def test_recarray():
# check roundtrip of structured array
dt = [('f1', 'f8'),
('f2', 'S10')]
arr = np.zeros((2,), dtype=dt)
arr[0]['f1'] = 0.5
arr[0]['f2'] = 'python'
arr[1]['f1'] = 99
arr[1]['f2'] = 'not perl'
stream = BytesIO()
savemat(stream, {'arr': arr})
d = loadmat(stream, struct_as_record=False)
a20 = d['arr'][0,0]
yield assert_equal, a20.f1, 0.5
yield assert_equal, a20.f2, 'python'
d = loadmat(stream, struct_as_record=True)
a20 = d['arr'][0,0]
yield assert_equal, a20['f1'], 0.5
yield assert_equal, a20['f2'], 'python'
# structs always come back as object types
yield assert_equal, a20.dtype, np.dtype([('f1', 'O'),
('f2', 'O')])
a21 = d['arr'].flat[1]
yield assert_equal, a21['f1'], 99
yield assert_equal, a21['f2'], 'not perl'
def test_read_opts():
# tests if read is seeing option sets, at initialization and after
# initialization
arr = np.arange(6).reshape(1,6)
stream = BytesIO()
savemat(stream, {'a': arr})
rdr = MatFile5Reader(stream)
back_dict = rdr.get_variables()
rarr = back_dict['a']
assert_array_equal(rarr, arr)
rdr = MatFile5Reader(stream, squeeze_me=True)
assert_array_equal(rdr.get_variables()['a'], arr.reshape((6,)))
rdr.squeeze_me = False
assert_array_equal(rarr, arr)
rdr = MatFile5Reader(stream, byte_order=boc.native_code)
assert_array_equal(rdr.get_variables()['a'], arr)
# inverted byte code leads to error on read because of swapped
# header etc
rdr = MatFile5Reader(stream, byte_order=boc.swapped_code)
assert_raises(Exception, rdr.get_variables)
rdr.byte_order = boc.native_code
assert_array_equal(rdr.get_variables()['a'], arr)
arr = np.array(['a string'])
stream.truncate(0)
stream.seek(0)
savemat(stream, {'a': arr})
rdr = MatFile5Reader(stream)
assert_array_equal(rdr.get_variables()['a'], arr)
rdr = MatFile5Reader(stream, chars_as_strings=False)
carr = np.atleast_2d(np.array(list(arr.item()), dtype='U1'))
assert_array_equal(rdr.get_variables()['a'], carr)
rdr.chars_as_strings = True
assert_array_equal(rdr.get_variables()['a'], arr)
def save(self, path=None, full_save=False):
import cPickle as pickle
import os
path = Analyzer.save(self, path=path)
for roi, scores in self.scores.items():
# Save full object!
if full_save:
with open(os.path.join(path, '%s_scores.pickle' %(roi)), 'wb') as output:
pickle.dump(scores, output)
for p, score in enumerate(scores):
mat_score = self._save_score(score)
# TODO: Better use of cv and attributes for leave-one-subject-out
filename = "%s_perm_%04d_data.mat" %(roi, p)
logger.info("Saving %s" %(filename))
savemat(os.path.join(path, filename), mat_score)
return
def test_gzip_simple():
xdense = np.zeros((20,20))
xdense[2,3] = 2.3
xdense[4,5] = 4.5
x = SP.csc_matrix(xdense)
name = 'gzip_test'
expected = {'x':x}
format = '4'
tmpdir = mkdtemp()
try:
fname = pjoin(tmpdir,name)
mat_stream = gzip.open(fname,mode='wb')
savemat(mat_stream, expected, format=format)
mat_stream.close()
mat_stream = gzip.open(fname,mode='rb')
actual = loadmat(mat_stream, struct_as_record=True)
mat_stream.close()
finally:
shutil.rmtree(tmpdir)
assert_array_almost_equal(actual['x'].todense(),
expected['x'].todense(),
err_msg=repr(actual))
def test_empty_string():
# make sure reading empty string does not raise error
estring_fname = pjoin(test_data_path, 'single_empty_string.mat')
fp = open(estring_fname, 'rb')
rdr = MatFile5Reader(fp)
d = rdr.get_variables()
fp.close()
assert_array_equal(d['a'], np.array([], dtype='U1'))
