30
'data': data,
'label': kwargs['label'],
'meta': {
2021-03-25 23:52:59 +05:30
'unit': kwargs['unit'],
'description': f"{kwargs['description']} (formula: {kwargs['formula']})",
'creator': 'add_calculation'
}
}
def add_calculation(self,
formula: str,
name: str,
unit: str = 'n/a',
2022-11-23 02:56:15 +05:30
description: Optional[str] = None):
2020-02-21 12:15:05 +05:30
"""
Add result of a general formula.
Parameters
----------
formula : str
2021-04-25 11:17:00 +05:30
Formula to calculate resulting dataset.
Existing datasets are referenced by '#TheirName#'.
name : str
Name of resulting dataset.
2020-02-21 12:15:05 +05:30
unit : str, optional
2020-03-19 12:15:31 +05:30
Physical unit of the result.
2020-02-21 12:15:05 +05:30
description : str, optional
2020-03-19 12:15:31 +05:30
Human-readable description of the result.
2020-02-21 12:15:05 +05:30
Examples
--------
Add total dislocation density, i.e. the sum of mobile dislocation
density 'rho_mob' and dislocation dipole density 'rho_dip' over
all slip systems:
>>> import damask
>>> r = damask.Result('my_file.hdf5')
2021-04-25 11:17:00 +05:30
>>> r.add_calculation('np.sum(#rho_mob#,axis=1)','rho_mob_total',
... '1/m²','total mobile dislocation density')
2021-04-25 21:04:41 +05:30
>>> r.add_calculation('np.sum(#rho_dip#,axis=1)','rho_dip_total',
... '1/m²','total dislocation dipole density')
2021-04-25 11:17:00 +05:30
>>> r.add_calculation('#rho_dip_total#+#rho_mob_total','rho_total',
... '1/m²','total dislocation density')
Add Mises equivalent of the Cauchy stress without storage of
intermediate results. Define a user function for better readability:
>>> import damask
>>> def equivalent_stress(F,P):
... sigma = damask.mechanics.stress_Cauchy(F=F,P=P)
... return damask.mechanics.equivalent_stress_Mises(sigma)
>>> r = damask.Result('my_file.hdf5')
>>> r.enable_user_function(equivalent_stress)
2021-04-25 11:17:00 +05:30
>>> r.add_calculation('equivalent_stress(#F#,#P#)','sigma_vM','Pa',
... 'Mises equivalent of the Cauchy stress')
2020-02-21 12:15:05 +05:30
"""
2022-05-23 11:31:17 +05:30
dataset_mapping = {d:d for d in set(re.findall(r'#(.*?)#',formula))} # datasets used in the formula
args = {'formula':formula,'label':name,'unit':unit,'description':description}
self._add_generic_pointwise(self._add_calculation,dataset_mapping,args)
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_stress_Cauchy(P: Dict[str, Any], F: Dict[str, Any]) -> Dict[str, Any]:
return {
2020-11-18 03:26:22 +05:30
'data': mechanics.stress_Cauchy(P['data'],F['data']),
'label': 'sigma',
'meta': {
2021-03-25 23:52:59 +05:30
'unit': P['meta']['unit'],
'description': "Cauchy stress calculated "
f"from {P['label']} ({P['meta']['description']})"
f" and {F['label']} ({F['meta']['description']})",
'creator': 'add_stress_Cauchy'
}
}
def add_stress_Cauchy(self,
P: str = 'P',
F: str = 'F'):
2020-02-21 12:15:05 +05:30
"""
Add Cauchy stress calculated from first Piola-Kirchhoff stress and deformation gradient.
2020-02-21 12:15:05 +05:30
Parameters
----------
P : str, optional
Name of the dataset containing the first Piola-Kirchhoff stress. Defaults to 'P'.
2020-02-21 12:15:05 +05:30
F : str, optional
Name of the dataset containing the deformation gradient. Defaults to 'F'.
2020-02-21 12:15:05 +05:30
"""
2020-11-18 03:26:22 +05:30
self._add_generic_pointwise(self._add_stress_Cauchy,{'P':P,'F':F})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_determinant(T: Dict[str, Any]) -> Dict[str, Any]:
return {
'data': np.linalg.det(T['data']),
2020-06-25 01:04:51 +05:30
'label': f"det({T['label']})",
'meta': {
2021-03-25 23:52:59 +05:30
'unit': T['meta']['unit'],
'description': f"determinant of tensor {T['label']} ({T['meta']['description']})",
'creator': 'add_determinant'
}
}
def add_determinant(self, T: str):
2020-02-21 12:15:05 +05:30
"""
Add the determinant of a tensor.
Parameters
----------
T : str
Name of tensor dataset.
Examples
--------
Add the determinant of plastic deformation gradient 'F_p':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
>>> r.add_determinant('F_p')
2020-02-21 12:15:05 +05:30
"""
self._add_generic_pointwise(self._add_determinant,{'T':T})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_deviator(T: Dict[str, Any]) -> Dict[str, Any]:
return {
2020-11-19 19:08:54 +05:30
'data': tensor.deviatoric(T['data']),
2020-06-25 01:04:51 +05:30
'label': f"s_{T['label']}",
'meta': {
2021-03-25 23:52:59 +05:30
'unit': T['meta']['unit'],
'description': f"deviator of tensor {T['label']} ({T['meta']['description']})",
'creator': 'add_deviator'
}
}
def add_deviator(self, T: str):
2020-02-21 12:15:05 +05:30
"""
Add the deviatoric part of a tensor.
Parameters
----------
T : str
Name of tensor dataset.
2020-02-21 12:15:05 +05:30
Examples
--------
Add the deviatoric part of Cauchy stress 'sigma':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
>>> r.add_deviator('sigma')
2020-02-21 12:15:05 +05:30
"""
self._add_generic_pointwise(self._add_deviator,{'T':T})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_eigenvalue(T_sym: Dict[str, Any], eigenvalue: Literal['max, mid, min']) -> Dict[str, Any]:
if eigenvalue == 'max':
2021-03-25 23:52:59 +05:30
label,p = 'maximum',2
elif eigenvalue == 'mid':
2021-03-25 23:52:59 +05:30
label,p = 'intermediate',1
elif eigenvalue == 'min':
2021-03-25 23:52:59 +05:30
label,p = 'minimum',0
else:
raise ValueError(f'invalid eigenvalue: {eigenvalue}')
return {
'data': tensor.eigenvalues(T_sym['data'])[:,p],
2020-06-25 01:04:51 +05:30
'label': f"lambda_{eigenvalue}({T_sym['label']})",
'meta' : {
2021-03-25 23:52:59 +05:30
'unit': T_sym['meta']['unit'],
'description': f"{label} eigenvalue of {T_sym['label']} ({T_sym['meta']['description']})",
'creator': 'add_eigenvalue'
}
}
def add_eigenvalue(self,
T_sym: str,
2022-05-09 03:51:53 +05:30
eigenvalue: Literal['max', 'mid', 'min'] = 'max'):
2020-02-21 12:15:05 +05:30
"""
Add eigenvalues of symmetric tensor.