# empty string round trip. Matlab cannot distiguish
# between a string array that is empty, and a string array
# containing a single empty string, because it stores strings as
# arrays of char. There is no way of having an array of char that
# is not empty, but contains an empty string.
stream = BytesIO()
savemat(stream, {'a': np.array([''])})
rdr = MatFile5Reader(stream)
d = rdr.get_variables()
assert_array_equal(d['a'], np.array([], dtype='U1'))
stream.truncate(0)
stream.seek(0)
savemat(stream, {'a': np.array([], dtype='U1')})
rdr = MatFile5Reader(stream)
d = rdr.get_variables()
assert_array_equal(d['a'], np.array([], dtype='U1'))
stream.close()
def test_unicode_mat4():
# Mat4 should save unicode as latin1
bio = BytesIO()
var = {'second_cat': u('Schrödinger')}
savemat(bio, var, format='4')
var_back = loadmat(bio)
assert_equal(var_back['second_cat'], var['second_cat'])
def test_fieldnames():
# Check that field names are as expected
stream = BytesIO()
savemat(stream, {"a": {"a": 1, "b": 2}})
res = loadmat(stream)
field_names = res["a"].dtype.names
assert_equal(set(field_names), set(("a", "b")))
def test_unicode_mat4():
# Mat4 should save unicode as latin1
bio = BytesIO()
var = {"second_cat": u("Schrödinger")}
savemat(bio, var, format="4")
var_back = loadmat(bio)
assert_equal(var_back["second_cat"], var["second_cat"])
def test_save_dict():
# Test that dict can be saved (as recarray), loaded as matstruct
d = {'a':1, 'b':2}
stream = StringIO()
savemat(stream, {'dict':d})
stream.seek(0)
vals = loadmat(stream)
def test_fieldnames():
# Check that field names are as expected
stream = BytesIO()
savemat(stream, {'a': {'a':1, 'b':2}})
res = loadmat(stream)
field_names = res['a'].dtype.names
assert_equal(set(field_names), set(('a', 'b')))
def test_save_dict():
# Test that dict can be saved (as recarray), loaded as matstruct
dict_types = ((dict, False),)
try:
from collections import OrderedDict
except ImportError:
pass
else:
dict_types += ((OrderedDict, True),)
ab_exp = np.array([[(1, 2)]], dtype=[('a', object), ('b', object)])
ba_exp = np.array([[(2, 1)]], dtype=[('b', object), ('a', object)])
for dict_type, is_ordered in dict_types:
# Initialize with tuples to keep order for OrderedDict
d = dict_type([('a', 1), ('b', 2)])
stream = BytesIO()
savemat(stream, {'dict': d})
stream.seek(0)
vals = loadmat(stream)['dict']
assert_equal(set(vals.dtype.names), set(['a', 'b']))
if is_ordered: # Input was ordered, output in ab order
assert_array_equal(vals, ab_exp)
else: # Not ordered input, either order output
if vals.dtype.names[0] == 'a':
assert_array_equal(vals, ab_exp)
else:
assert_array_equal(vals, ba_exp)
def test_scalar_squeeze():
stream = BytesIO()
in_d = {"scalar": [[0.1]], "string": "my name", "st": {"one": 1, "two": 2}}
savemat(stream, in_d)
out_d = loadmat(stream, squeeze_me=True)
assert_(isinstance(out_d["scalar"], float))
assert_(isinstance(out_d["string"], string_types))
assert_(isinstance(out_d["st"], np.ndarray))
def test_mat_struct_squeeze():
stream = BytesIO()
in_d = {"st": {"one": 1, "two": 2}}
savemat(stream, in_d)
# no error without squeeze
out_d = loadmat(stream, struct_as_record=False)
# previous error was with squeeze, with mat_struct
out_d = loadmat(stream, struct_as_record=False, squeeze_me=True)
def test_scalar_squeeze():
stream = BytesIO()
in_d = {'scalar': [[0.1]], 'string': 'my name', 'st':{'one':1, 'two':2}}
savemat(stream, in_d)
out_d = loadmat(stream, squeeze_me=True)
assert_(isinstance(out_d['scalar'], float))
assert_(isinstance(out_d['string'], string_types))
assert_(isinstance(out_d['st'], np.ndarray))
def test_sparse_in_struct():
# reproduces bug found by DC where Cython code was insisting on
# ndarray return type, but getting sparse matrix
st = {'sparsefield': SP.coo_matrix(np.eye(4))}
stream = BytesIO()
savemat(stream, {'a':st})
d = loadmat(stream, struct_as_record=True)
yield assert_array_equal, d['a'][0,0]['sparsefield'].todense(), np.eye(4)
def test_long_field_names():
# Test limit for length of field names in structs
lim = 63
fldname = "a" * lim
st1 = np.zeros((1, 1), dtype=[(fldname, object)])
savemat(BytesIO(), {"longstruct": st1}, format="5", long_field_names=True)
fldname = "a" * (lim + 1)
st1 = np.zeros((1, 1), dtype=[(fldname, object)])
assert_raises(ValueError, savemat, BytesIO(), {"longstruct": st1}, format="5", long_field_names=True)
def test_save_empty_dict():
# saving empty dict also gives empty struct
stream = BytesIO()
savemat(stream, {'arr': {}})
d = loadmat(stream)
a = d['arr']
assert_equal(a.shape, (1,1))
assert_equal(a.dtype, np.dtype(object))
assert_(a[0,0] is None)
def test_regression_653():