Parameters
----------
2020-03-03 03:44:59 +05:30
T_sym : str
Name of symmetric tensor dataset.
eigenvalue : {'max', 'mid', 'min'}, optional
Eigenvalue. Defaults to 'max'.
2020-02-21 12:15:05 +05:30
Examples
--------
Add the minimum eigenvalue of Cauchy stress 'sigma':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
>>> r.add_eigenvalue('sigma','min')
2020-02-21 12:15:05 +05:30
"""
self._add_generic_pointwise(self._add_eigenvalue,{'T_sym':T_sym},{'eigenvalue':eigenvalue})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_eigenvector(T_sym: Dict[str, Any], eigenvalue: Literal['max', 'mid', 'min']) -> Dict[str, Any]:
if eigenvalue == 'max':
label,p = 'maximum',2
elif eigenvalue == 'mid':
label,p = 'intermediate',1
elif eigenvalue == 'min':
label,p = 'minimum',0
2022-05-09 03:51:53 +05:30
else:
raise ValueError(f'invalid eigenvalue: {eigenvalue}')
2022-05-09 03:51:53 +05:30
return {
'data': tensor.eigenvectors(T_sym['data'])[:,p],
2020-06-25 01:04:51 +05:30
'label': f"v_{eigenvalue}({T_sym['label']})",
'meta' : {
2021-03-25 23:52:59 +05:30
'unit': '1',
'description': f"eigenvector corresponding to {label} eigenvalue"
f" of {T_sym['label']} ({T_sym['meta']['description']})",
'creator': 'add_eigenvector'
}
2020-06-25 01:04:51 +05:30
}
def add_eigenvector(self,
T_sym: str,
2022-05-09 03:51:53 +05:30
eigenvalue: Literal['max', 'mid', 'min'] = 'max'):
2020-02-21 12:15:05 +05:30
"""
Add eigenvector of symmetric tensor.
2020-02-21 12:15:05 +05:30
Parameters
----------
2020-03-03 03:44:59 +05:30
T_sym : str
Name of symmetric tensor dataset.
eigenvalue : {'max', 'mid', 'min'}, optional
2021-04-06 21:09:44 +05:30
Eigenvalue to which the eigenvector corresponds.
Defaults to 'max'.
2020-02-21 12:15:05 +05:30
"""
self._add_generic_pointwise(self._add_eigenvector,{'T_sym':T_sym},{'eigenvalue':eigenvalue})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_IPF_color(l: FloatSequence, q: Dict[str, Any]) -> Dict[str, Any]:
m = util.scale_to_coprime(np.array(l))
lattice = q['meta']['lattice']
o = Orientation(rotation = q['data'],lattice=lattice)
return {
'data': np.uint8(o.IPF_color(l)*255),
'label': 'IPFcolor_({} {} {})'.format(*m),
'meta' : {
2021-03-25 23:52:59 +05:30
'unit': '8-bit RGB',
'lattice': q['meta']['lattice'],
'description': 'Inverse Pole Figure (IPF) colors along sample direction ({} {} {})'.format(*m),
2021-03-25 23:52:59 +05:30
'creator': 'add_IPF_color'
}
}
def add_IPF_color(self,
2022-05-09 03:51:53 +05:30
l: FloatSequence,
q: str = 'O'):
2020-02-21 12:15:05 +05:30
"""
Add RGB color tuple of inverse pole figure (IPF) color.
Parameters
----------
2020-02-21 22:12:01 +05:30
l : numpy.array of shape (3)
2020-03-19 12:15:31 +05:30
Lab frame direction for inverse pole figure.
q : str, optional
Name of the dataset containing the crystallographic orientation as quaternions.
2020-12-02 19:15:47 +05:30
Defaults to 'O'.
2020-02-21 12:15:05 +05:30
Examples
--------
Add the IPF color along [0,1,1] for orientation 'O':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
>>> r.add_IPF_color(np.array([0,1,1]))
2020-02-21 12:15:05 +05:30
"""
2020-07-01 01:13:57 +05:30
self._add_generic_pointwise(self._add_IPF_color,{'q':q},{'l':l})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-02 14:48:35 +05:30
def _add_maximum_shear(T_sym: Dict[str, Any]) -> Dict[str, Any]:
return {
2020-03-03 03:44:59 +05:30
'data': mechanics.maximum_shear(T_sym['data']),
2020-06-25 01:04:51 +05:30
'label': f"max_shear({T_sym['label']})",
'meta': {
2021-03-25 23:52:59 +05:30
'unit': T_sym['meta']['unit'],
'description': f"maximum shear component of {T_sym['label']} ({T_sym['meta']['description']})",
'creator': 'add_maximum_shear'
}
2020-02-21 12:15:05 +05:30
}
def add_maximum_shear(self, T_sym: str):
2020-02-21 12:15:05 +05:30
"""
Add maximum shear components of symmetric tensor.
Parameters
----------
2020-03-03 03:44:59 +05:30
T_sym : str
Name of symmetric tensor dataset.
2020-02-21 12:15:05 +05:30
"""
2020-03-03 03:44:59 +05:30
self._add_generic_pointwise(self._add_maximum_shear,{'T_sym':T_sym})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_equivalent_Mises(T_sym: Dict[str, Any], kind: str) -> Dict[str, Any]:
2020-11-06 04:17:37 +05:30
k = kind
if k is None:
2021-03-25 23:52:59 +05:30
if T_sym['meta']['unit'] == '1':
k = 'strain'
2021-03-25 23:52:59 +05:30
elif T_sym['meta']['unit'] == 'Pa':
k = 'stress'
if k not in ['stress', 'strain']:
raise ValueError(f'invalid von Mises kind "{kind}"')
2020-02-21 12:15:05 +05:30
return {
2020-11-18 03:26:22 +05:30
'data': (mechanics.equivalent_strain_Mises if k=='strain' else \
mechanics.equivalent_stress_Mises)(T_sym['data']),
2020-06-25 01:04:51 +05:30
'label': f"{T_sym['label']}_vM",
'meta': {
2021-03-25 23:52:59 +05:30
'unit': T_sym['meta']['unit'],
'description': f"Mises equivalent {k} of {T_sym['label']} ({T_sym['meta']['description']})",
'creator': 'add_Mises'
}
}
def add_equivalent_Mises(self,
T_sym: str,
2022-11-23 02:56:15 +05:30
kind: Optional[str] = None):
2020-02-21 12:15:05 +05:30
"""
Add the equivalent Mises stress or strain of a symmetric tensor.
Parameters
----------
2020-03-03 03:44:59 +05:30
T_sym : str
Name of symmetric tensorial stress or strain dataset.
kind : {'stress', 'strain', None}, optional
Kind of the von Mises equivalent. Defaults to None, in which case
2021-04-06 21:09:44 +05:30
it is selected based on the unit of the dataset ('1' -> strain, 'Pa' -> stress).
2020-02-21 12:15:05 +05:30
Examples
--------
Add the Mises equivalent of the Cauchy stress 'sigma':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
>>> r.add_equivalent_Mises('sigma')
Add the Mises equivalent of the spatial logarithmic strain 'epsilon_V^0.0(F)':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
>>> r.add_equivalent_Mises('epsilon_V^0.0(F)')
2020-02-21 12:15:05 +05:30
"""
2020-11-18 03:26:22 +05:30
self._add_generic_pointwise(self._add_equivalent_Mises,{'T_sym':T_sym},{'kind':kind})
2020-02-21 12:15:05 +05:30
@staticmethod
def _add_norm(x: Dict[str, Any], ord: Union[int, float, Literal['fro', 'nuc']]) -> Dict[str, Any]:
o = ord
if len(x['data'].shape) == 2:
axis: Union[int, Tuple[int, int]] = 1
t = 'vector'
if o is None: o = 2
elif len(x['data'].shape) == 3:
axis = (1,2)
t = 'tensor'
if o is None: o = 'fro'
else:
raise ValueError(f'invalid shape of {x["label"]}')
2020-02-21 12:15:05 +05:30
return {
'data': np.linalg.norm(x['data'],ord=o,axis=axis,keepdims=True),
2020-06-25 01:04:51 +05:30
'label': f"|{x['label']}|_{o}",
'meta': {
2021-03-25 23:52:59 +05:30
'unit': x['meta']['unit'],
'description': f"{o}-norm of {t} {x['label']} ({x['meta']['description']})",
'creator': 'add_norm'
}
}
def add_norm(self,
x: str,
2022-11-23 02:56:15 +05:30
ord: Union[None, int, float, Literal['fro', 'nuc']] = None):
2020-02-21 12:15:05 +05:30
"""
Add the norm of vector or tensor.
Parameters
----------
x : str
Name of vector or tensor dataset.
2021-04-06 21:09:44 +05:30
ord : {non-zero int, inf, -inf, 'fro', 'nuc'}, optional
2020-03-19 12:15:31 +05:30
Order of the norm. inf means NumPys inf object. For details refer to numpy.linalg.norm.
2020-02-21 12:15:05 +05:30
"""
self._add_generic_pointwise(self._add_norm,{'x':x},{'ord':ord})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_stress_second_Piola_Kirchhoff(P: Dict[str, Any], F: Dict[str, Any]) -> Dict[str, Any]:
return {
2020-11-18 03:26:22 +05:30
'data': mechanics.stress_second_Piola_Kirchhoff(P['data'],F['data']),
'label': 'S',
'meta': {
2021-03-25 23:52:59 +05:30
'unit': P['meta']['unit'],
'description': "second Piola-Kirchhoff stress calculated "
f"from {P['label']} ({P['meta']['description']})"
f" and {F['label']} ({F['meta']['description']})",
'creator': 'add_stress_second_Piola_Kirchhoff'
}
}
def add_stress_second_Piola_Kirchhoff(self,
P: str = 'P',
F: str = 'F'):
2020-02-21 12:15:05 +05:30
"""
2020-06-25 01:04:51 +05:30
Add second Piola-Kirchhoff stress calculated from first Piola-Kirchhoff stress and deformation gradient.
2020-02-21 12:15:05 +05:30
Parameters
----------
P : str, optional
Name of first Piola-Kirchhoff stress dataset. Defaults to 'P'.
2020-02-21 12:15:05 +05:30
F : str, optional
Name of deformation gradient dataset. Defaults to 'F'.
2020-02-21 12:15:05 +05:30
2021-04-25 21:04:41 +05:30
Notes
-----
The definition of the second Piola-Kirchhoff stress (S = [F^-1 P]_sym)
follows the standard definition in nonlinear continuum mechanics.
As such, no intermediate configuration, for instance that reached by F_p,
is taken into account.
2021-04-25 21:04:41 +05:30
2020-02-21 12:15:05 +05:30
"""
2020-11-18 03:26:22 +05:30
self._add_generic_pointwise(self._add_stress_second_Piola_Kirchhoff,{'P':P,'F':F})
2020-02-21 12:15:05 +05:30
2021-07-18 21:33:36 +05:30
@staticmethod
def _add_pole(q: Dict[str, Any],
2022-05-02 14:48:35 +05:30
uvw: FloatSequence,
hkl: FloatSequence,
with_symmetry: bool,
normalize: bool) -> Dict[str, Any]:
2021-07-18 21:33:36 +05:30
c = q['meta']['c/a'] if 'c/a' in q['meta'] else 1
2022-05-11 10:32:20 +05:30
brackets = ['[]','()','⟨⟩','{}'][(uvw is None)*1+with_symmetry*2]
label = 'p^' + '{}{} {} {}{}'.format(brackets[0],
*(uvw if uvw else hkl),
brackets[-1],)
2022-05-13 13:23:44 +05:30
ori = Orientation(q['data'],lattice=q['meta']['lattice'],a=1,c=c)
2021-07-18 21:33:36 +05:30
return {
2022-05-13 13:23:44 +05:30
'data': ori.to_pole(uvw=uvw,hkl=hkl,with_symmetry=with_symmetry,normalize=normalize),
'label': label,
2021-07-18 21:33:36 +05:30
'meta' : {
'unit': '1',
2022-05-13 13:23:44 +05:30
'description': f'{"normalized " if normalize else ""}lab frame vector along lattice ' \
+ ('direction' if uvw is not None else 'plane') \
+ ('s' if with_symmetry else ''),
'creator': 'add_pole'
2021-07-18 21:33:36 +05:30
}
}
def add_pole(self,
q: str = 'O',
*,
2022-11-23 02:56:15 +05:30
uvw: Optional[FloatSequence] = None,
hkl: Optional[FloatSequence] = None,
with_symmetry: bool = False,
normalize: bool = True):
2021-07-18 21:33:36 +05:30
"""
Add lab frame vector along lattice direction [uvw] or plane normal (hkl).
Parameters
----------
q : str, optional
2021-07-18 21:33:36 +05:30
Name of the dataset containing the crystallographic orientation as quaternions.
Defaults to 'O'.
2022-05-09 03:51:53 +05:30
uvw|hkl : numpy.ndarray of shape (3)
2021-07-18 21:33:36 +05:30
Miller indices of crystallographic direction or plane normal.
with_symmetry : bool, optional
Calculate all N symmetrically equivalent vectors.
2022-05-13 13:23:44 +05:30
Defaults to True.
normalize : bool, optional
Normalize output vector.
Defaults to True.
2021-07-18 21:33:36 +05:30
"""
2022-05-13 13:23:44 +05:30
self._add_generic_pointwise(self._add_pole,
{'q':q},
{'uvw':uvw,'hkl':hkl,'with_symmetry':with_symmetry,'normalize':normalize})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_rotation(F: Dict[str, Any]) -> Dict[str, Any]:
return {
2020-11-20 03:16:52 +05:30
'data': mechanics.rotation(F['data']).as_matrix(),
2020-06-25 01:04:51 +05:30
'label': f"R({F['label']})",
'meta': {
2021-03-25 23:52:59 +05:30
'unit': F['meta']['unit'],
'description': f"rotational part of {F['label']} ({F['meta']['description']})",
'creator': 'add_rotation'
}
2020-02-21 12:15:05 +05:30
}
def add_rotation(self, F: str):
2020-02-21 12:15:05 +05:30
"""
Add rotational part of a deformation gradient.
Parameters
----------
F : str
Name of deformation gradient dataset.