# Saving a dictionary with only invalid keys used to raise an error. Now we
# save this as an empty struct in matlab space.
sio = BytesIO()
savemat(sio, {'d':{1:2}}, format='5')
back = loadmat(sio)['d']
# Check we got an empty struct equivalent
assert_equal(back.shape, (1,1))
assert_equal(back.dtype, np.dtype(object))
assert_(back[0,0] is None)
def test_structname_len():
# Test limit for length of field names in structs
lim = 31
fldname = "a" * lim
st1 = np.zeros((1, 1), dtype=[(fldname, object)])
mat_stream = BytesIO()
savemat(BytesIO(), {"longstruct": st1}, format="5")
fldname = "a" * (lim + 1)
st1 = np.zeros((1, 1), dtype=[(fldname, object)])
assert_raises(ValueError, savemat, BytesIO(), {"longstruct": st1}, format="5")
def test_round_types():
# Check that saving, loading preserves dtype in most cases
arr = np.arange(10)
stream = BytesIO()
for dts in ("f8", "f4", "i8", "i4", "i2", "i1", "u8", "u4", "u2", "u1", "c16", "c8"):
stream.truncate(0)
stream.seek(0) # needed for BytesIO in python 3
savemat(stream, {"arr": arr.astype(dts)})
vars = loadmat(stream)
assert_equal(np.dtype(dts), vars["arr"].dtype)
def test_empty_sparse():
# Can we read empty sparse matrices?
sio = BytesIO()
import scipy.sparse
empty_sparse = scipy.sparse.csr_matrix([[0,0],[0,0]])
savemat(sio, dict(x=empty_sparse))
sio.seek(0)
res = loadmat(sio)
assert_array_equal(res['x'].shape, empty_sparse.shape)
assert_array_equal(res['x'].todense(), 0)
def test_structname_len():
# Test limit for length of field names in structs
lim = 31
fldname = 'a' * lim
st1 = np.zeros((1,1), dtype=[(fldname, object)])
savemat(BytesIO(), {'longstruct': st1}, format='5')
fldname = 'a' * (lim+1)
st1 = np.zeros((1,1), dtype=[(fldname, object)])
assert_raises(ValueError, savemat, BytesIO(),
{'longstruct': st1}, format='5')
def test_mat4_3d():
# test behavior when writing 3D arrays to matlab 4 files
stream = StringIO()
arr = np.arange(24).reshape((2,3,4))
warnings.simplefilter('error')
yield (assert_raises, DeprecationWarning, savemat,
stream, {'a': arr}, True, '4')
warnings.resetwarnings()
savemat(stream, {'a': arr}, format='4')
d = loadmat(stream)
yield assert_array_equal, d['a'], arr.reshape((6,4))
def test_round_types():
# Check that saving, loading preserves dtype in most cases
arr = np.arange(10)
stream = BytesIO()
for dts in ('f8','f4','i8','i4','i2','i1',
'u8','u4','u2','u1','c16','c8'):
stream.truncate(0)
stream.seek(0) # needed for BytesIO in python 3
savemat(stream, {'arr': arr.astype(dts)})
vars = loadmat(stream)
assert_equal(np.dtype(dts), vars['arr'].dtype)
def test_long_field_names():
# Test limit for length of field names in structs
lim = 63
fldname = 'a' * lim
st1 = np.zeros((1,1), dtype=[(fldname, object)])
mat_stream = StringIO()
savemat(StringIO(), {'longstruct': st1}, format='5',long_field_names=True)
fldname = 'a' * (lim+1)
st1 = np.zeros((1,1), dtype=[(fldname, object)])
assert_raises(ValueError, savemat, StringIO(),
{'longstruct': st1}, format='5',long_field_names=True)
def test_mat_struct_squeeze():
stream = BytesIO()
in_d = {'st':{'one':1, 'two':2}}
savemat(stream, in_d)
# no error without squeeze
out_d = loadmat(stream, struct_as_record=False)
# previous error was with squeeze, with mat_struct
out_d = loadmat(stream,
struct_as_record=False,
squeeze_me=True,
)
def test_cell_with_one_thing_in_it():