Examples
--------
Add the rotational part of deformation gradient 'F':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
>>> r.add_rotation('F')
2020-02-21 12:15:05 +05:30
"""
2020-11-20 03:16:52 +05:30
self._add_generic_pointwise(self._add_rotation,{'F':F})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_spherical(T: Dict[str, Any]) -> Dict[str, Any]:
return {
2020-11-19 19:08:54 +05:30
'data': tensor.spherical(T['data'],False),
2020-06-25 01:04:51 +05:30
'label': f"p_{T['label']}",
'meta': {
2021-03-25 23:52:59 +05:30
'unit': T['meta']['unit'],
'description': f"spherical component of tensor {T['label']} ({T['meta']['description']})",
'creator': 'add_spherical'
}
}
def add_spherical(self, T: str):
2020-02-21 12:15:05 +05:30
"""
Add the spherical (hydrostatic) part of a tensor.
Parameters
----------
T : str
Name of tensor dataset.
Examples
--------
Add the hydrostatic part of the Cauchy stress 'sigma':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
>>> r.add_spherical('sigma')
2020-02-21 12:15:05 +05:30
"""
self._add_generic_pointwise(self._add_spherical,{'T':T})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_strain(F: Dict[str, Any], t: Literal['V', 'U'], m: float) -> Dict[str, Any]:
return {
2020-11-16 05:42:23 +05:30
'data': mechanics.strain(F['data'],t,m),
2020-06-25 01:04:51 +05:30
'label': f"epsilon_{t}^{m}({F['label']})",
'meta': {
2021-03-25 23:52:59 +05:30
'unit': F['meta']['unit'],
'description': f"strain tensor of {F['label']} ({F['meta']['description']})",
'creator': 'add_strain'
}
}
def add_strain(self,
F: str = 'F',
t: Literal['V', 'U'] = 'V',
m: float = 0.0):
2020-02-21 12:15:05 +05:30
"""
Add strain tensor of a deformation gradient.
2020-02-21 12:15:05 +05:30
2021-04-06 21:09:44 +05:30
For details, see damask.mechanics.strain.
2020-02-21 12:15:05 +05:30
Parameters
----------
F : str, optional
Name of deformation gradient dataset. Defaults to 'F'.
2021-04-06 21:09:44 +05:30
t : {'V', 'U'}, optional
Type of the polar decomposition, 'V' for left stretch tensor and 'U' for right stretch tensor.
Defaults to 'V'.
2020-02-21 12:15:05 +05:30
m : float, optional
2021-04-06 21:09:44 +05:30
Order of the strain calculation. Defaults to 0.0.
2020-02-21 12:15:05 +05:30
Examples
--------
Add the Biot strain based on the deformation gradient 'F':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
2022-04-12 15:32:00 +05:30
>>> r.add_strain(t='U',m=0.5)
Add the plastic Euler-Almansi strain based on the
plastic deformation gradient 'F_p':
>>> import damask
>>> r = damask.Result('my_file.hdf5')
2022-04-12 15:32:00 +05:30
>>> r.add_strain('F_p','V',-1)
2020-02-21 12:15:05 +05:30
"""
2020-11-16 05:42:23 +05:30
self._add_generic_pointwise(self._add_strain,{'F':F},{'t':t,'m':m})
2020-02-21 12:15:05 +05:30
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_stretch_tensor(F: Dict[str, Any], t: str) -> Dict[str, Any]:
return {
'data': (mechanics.stretch_left if t.upper() == 'V' else mechanics.stretch_right)(F['data']),
2020-06-25 01:04:51 +05:30
'label': f"{t}({F['label']})",
'meta': {
2021-03-25 23:52:59 +05:30
'unit': F['meta']['unit'],
2021-05-28 16:50:56 +05:30
'description': f"{'left' if t.upper() == 'V' else 'right'} stretch tensor "\
+f"of {F['label']} ({F['meta']['description']})", # noqa
2021-03-25 23:52:59 +05:30
'creator': 'add_stretch_tensor'
}
}
def add_stretch_tensor(self,
F: str = 'F',
t: Literal['V', 'U'] = 'V'):
2020-02-21 12:15:05 +05:30
"""
Add stretch tensor of a deformation gradient.
2020-02-21 12:15:05 +05:30
Parameters
----------
F : str, optional
Name of deformation gradient dataset. Defaults to 'F'.
2021-04-06 21:09:44 +05:30
t : {'V', 'U'}, optional
Type of the polar decomposition, 'V' for left stretch tensor and 'U' for right stretch tensor.
Defaults to 'V'.
2020-02-21 12:15:05 +05:30
"""
self._add_generic_pointwise(self._add_stretch_tensor,{'F':F},{'t':t})
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_curl(f: Dict[str, Any], size: np.ndarray) -> Dict[str, Any]:
return {
'data': grid_filters.curl(size,f['data']),
'label': f"curl({f['label']})",
'meta': {
'unit': f['meta']['unit']+'/m',
'description': f"curl of {f['label']} ({f['meta']['description']})",
'creator': 'add_curl'
}
}
def add_curl(self, f: str):
"""
Add curl of a field.
Parameters
----------
f : str
Name of vector or tensor field dataset.
2021-05-29 00:27:44 +05:30
Notes
-----
This function is only available for structured grids,
i.e. results from the grid solver.
"""
2021-05-29 00:27:44 +05:30
self._add_generic_grid(self._add_curl,{'f':f},{'size':self.size})
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_divergence(f: Dict[str, Any], size: np.ndarray) -> Dict[str, Any]:
return {
'data': grid_filters.divergence(size,f['data']),
'label': f"divergence({f['label']})",
'meta': {
'unit': f['meta']['unit']+'/m',
'description': f"divergence of {f['label']} ({f['meta']['description']})",
'creator': 'add_divergence'
}
}
def add_divergence(self, f: str):
"""
Add divergence of a field.
Parameters
----------
f : str
Name of vector or tensor field dataset.
2021-05-29 00:27:44 +05:30
Notes
-----
This function is only available for structured grids,
i.e. results from the grid solver.
"""
2021-05-29 00:27:44 +05:30
self._add_generic_grid(self._add_divergence,{'f':f},{'size':self.size})
@staticmethod
2022-05-09 03:51:53 +05:30
def _add_gradient(f: Dict[str, Any], size: np.ndarray) -> Dict[str, Any]:
return {
'data': grid_filters.gradient(size,f['data'] if len(f['data'].shape) == 4 else \
f['data'].reshape(f['data'].shape+(1,))),
'label': f"gradient({f['label']})",
'meta': {
'unit': f['meta']['unit']+'/m',
'description': f"gradient of {f['label']} ({f['meta']['description']})",
'creator': 'add_gradient'
}
}
def add_gradient(self, f: str):
"""
Add gradient of a field.
Parameters
----------
f : str
Name of scalar or vector field dataset.
2021-05-29 00:27:44 +05:30
Notes
-----
This function is only available for structured grids,
i.e. results from the grid solver.
"""
2021-05-29 00:27:44 +05:30
self._add_generic_grid(self._add_gradient,{'f':f},{'size':self.size})
def _add_generic_grid(self,
func: Callable,
datasets: Dict[str, str],
args: Dict[str, str] = {},
2022-05-02 14:48:35 +05:30
constituents = None):
"""
General function to add data on a regular grid.
Parameters
----------
func : function
Callback function that calculates a new dataset from one or
more datasets per HDF5 group.
datasets : dictionary
Details of the datasets to be used:
{arg (name to which the data is passed in func): label (in HDF5 file)}.
args : dictionary, optional
Arguments parsed to func.
"""
2021-08-18 14:14:04 +05:30
if len(datasets) != 1 or self.N_constituents != 1:
raise NotImplementedError
at_cell_ph,in_data_ph,at_cell_ho,in_data_ho = self._mappings()
increments = self.place(list(datasets.values()),False)
if not increments: raise RuntimeError("received invalid dataset")
with h5py.File(self.fname, 'a') as f:
for increment in increments.items():
for ty in increment[1].items():
for field in ty[1].items():
d: np.ma.MaskedArray = list(field[1].values())[0]
if np.any(d.mask): continue
dataset = {'f':{'data':np.reshape(d.data,tuple(self.cells)+d.data.shape[1:]),
'label':list(datasets.values())[0],
'meta':d.data.dtype.metadata}}
r = func(**dataset,**args)
result = r['data'].reshape((-1,)+r['data'].shape[3:])
for x in self.visible[ty[0]+'s']:
if ty[0] == 'phase':
result1 = result[at_cell_ph[0][x]]
if ty[0] == 'homogenization':
result1 = result[at_cell_ho[x]]
path = '/'.join(['/',increment[0],ty[0],x,field[0]])
h5_dataset = f[path].create_dataset(r['label'],data=result1)
now = datetime.datetime.now().astimezone()
h5_dataset.attrs['created'] = now.strftime('%Y-%m-%d %H:%M:%S%z') if h5py3 else \
now.strftime('%Y-%m-%d %H:%M:%S%z').encode()
for l,v in r['meta'].items():
h5_dataset.attrs[l.lower()]=v if h5py3 else v.encode()
creator = h5_dataset.attrs['creator'] if h5py3 else \
h5_dataset.attrs['creator'].decode()
h5_dataset.attrs['creator'] = f'damask.Result.{creator} v{damask.version}' if h5py3 else \
f'damask.Result.{creator} v{damask.version}'.encode()
def _job_pointwise(self,
group: str,
callback: Callable,
datasets: Dict[str, str],
args: Dict[str, str],
2022-05-09 03:51:53 +05:30
lock: Lock) -> List[Union[None, Any]]:
2020-02-22 04:29:33 +05:30
"""Execute job for _add_generic_pointwise."""
try:
2020-02-22 03:46:25 +05:30
datasets_in = {}
lock.acquire()
with h5py.File(self.fname,'r') as f:
2020-03-12 03:05:58 +05:30
for arg,label in datasets.items():
loc = f[group+'/'+label]
datasets_in[arg]={'data' :loc[()],
'label':label,
'meta': {k:(v.decode() if not h5py3 and type(v) is bytes else v) \
for k,v in loc.attrs.items()}}
2020-02-22 03:46:25 +05:30
lock.release()
r = callback(**datasets_in,**args)
return [group,r]
except Exception as err:
2020-06-25 01:04:51 +05:30
print(f'Error during calculation: {err}.')
return [None,None]
def _add_generic_pointwise(self,
func: Callable,
datasets: Dict[str, Any],
args: Dict[str, Any] = {}):
2020-02-22 03:46:25 +05:30
"""
General function to add pointwise data.
2020-02-22 03:46:25 +05:30
Parameters
----------
callback : function
Callback function that calculates a new dataset from one or
more datasets per HDF5 group.
2020-02-22 03:46:25 +05:30
datasets : dictionary
2021-04-06 21:09:44 +05:30
Details of the datasets to be used:
{arg (name to which the data is passed in func): label (in HDF5 file)}.
2020-02-22 03:46:25 +05:30
args : dictionary, optional
2020-03-19 12:15:31 +05:30
Arguments parsed to func.
2020-02-22 04:29:33 +05:30
2020-02-22 03:46:25 +05:30
"""
pool = mp.Pool(int(os.environ.get('OMP_NUM_THREADS',4)))
2020-06-26 15:15:06 +05:30
lock = mp.Manager().Lock()
2020-02-22 03:46:25 +05:30
2021-04-04 22:02:17 +05:30
groups = []
with h5py.File(self.fname,'r') as f:
for inc in self.visible['increments']:
2021-04-05 11:23:19 +05:30
for ty in ['phase','homogenization']:
for label in self.visible[ty+'s']:
for field in _match(self.visible['fields'],f['/'.join([inc,ty,label])].keys()):
2021-04-05 11:23:19 +05:30
group = '/'.join([inc,ty,label,field])
2021-04-04 22:02:17 +05:30
if set(datasets.values()).issubset(f[group].keys()): groups.append(group)
if len(groups) == 0:
print('No matching dataset found, no data was added.')
return
default_arg = partial(self._job_pointwise,callback=func,datasets=datasets,args=args,lock=lock)
2020-02-22 03:46:25 +05:30
2022-05-23 11:31:17 +05:30
for group,result in util.show_progress(pool.imap_unordered(default_arg,groups),len(groups)):# type: ignore
if not result:
continue
2020-02-22 03:46:25 +05:30
lock.acquire()
with h5py.File(self.fname, 'a') as f:
try:
2021-12-17 15:07:45 +05:30
if not self._protected and '/'.join([group,result['label']]) in f:
2021-05-28 16:50:56 +05:30
dataset = f['/'.join([group,result['label']])]
dataset[...] = result['data']
2021-03-25 23:52:59 +05:30
dataset.attrs['overwritten'] = True
else:
shape = result['data'].shape
2022-05-23 11:31:17 +05:30
if compress := (result['data'].size >= chunk_size*2):
chunks = (chunk_size//np.prod(shape[1:]),)+shape[1:]
else:
chunks = shape
dataset = f[group].create_dataset(result['label'],data=result['data'],
maxshape=shape, chunks=chunks,
compression = 'gzip' if compress else None,
compression_opts = 6 if compress else None,
shuffle=True,fletcher32=True)
2020-05-25 22:20:31 +05:30
now = datetime.datetime.now().astimezone()
2021-03-25 23:52:59 +05:30
dataset.attrs['created'] = now.strftime('%Y-%m-%d %H:%M:%S%z') if h5py3 else \
now.strftime('%Y-%m-%d %H:%M:%S%z').encode()
2021-05-28 16:50:56 +05:30
for l,v in result['meta'].items():
dataset.attrs[l.lower()]=v.encode() if not h5py3 and type(v) is str else v
2021-03-25 23:52:59 +05:30
creator = dataset.attrs['creator'] if h5py3 else \
dataset.attrs['creator'].decode()
2021-04-03 11:38:48 +05:30
dataset.attrs['creator'] = f'damask.Result.{creator} v{damask.version}' if h5py3 else \
f'damask.Result.{creator} v{damask.version}'.encode()
except (OSError,RuntimeError) as err:
2020-06-25 01:04:51 +05:30
print(f'Could not add dataset: {err}.')