# Regression test - make a cell array that's 1 x 2 and put two
# strings in it. It works. Make a cell array that's 1 x 1 and put
# a string in it. It should work but, in the old days, it didn't.
cells = np.ndarray((1,2),dtype=object)
cells[0,0] = 'Hello'
cells[0,1] = 'World'
savemat(BytesIO(), {'x': cells}, format='5')
cells = np.ndarray((1,1),dtype=object)
cells[0,0] = 'Hello, world'
savemat(BytesIO(), {'x': cells}, format='5')
def test_1d_shape():
# Current 5 behavior is 1D -> column vector
arr = np.arange(5)
stream = StringIO()
savemat(stream, {'oned':arr}, format='5')
vals = loadmat(stream)
yield assert_equal, vals['oned'].shape, (5,1)
# Current 4 behavior is 1D -> row vector
arr = np.arange(5)
stream = StringIO()
savemat(stream, {'oned':arr}, format='4')
vals = loadmat(stream)
yield assert_equal, vals['oned'].shape, (1, 5)
for format in ('4', '5'):
# can be explicitly 'column' for oned_as
stream = StringIO()
savemat(stream, {'oned':arr},
format=format,
oned_as='column')
vals = loadmat(stream)
yield assert_equal, vals['oned'].shape, (5,1)
# but different from 'row'
stream = StringIO()
savemat(stream, {'oned':arr},
format=format,
oned_as='row')
vals = loadmat(stream)
yield assert_equal, vals['oned'].shape, (1,5)
def test_1d_shape():
# Current 5 behavior is 1D -> column vector
arr = np.arange(5)
stream = BytesIO()
warn_ctx = WarningManager()
warn_ctx.__enter__()
try:
# silence warnings for tests
warnings.simplefilter('ignore')
savemat(stream, {'oned':arr}, format='5')
vals = loadmat(stream)
assert_equal(vals['oned'].shape, (5,1))
# Current 4 behavior is 1D -> row vector
stream = BytesIO()
savemat(stream, {'oned':arr}, format='4')
vals = loadmat(stream)
assert_equal(vals['oned'].shape, (1, 5))
for format in ('4', '5'):
# can be explicitly 'column' for oned_as
stream = BytesIO()
savemat(stream, {'oned':arr},
format=format,
oned_as='column')
vals = loadmat(stream)
assert_equal(vals['oned'].shape, (5,1))
# but different from 'row'
stream = BytesIO()
savemat(stream, {'oned':arr},
format=format,
oned_as='row')
vals = loadmat(stream)
assert_equal(vals['oned'].shape, (1,5))
finally:
warn_ctx.__exit__()
def test_compression():
arr = np.zeros(100).reshape((5,20))
arr[2,10] = 1
stream = BytesIO()
savemat(stream, {'arr':arr})
raw_len = len(stream.getvalue())
vals = loadmat(stream)
yield assert_array_equal, vals['arr'], arr
stream = BytesIO()
savemat(stream, {'arr':arr}, do_compression=True)
compressed_len = len(stream.getvalue())
vals = loadmat(stream)
yield assert_array_equal, vals['arr'], arr
yield assert_, raw_len > compressed_len
# Concatenate, test later
arr2 = arr.copy()
arr2[0,0] = 1
stream = BytesIO()
savemat(stream, {'arr':arr, 'arr2':arr2}, do_compression=False)
vals = loadmat(stream)
yield assert_array_equal, vals['arr2'], arr2
stream = BytesIO()
savemat(stream, {'arr':arr, 'arr2':arr2}, do_compression=True)
vals = loadmat(stream)
yield assert_array_equal, vals['arr2'], arr2
def test_write_opposite_endian():
# We don't support writing opposite endian .mat files, but we need to behave
# correctly if the user supplies an other-endian NumPy array to write out.
float_arr = np.array([[2., 3.], [3., 4.]])