2020-02-22 03:46:25 +05:30
lock.release()
pool.close()
pool.join()
2020-02-21 12:15:05 +05:30
def _mappings(self):
"""Mappings to place data spatially."""
with h5py.File(self.fname,'r') as f:
at_cell_ph = []
in_data_ph = []
for c in range(self.N_constituents):
at_cell_ph.append({label: np.where(self.phase[:,c] == label)[0] \
for label in self.visible['phases']})
in_data_ph.append({label: f['/'.join(['cell_to','phase'])]['entry'][at_cell_ph[c][label]][:,c] \
for label in self.visible['phases']})
at_cell_ho = {label: np.where(self.homogenization[:] == label)[0] \
for label in self.visible['homogenizations']}
in_data_ho = {label: f['/'.join(['cell_to','homogenization'])]['entry'][at_cell_ho[label]] \
for label in self.visible['homogenizations']}
return at_cell_ph,in_data_ph,at_cell_ho,in_data_ho
def get(self,
output: Union[str, List[str]] = '*',
flatten: bool = True,
2022-11-11 11:33:14 +05:30
prune: bool = True) -> Optional[Dict[str,Any]]:
"""
Collect data per phase/homogenization reflecting the group/folder structure in the DADF5 file.
Parameters
----------
output : (list of) str, optional
Names of the datasets to read.
Defaults to '*', in which case all datasets are read.
flatten : bool, optional
Remove singular levels of the folder hierarchy.
This might be beneficial in case of single increment,
phase/homogenization, or field. Defaults to True.
prune : bool, optional
Remove branches with no data. Defaults to True.
Returns
-------
data : dict of numpy.ndarray
Datasets structured by phase/homogenization and according to selected view.
"""
2022-11-11 11:33:14 +05:30
r: Dict[str,Any] = {}
with h5py.File(self.fname,'r') as f:
for inc in util.show_progress(self.visible['increments']):
r[inc] = {'phase':{},'homogenization':{},'geometry':{}}
for out in _match(output,f['/'.join([inc,'geometry'])].keys()):
r[inc]['geometry'][out] = _read(f['/'.join([inc,'geometry',out])])
for ty in ['phase','homogenization']:
for label in self.visible[ty+'s']:
r[inc][ty][label] = {}
for field in _match(self.visible['fields'],f['/'.join([inc,ty,label])].keys()):
r[inc][ty][label][field] = {}
for out in _match(output,f['/'.join([inc,ty,label,field])].keys()):
r[inc][ty][label][field][out] = _read(f['/'.join([inc,ty,label,field,out])])
if prune: r = util.dict_prune(r)
if flatten: r = util.dict_flatten(r)
return None if (type(r) == dict and r == {}) else r
def place(self,
output: Union[str, List[str]] = '*',
flatten: bool = True,
prune: bool = True,
2022-11-23 02:56:15 +05:30
constituents: Optional[IntSequence] = None,
fill_float: float = np.nan,
2022-11-11 11:33:14 +05:30
fill_int: int = 0) -> Optional[Dict[str,Any]]:
"""
Merge data into spatial order that is compatible with the damask.VTK geometry representation.
The returned data structure reflects the group/folder structure in the DADF5 file.
Multi-phase data is fused into a single output.
`place` is equivalent to `get` if only one phase/homogenization
and one constituent is present.
Parameters
----------
output : (list of) str, optional
Names of the datasets to read.
Defaults to '*', in which case all visible datasets are placed.
flatten : bool, optional
Remove singular levels of the folder hierarchy.
This might be beneficial in case of single increment or field.
Defaults to True.
prune : bool, optional
Remove branches with no data. Defaults to True.
constituents : (list of) int, optional
Constituents to consider.
Defaults to None, in which case all constituents are considered.
fill_float : float, optional
Fill value for non-existent entries of floating point type.
Defaults to NaN.
fill_int : int, optional
Fill value for non-existent entries of integer type.
Defaults to 0.
Returns
-------
data : dict of numpy.ma.MaskedArray
Datasets structured by spatial position and according to selected view.
"""
2022-11-11 11:33:14 +05:30
r: Dict[str,Any] = {}
2022-11-11 11:33:14 +05:30
constituents_ = map(int,constituents) if isinstance(constituents,Iterable) else \
(range(self.N_constituents) if constituents is None else [constituents]) # type: ignore
suffixes = [''] if self.N_constituents == 1 or isinstance(constituents,int) else \
[f'#{c}' for c in constituents_]
at_cell_ph,in_data_ph,at_cell_ho,in_data_ho = self._mappings()
with h5py.File(self.fname,'r') as f:
for inc in util.show_progress(self.visible['increments']):
r[inc] = {'phase':{},'homogenization':{},'geometry':{}}
for out in _match(output,f['/'.join([inc,'geometry'])].keys()):
r[inc]['geometry'][out] = ma.array(_read(f['/'.join([inc,'geometry',out])]),fill_value = fill_float)
for ty in ['phase','homogenization']:
for label in self.visible[ty+'s']:
for field in _match(self.visible['fields'],f['/'.join([inc,ty,label])].keys()):
if field not in r[inc][ty].keys():
r[inc][ty][field] = {}
for out in _match(output,f['/'.join([inc,ty,label,field])].keys()):
data = ma.array(_read(f['/'.join([inc,ty,label,field,out])]))
if ty == 'phase':
if out+suffixes[0] not in r[inc][ty][field].keys():
for c,suffix in zip(constituents_,suffixes):
r[inc][ty][field][out+suffix] = \
_empty_like(data,self.N_materialpoints,fill_float,fill_int)
for c,suffix in zip(constituents_,suffixes):
r[inc][ty][field][out+suffix][at_cell_ph[c][label]] = data[in_data_ph[c][label]]
if ty == 'homogenization':
if out not in r[inc][ty][field].keys():
r[inc][ty][field][out] = \
_empty_like(data,self.N_materialpoints,fill_float,fill_int)
r[inc][ty][field][out][at_cell_ho[label]] = data[in_data_ho[label]]
if prune: r = util.dict_prune(r)
if flatten: r = util.dict_flatten(r)
return None if (type(r) == dict and r == {}) else r
def export_XDMF(self,
output: Union[str, List[str]] = '*',
2022-11-23 02:56:15 +05:30
target_dir: Union[None, str, Path] = None,
absolute_path: bool = False):
"""
Write XDMF file to directly visualize data from DADF5 file.
The XDMF format is only supported for structured grids
with single phase and single constituent.
2021-06-17 21:56:37 +05:30
For other cases use `export_VTK`.
2021-04-05 13:59:34 +05:30
Parameters
----------
output : (list of) str, optional
Names of the datasets included in the XDMF file.
2021-04-06 21:09:44 +05:30
Defaults to '*', in which case all datasets are considered.
target_dir : str or pathlib.Path, optional
Directory to save XDMF file. Will be created if non-existent.
absolute_path : bool, optional
Store absolute (instead of relative) path to DADF5 file.
Defaults to False, i.e. the XDMF file expects the
DADF5 file at a stable relative path.