int_arr = np.arange(6).reshape((2, 3))
uni_arr = np.array(['hello', 'world'], dtype='U')
stream = BytesIO()
savemat(
stream, {
'floats': float_arr.byteswap().newbyteorder(),
'ints': int_arr.byteswap().newbyteorder(),
'uni_arr': uni_arr.byteswap().newbyteorder()
})
rdr = MatFile5Reader(stream)
d = rdr.get_variables()
assert_array_equal(d['floats'], float_arr)
assert_array_equal(d['ints'], int_arr)
assert_array_equal(d['uni_arr'], uni_arr)
stream.close()
def test_long_field_names_in_struct():
# Regression test - long_field_names was erased if you passed a struct
# within a struct
lim = 63
fldname = 'a' * lim
cell = np.ndarray((1, 2), dtype=object)
st1 = np.zeros((1, 1), dtype=[(fldname, object)])
cell[0, 0] = st1
cell[0, 1] = st1
mat_stream = BytesIO()
savemat(BytesIO(), {'longstruct': cell}, format='5', long_field_names=True)
#
# Check to make sure it fails with long field names off
#
assert_raises(ValueError,
savemat,
BytesIO(), {'longstruct': cell},
format='5',
long_field_names=False)
def test_save_dict():
# Test that dict can be saved (as recarray), loaded as matstruct
dict_types = ((dict, False), (OrderedDict, True),)
ab_exp = np.array([[(1, 2)]], dtype=[('a', object), ('b', object)])
ba_exp = np.array([[(2, 1)]], dtype=[('b', object), ('a', object)])
for dict_type, is_ordered in dict_types:
# Initialize with tuples to keep order for OrderedDict
d = dict_type([('a', 1), ('b', 2)])
stream = BytesIO()
savemat(stream, {'dict': d})
stream.seek(0)
vals = loadmat(stream)['dict']
assert_equal(set(vals.dtype.names), set(['a', 'b']))
if is_ordered: # Input was ordered, output in ab order
assert_array_equal(vals, ab_exp)
else: # Not ordered input, either order output
if vals.dtype.names[0] == 'a':
assert_array_equal(vals, ab_exp)
else:
assert_array_equal(vals, ba_exp)
def test_varmats_from_mat():
# Make a mat file with several variables, write it, read it back
names_vars = (('arr', mlarr(np.arange(10))), ('mystr', mlarr('a string')),
('mynum', mlarr(10)))
# Dict like thing to give variables in defined order
class C(object):
def items(self):
return names_vars
stream = BytesIO()
savemat(stream, C())
varmats = varmats_from_mat(stream)
assert_equal(len(varmats), 3)
for i in range(3):
name, var_stream = varmats[i]
exp_name, exp_res = names_vars[i]
assert_equal(name, exp_name)
res = loadmat(var_stream)
assert_array_equal(res[name], exp_res)
def test_multiple_open():
# Ticket #1039, on Windows: check that files are not left open
tmpdir = mkdtemp()
try:
x = dict(x=np.zeros((2, 2)))
fname = pjoin(tmpdir, "a.mat")
# Check that file is not left open
savemat(fname, x, oned_as='row')
os.unlink(fname)
savemat(fname, x, oned_as='row')
loadmat(fname)
os.unlink(fname)
# Check that stream is left open
f = open(fname, 'wb')
savemat(f, x, oned_as='column')
f.seek(0)
f.close()
f = open(fname, 'rb')
loadmat(f)
f.seek(0)
f.close()
finally:
shutil.rmtree(tmpdir)
def test_str_round():
# from report by Angus McMorland on mailing list 3 May 2010
stream = BytesIO()
in_arr = np.array(['Hello', 'Foob'])
out_arr = np.array(['Hello', 'Foob '])
savemat(stream, dict(a=in_arr))
res = loadmat(stream)
# resulted in ['HloolFoa', 'elWrdobr']
assert_array_equal(res['a'], out_arr)
stream.truncate(0)
stream.seek(0)
# Make Fortran ordered version of string
in_str = in_arr.tostring(order='F')
in_from_str = np.ndarray(shape=a.shape,
dtype=in_arr.dtype,
order='F',
buffer=in_str)
savemat(stream, dict(a=in_from_str))
assert_array_equal(res['a'], out_arr)
# unicode save did lead to buffer too small error
stream.truncate(0)
stream.seek(0)
in_arr_u = in_arr.astype('U')
out_arr_u = out_arr.astype('U')
savemat(stream, {'a': in_arr_u})
res = loadmat(stream)
assert_array_equal(res['a'], out_arr_u)
def test_recarray():
# check roundtrip of structured array
dt = [('f1', 'f8'), ('f2', 'S10')]
arr = np.zeros((2, ), dtype=dt)
arr[0]['f1'] = 0.5
arr[0]['f2'] = 'python'
arr[1]['f1'] = 99
arr[1]['f2'] = 'not perl'
stream = BytesIO()
savemat(stream, {'arr': arr})
d = loadmat(stream, struct_as_record=False)
a20 = d['arr'][0, 0]
assert_equal(a20.f1, 0.5)
assert_equal(a20.f2, 'python')
d = loadmat(stream, struct_as_record=True)
a20 = d['arr'][0, 0]
assert_equal(a20['f1'], 0.5)
assert_equal(a20['f2'], 'python')
# structs always come back as object types
assert_equal(a20.dtype, np.dtype([('f1', 'O'), ('f2', 'O')]))
a21 = d['arr'].flat[1]
assert_equal(a21['f1'], 99)
assert_equal(a21['f2'], 'not perl')
def test_gzip_simple():
xdense = np.zeros((20, 20))
xdense[2, 3] = 2.3
xdense[4, 5] = 4.5
x = SP.csc_matrix(xdense)
name = 'gzip_test'
expected = {'x': x}
format = '4'
tmpdir = mkdtemp()
try:
fname = join(tmpdir, name)
mat_stream = gzip.open(fname, mode='wb')
savemat(mat_stream, expected, format=format)
mat_stream.close()
mat_stream = gzip.open(fname, mode='rb')
actual = loadmat(mat_stream, struct_as_record=True)
mat_stream.close()
finally:
shutil.rmtree(tmpdir)
assert_array_almost_equal(actual['x'].todense(), expected['x'].todense())
def test_1d_shape():
# Current 5 behavior is 1D -> column vector
arr = np.arange(5)
stream = StringIO()
savemat(stream, {'oned': arr}, format='5')
vals = loadmat(stream)
yield assert_equal, vals['oned'].shape, (5, 1)
# Current 4 behavior is 1D -> row vector
arr = np.arange(5)
stream = StringIO()
savemat(stream, {'oned': arr}, format='4')
vals = loadmat(stream)
yield assert_equal, vals['oned'].shape, (1, 5)
for format in ('4', '5'):
# can be explicitly 'column' for oned_as
stream = StringIO()
savemat(stream, {'oned': arr}, format=format, oned_as='column')
vals = loadmat(stream)
yield assert_equal, vals['oned'].shape, (5, 1)
# but different from 'row'
stream = StringIO()
savemat(stream, {'oned': arr}, format=format, oned_as='row')
vals = loadmat(stream)
yield assert_equal, vals['oned'].shape, (1, 5)
def test_1d_shape():
# New 5 behavior is 1D -> row vector
arr = np.arange(5)
for format in ('4', '5'):
# Column is the default
stream = BytesIO()
savemat(stream, {'oned': arr}, format=format)
vals = loadmat(stream)
assert_equal(vals['oned'].shape, (1, 5))
# can be explicitly 'column' for oned_as
stream = BytesIO()
savemat(stream, {'oned': arr}, format=format, oned_as='column')
vals = loadmat(stream)
assert_equal(vals['oned'].shape, (5, 1))
# but different from 'row'
stream = BytesIO()
savemat(stream, {'oned': arr}, format=format, oned_as='row')
vals = loadmat(stream)
assert_equal(vals['oned'].shape, (1, 5))
if each in exclude_list:
continue
mdict[each] = input_file.variables[each][:]
print each,
except Exception:
missing.append(each)
continue
print ""
if len(missing) > 0:
print "Warning! the following variable(s) was(were) not found in '%s':" % (input_name), missing
print "Excluded variables: ", exclude_list
if mdict.keys() == []:
print "WARNING: No data to write. Exiting ..."
print_options_and_exit(-2)
print "Writing data to %s..." % (output_name)
try:
from scipy.io.matlab.mio import savemat
except:
print "ERROR! Can't import 'savemat' from scipy.io.matlab.mio. Exiting ..."