2021-04-05 13:59:34 +05:30
"""
if self.N_constituents != 1 or len(self.phases) != 1 or not self.structured:
2021-04-06 21:09:44 +05:30
raise TypeError('XDMF output requires structured grid with single phase and single constituent.')
attribute_type_map = defaultdict(lambda:'Matrix', ( ((),'Scalar'), ((3,),'Vector'), ((3,3),'Tensor')) )
def number_type_map(dtype):
if dtype in np.sctypes['int']: return 'Int'
if dtype in np.sctypes['uint']: return 'UInt'
if dtype in np.sctypes['float']: return 'Float'
2021-04-06 21:09:44 +05:30
xdmf = ET.Element('Xdmf')
xdmf.attrib = {'Version': '2.0',
'xmlns:xi': 'http://www.w3.org/2001/XInclude'}
2021-04-06 21:09:44 +05:30
domain = ET.SubElement(xdmf, 'Domain')
collection = ET.SubElement(domain, 'Grid')
collection.attrib = {'GridType': 'Collection',
'CollectionType': 'Temporal',
'Name': 'Increments'}
time = ET.SubElement(collection, 'Time')
time.attrib = {'TimeType': 'List'}
time_data = ET.SubElement(time, 'DataItem')
2021-04-05 13:59:34 +05:30
times = [self.times[self.increments.index(i)] for i in self.visible['increments']]
time_data.attrib = {'Format': 'XML',
'NumberType': 'Float',
'Dimensions': f'{len(times)}'}
2021-04-05 13:59:34 +05:30
time_data.text = ' '.join(map(str,times))
attributes = []
data_items = []
hdf5_name = self.fname.name
hdf5_dir = self.fname.parent
xdmf_dir = Path.cwd() if target_dir is None else Path(target_dir)
hdf5_link = (hdf5_dir if absolute_path else Path(os.path.relpath(hdf5_dir,xdmf_dir.resolve())))/hdf5_name
2021-04-05 19:28:10 +05:30
with h5py.File(self.fname,'r') as f:
for inc in self.visible['increments']:
2021-04-06 21:09:44 +05:30
grid = ET.SubElement(collection,'Grid')
2021-04-05 19:28:10 +05:30
grid.attrib = {'GridType': 'Uniform',
'Name': inc}
2021-04-06 21:09:44 +05:30
topology = ET.SubElement(grid, 'Topology')
topology.attrib = {'TopologyType': '3DCoRectMesh',
'Dimensions': '{} {} {}'.format(*(self.cells[::-1]+1))}
2021-04-06 21:09:44 +05:30
geometry = ET.SubElement(grid, 'Geometry')
geometry.attrib = {'GeometryType':'Origin_DxDyDz'}
2021-04-06 21:09:44 +05:30
origin = ET.SubElement(geometry, 'DataItem')
origin.attrib = {'Format': 'XML',
'NumberType': 'Float',
'Dimensions': '3'}
origin.text = "{} {} {}".format(*self.origin[::-1])
2021-04-06 21:09:44 +05:30
delta = ET.SubElement(geometry, 'DataItem')
delta.attrib = {'Format': 'XML',
'NumberType': 'Float',
'Dimensions': '3'}
delta.text="{} {} {}".format(*(self.size/self.cells)[::-1])
attributes.append(ET.SubElement(grid, 'Attribute'))
2021-04-06 21:09:44 +05:30
attributes[-1].attrib = {'Name': 'u / m',
'Center': 'Node',
'AttributeType': 'Vector'}
data_items.append(ET.SubElement(attributes[-1], 'DataItem'))
2021-04-06 21:09:44 +05:30
data_items[-1].attrib = {'Format': 'HDF',
'Precision': '8',
'Dimensions': '{} {} {} 3'.format(*(self.cells[::-1]+1))}
data_items[-1].text = f'{hdf5_link}:/{inc}/geometry/u_n'
2021-04-05 11:23:19 +05:30
for ty in ['phase','homogenization']:
for label in self.visible[ty+'s']:
for field in _match(self.visible['fields'],f['/'.join([inc,ty,label])].keys()):
for out in _match(output,f['/'.join([inc,ty,label,field])].keys()):
2021-04-05 11:23:19 +05:30
name = '/'.join([inc,ty,label,field,out])
shape = f[name].shape[1:]
dtype = f[name].dtype
2021-04-14 10:36:24 +05:30
unit = f[name].attrs['unit'] if h5py3 else \
f[name].attrs['unit'].decode()
attributes.append(ET.SubElement(grid, 'Attribute'))
attributes[-1].attrib = {'Name': '/'.join([ty,field,out])+f' / {unit}',
2021-04-06 21:09:44 +05:30
'Center': 'Cell',
'AttributeType': attribute_type_map[shape]}
data_items.append(ET.SubElement(attributes[-1], 'DataItem'))
2021-04-06 21:09:44 +05:30
data_items[-1].attrib = {'Format': 'HDF',
'NumberType': number_type_map(dtype),
'Precision': f'{dtype.itemsize}',
'Dimensions': '{} {} {} {}'.format(*self.cells[::-1],1 if shape == () else
2021-04-06 21:09:44 +05:30
np.prod(shape))}
data_items[-1].text = f'{hdf5_link}:{name}'
xdmf_dir.mkdir(parents=True,exist_ok=True)
with util.open_text((xdmf_dir/hdf5_name).with_suffix('.xdmf'),'w') as f:
f.write(xml.dom.minidom.parseString(ET.tostring(xdmf).decode()).toprettyxml())
def export_VTK(self,
output: Union[str,List[str]] = '*',
mode: str = 'cell',
2022-11-23 02:56:15 +05:30
constituents: Optional[IntSequence] = None,
target_dir: Union[None, str, Path] = None,
fill_float: float = np.nan,
fill_int: int = 0,
parallel: bool = True):
2020-02-21 12:15:05 +05:30
"""
2021-04-06 21:09:44 +05:30
Export to VTK cell/point data.
2020-02-21 12:15:05 +05:30
One VTK file per visible increment is created.
For point data, the VTK format is poly data (.vtp).
2021-06-19 00:41:01 +05:30
For cell data, either an image (.vti) or unstructured (.vtu) dataset
is written for grid-based or mesh-based simulations, respectively.
2020-02-21 12:15:05 +05:30
Parameters
----------
2021-04-06 21:09:44 +05:30
output : (list of) str, optional
2021-04-24 22:42:44 +05:30
Names of the datasets to export to the VTK file.
Defaults to '*', in which case all visible datasets are exported.
mode : {'cell', 'point'}, optional
2020-03-19 12:15:31 +05:30
Export in cell format or point format.
Defaults to 'cell'.
2021-04-06 21:09:44 +05:30
constituents : (list of) int, optional
Constituents to consider.
Defaults to None, in which case all constituents are considered.
target_dir : str or pathlib.Path, optional
Directory to save VTK files. Will be created if non-existent.
fill_float : float, optional
2021-04-02 15:51:27 +05:30
Fill value for non-existent entries of floating point type.
2021-04-06 21:09:44 +05:30
Defaults to NaN.
fill_int : int, optional
2021-04-02 15:51:27 +05:30
Fill value for non-existent entries of integer type.
Defaults to 0.
parallel : bool, optional
Write VTK files in parallel in a separate background process.
2021-04-05 13:59:34 +05:30
Defaults to True.