exit(-3)
from scipy.io.matlab.mio import savemat
try:
savemat(output_name, mdict, appendmat=False, format='5')
except:
print "ERROR! Can't write to %s. Exiting ..." % (output_name)
exit(-4)
def setUp(self):
self.prior = np.array([1, 0, 0])
self.transmat = np.matrix([[0, 1, 0], [0, 0, 1], [0, 0, 1]])
self.mu = np.zeros((2, 3, 2))
self.mu[:, :, 0] = np.array([[1, 2, 3], [1, 2, 3]])
self.mu[:, :, 1] = np.array([[4, 5, 6], [4, 5, 6]])
self.Sigma = np.zeros((2, 2, 3, 2))
for i in range(3):
self.Sigma[:, :, i, 0] = np.diag(np.ones((2, )) * 0.01)
self.Sigma[:, :, i, 1] = np.diag(np.ones((2, )) * 0.01)
self.mixmat = np.array([[.5, .5], [.5, .5], [.5, .5]])
try:
with open('MhmmEM2DTest.cache', 'rb') as f:
cache = load(f)
self.obs = cache['obs']
self.prior0 = cache['prior0']
self.transmat0 = cache['transmat0']
self.mu0 = cache['mu0']
self.Sigma0 = cache['Sigma0']
self.mixmat0 = cache['mixmat0']
except:
self.obs, hidden = mhmm_sample(T=4,
numex=100,
initial_prob=self.prior,
transmat=self.transmat,
mu=self.mu,
Sigma=self.Sigma,
mixmat=self.mixmat)
self.prior0, _ = mk_stochastic(np.random.rand(3))
self.transmat0, _ = mk_stochastic(np.random.rand(3, 3))
self.mu0 = np.zeros((2, 3, 2))
self.mu0[:, :, 0] = np.array([[1.5, 2.5, 3.5], [1.5, 2.5, 3.5]])
self.mu0[:, :, 1] = np.array([[4.5, 5.5, 6.5], [4.5, 5.5, 6.5]])
self.Sigma0 = np.zeros((2, 2, 3, 2))
for i in range(3):
self.Sigma0[:, :, i, 0] = np.diag(np.ones((2, )) * 1.0)
self.Sigma0[:, :, i, 1] = np.diag(np.ones((2, )) * 1.0)
self.mixmat0 = np.array([[.2, .8], [.2, .8], [.2, .8]])
cache = {
'obs': self.obs,
'prior0': self.prior0,
'transmat0': self.transmat0,
'mu0': self.mu0,
'Sigma0': self.Sigma0,
'mixmat0': self.mixmat0
}
with open('MhmmEM2DTest.cache', 'wb') as f:
dump(cache, f)
savemat('MhmmEM2DTest.mat', cache)
# IApos=InertialAxis
# rho=1.02
AELAOPTS = AeroelasticOps(EApos, IApos, 0.08891)
# AELAOPTS = AeroelasticOps(EApos,IApos,rho)
Settings.OutputFileRoot = 'M' + str(M) + 'N' + str(N) + '_V' + str(
U_mag) + '_alpha' + str(alpha)
# solve static problem
PosDefor, PsiDefor, Zeta, ZetaStar, Gamma, GammaStar, iForceStep, NumNodes_tot, NumDof, XBELEM, XBNODE, PosIni, PsiIni, Uext = Solve_Py(
XBINPUT, XBOPTS, VMOPTS, VMINPUT, AELAOPTS)
# save reference gamma
if True:
fileName = Settings.OutputDir + Settings.OutputFileRoot + '_Gamma0'
savemat(fileName, {'Gamma0': Gamma}, True)
### linearized beam model
A, B, C, kMat, phiSort = genSSbeam(XBINPUT,
NumNodes_tot,
NumDof,
XBELEM,
XBNODE,
PosIni,
PsiIni,
PosDefor,
PsiDefor,
XBOPTS,
chord,
U_mag,
modal=10)
'ssim': ssim,
'CoordinateChannel2D': CoordinateChannel2D
})
output_path = './output_file/'
if not os.path.exists(output_path):
os.makedirs(output_path)
testImagePath = args.test_path
fileName = os.listdir(testImagePath)
for i in range(len(fileName)):
start_time = time.time()
img = cv2.imread(testImagePath + fileName[i])
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) / 255
results = generate_output(img, model)
end_time = time.time()
print('predicted time', end_time - start_time)
print(fileName[i].split('clean')[0][0:-1] + '.mat')
savemat(output_path + fileName[i].split('clean')[0][0:-1] + '.mat',
{'cube': results})
print(i)
print("output files saved in " + output_path)
n_im, h, w, c = udc_mat.shape
results = udc_mat.copy()
for i in range(n_im):