2020-02-21 12:15:05 +05:30
"""
if mode.lower()=='cell':
v = self.geometry0
2020-02-21 23:22:58 +05:30
elif mode.lower()=='point':
v = VTK.from_poly_data(self.coordinates0_point)
else:
raise ValueError(f'invalid mode "{mode}"')
2022-05-22 13:38:32 +05:30
v.comments = [util.execution_stamp('Result','export_VTK')]
2020-03-12 03:05:58 +05:30
2022-04-02 03:34:50 +05:30
N_digits = int(np.floor(np.log10(max(1,self.incs[-1]))))+1
2020-02-21 12:15:05 +05:30
2021-04-02 15:51:27 +05:30
constituents_ = constituents if isinstance(constituents,Iterable) else \
2022-05-09 03:51:53 +05:30
(range(self.N_constituents) if constituents is None else [constituents]) # type: ignore
2021-04-02 15:51:27 +05:30
suffixes = [''] if self.N_constituents == 1 or isinstance(constituents,int) else \
[f'#{c}' for c in constituents_]
2021-04-05 19:11:28 +05:30
at_cell_ph,in_data_ph,at_cell_ho,in_data_ho = self._mappings()
2021-04-02 15:51:27 +05:30
vtk_dir = Path.cwd() if target_dir is None else Path(target_dir)
vtk_dir.mkdir(parents=True,exist_ok=True)
2021-04-02 15:51:27 +05:30
with h5py.File(self.fname,'r') as f:
2021-04-26 02:22:13 +05:30
if self.version_minor >= 13:
creator = f.attrs['creator'] if h5py3 else f.attrs['creator'].decode()
created = f.attrs['created'] if h5py3 else f.attrs['created'].decode()
v.comments += [f'{creator} ({created})']
2021-04-02 15:51:27 +05:30
for inc in util.show_progress(self.visible['increments']):
u = _read(f['/'.join([inc,'geometry','u_n' if mode.lower() == 'cell' else 'u_p'])])
2022-05-12 04:24:03 +05:30
v = v.set('u',u)
2021-04-02 15:51:27 +05:30
for ty in ['phase','homogenization']:
for field in self.visible['fields']:
outs: Dict[str, np.ma.core.MaskedArray] = {}
2021-04-02 15:51:27 +05:30
for label in self.visible[ty+'s']:
if field not in f['/'.join([inc,ty,label])].keys(): continue
2021-04-02 15:51:27 +05:30
for out in _match(output,f['/'.join([inc,ty,label,field])].keys()):
data = ma.array(_read(f['/'.join([inc,ty,label,field,out])]))
2021-04-02 15:51:27 +05:30
if ty == 'phase':
if out+suffixes[0] not in outs.keys():
for c,suffix in zip(constituents_,suffixes):
outs[out+suffix] = \
2021-04-06 21:09:44 +05:30
_empty_like(data,self.N_materialpoints,fill_float,fill_int)
2021-04-02 15:51:27 +05:30
for c,suffix in zip(constituents_,suffixes):
outs[out+suffix][at_cell_ph[c][label]] = data[in_data_ph[c][label]]
if ty == 'homogenization':
if out not in outs.keys():
2021-04-06 21:09:44 +05:30
outs[out] = _empty_like(data,self.N_materialpoints,fill_float,fill_int)
2020-03-22 20:43:35 +05:30
2021-04-02 15:51:27 +05:30
outs[out][at_cell_ho[label]] = data[in_data_ho[label]]
2020-03-22 20:43:35 +05:30
2021-04-02 15:51:27 +05:30
for label,dataset in outs.items():
2022-05-12 04:24:03 +05:30
v = v.set(' / '.join(['/'.join([ty,field,label]),dataset.dtype.metadata['unit']]),dataset)
2020-03-22 20:43:35 +05:30
v.save(vtk_dir/f'{self.fname.stem}_inc{inc.split(prefix_inc)[-1].zfill(N_digits)}',
parallel=parallel)
def export_DADF5(self,
fname,
output: Union[str, List[str]] = '*'):
"""
Export visible components into a new DADF5 file.
2022-11-08 17:09:38 +05:30
A DADF5 (DAMASK HDF5) file contains DAMASK results.
Its group/folder structure reflects the layout in material.yaml.
Parameters
----------
fname : str or pathlib.Path
Name of the DADF5 file to be created.
output : (list of) str, optional
Names of the datasets to export.
Defaults to '*', in which case all visible datasets are exported.
"""
if Path(fname).expanduser().absolute() == self.fname:
raise PermissionError(f'cannot overwrite {self.fname}')
with h5py.File(self.fname,'r') as f_in, h5py.File(fname,'w') as f_out:
for k,v in f_in.attrs.items():
f_out.attrs.create(k,v)
for g in ['setup','geometry','cell_to']:
f_in.copy(g,f_out)
for inc in util.show_progress(self.visible['increments']):
f_in.copy(inc,f_out,shallow=True)
for out in _match(output,f_in['/'.join([inc,'geometry'])].keys()):
f_in[inc]['geometry'].copy(out,f_out[inc]['geometry'])
for label in self.homogenizations:
f_in[inc]['homogenization'].copy(label,f_out[inc]['homogenization'],shallow=True)
for label in self.phases:
f_in[inc]['phase'].copy(label,f_out[inc]['phase'],shallow=True)
for ty in ['phase','homogenization']:
for label in self.visible[ty+'s']:
for field in _match(self.visible['fields'],f_in['/'.join([inc,ty,label])].keys()):
p = '/'.join([inc,ty,label,field])
for out in _match(output,f_in[p].keys()):
f_in[p].copy(out,f_out[p])
def export_simulation_setup(self,
output: Union[str, List[str]] = '*',
2022-11-23 02:56:15 +05:30
target_dir: Union[None, str, Path] = None,
overwrite: bool = False,
):
"""
2022-11-09 17:18:48 +05:30
Export original simulation setup of the Result object.
Parameters
----------
output : (list of) str, optional
Names of the datasets to export to the file.
Defaults to '*', in which case all setup files are exported.
target_dir : str or pathlib.Path, optional
Directory to save setup files. Will be created if non-existent.
2022-01-13 03:43:38 +05:30
overwrite : bool, optional
Overwrite any existing setup files.
Defaults to False.
"""
def export(name: str,
obj: Union[h5py.Dataset,h5py.Group],
output: Union[str,List[str]],
cfg_dir: Path,
overwrite: bool):
cfg = cfg_dir/name
2022-11-10 04:00:31 +05:30
if type(obj) == h5py.Dataset and _match(output,[name]):
if cfg.exists() and not overwrite:
raise PermissionError(f'"{cfg}" exists')
else:
2022-11-10 04:00:31 +05:30
cfg.parent.mkdir(parents=True,exist_ok=True)
with util.open_text(cfg,'w') as f_out: f_out.write(obj[0].decode())
cfg_dir = (Path.cwd() if target_dir is None else Path(target_dir))
2021-08-02 14:08:59 +05:30
with h5py.File(self.fname,'r') as f_in:
f_in['setup'].visititems(partial(export,
output=output,
cfg_dir=cfg_dir,
overwrite=overwrite))