udc = np.reshape(udc_mat[i, :, :, :], (h, w, c))
restored, run_time = restoration(opt, generator, udc, i, run_time)
results[i, :, :, :] = restored
print(run_time)
# create results directory
res_dir = 'res_dir'
os.makedirs(os.path.join(work_dir, res_dir), exist_ok=True)
# save images in a .mat file with dictionary key "results"
res_fn = os.path.join(work_dir, res_dir, 'results.mat')
res_key = 'results' # Note: do not change this key, the evaluation code will look for this key
savemat(res_fn, {res_key: results})
# submission indormation
# TODO: update the values below; the evaluation code will parse them
runtime = run_time / 30 # seconds / image
cpu_or_gpu = 0 # 0: GPU, 1: CPU
method = 1 # 0: traditional methods, 1: deep learning method
other = '(optional) any additional description or information'
# prepare and save readme file
readme_fn = os.path.join(work_dir, res_dir,
'readme.txt') # Note: do not change 'readme.txt'
with open(readme_fn, 'w') as readme_file:
readme_file.write('Runtime (seconds / megapixel): %s\n' % str(runtime))
readme_file.write('CPU[1] / GPU[0]: %s\n' % str(cpu_or_gpu))
readme_file.write('Method: %s\n' % str(method))
def prep_Result(opt):
net = set_model(opt)
_, net, _ = load_model(opt, net)
if opt.n_channels == 1:
from skimage.external.tifffile import imsave, imread
else:
from skimage.io import imsave, imread
opt = set_gpu(opt)
if opt.use_cuda:
net = net.to(opt.device)
if opt.multi_gpu:
net = nn.DataParallel(net)
set_test_dir(opt)
if not os.path.exists(opt.test_result_dir):
os.makedirs(opt.test_result_dir)
res_img_dir = os.path.join(opt.test_result_dir, 'result_img_dir')
if not os.path.exists(res_img_dir):
os.makedirs(res_img_dir)
# create results directory
res_dir = os.path.join(opt.test_result_dir, 'res_dir')
os.makedirs(os.path.join(res_dir), exist_ok=True)
print('\ntest_result_dir : ', opt.test_result_dir)
print('\nresult_img_dir : ', res_img_dir)
print('\nresult_dir : ', res_dir)
# total_psnr = 0
# total_ssim = 0
loss_criterion = nn.MSELoss()
total_psnr = 0.0
if opt.n_channels == 1:
# load noisy images
noisy_fn = 'siddplus_test_noisy_raw.mat'
noisy_key = 'siddplus_test_noisy_raw'
noisy_mat = loadmat(os.path.join(opt.test_dir, opt.dataset,
noisy_fn))[noisy_key]
# denoise
n_im, h, w = noisy_mat.shape
results = noisy_mat.copy()
start_time = time.time()
for i in range(n_im):
print('\n[*]PROCESSING..{}/{}'.format(i, n_im))
noisy = np.reshape(noisy_mat[i, :, :], (h, w))
denoised = denoiser(opt, net, noisy)
results[i, :, :] = denoised
result_name = str(i) + '.tiff'
concat_img = np.concatenate((denoised, noisy), axis=1)
imsave(os.path.join(res_img_dir, result_name), concat_img)
denoised = torch.Tensor(denoised)
noisy = torch.Tensor(noisy)
mse_loss = loss_criterion(denoised, noisy)
psnr = 10 * math.log10(1 / mse_loss.item())
total_psnr += psnr
print('%.5fs .. [%d/%d] psnr : %.5f, avg_psnr : %.5f' %
(time.time() - start_time, i, n_im, psnr, total_psnr /
(i + 1)))
else:
# load noisy images
noisy_fn = 'siddplus_test_noisy_srgb.mat'
noisy_key = 'siddplus_test_noisy_srgb'
noisy_mat = loadmat(os.path.join(opt.test_dir, opt.dataset,
noisy_fn))[noisy_key]
# denoise
n_im, h, w, c = noisy_mat.shape
results = noisy_mat.copy()
start_time = time.time()
for i in range(n_im):
print('\n[*]PROCESSING..{}/{}'.format(i, n_im))
noisy = np.reshape(noisy_mat[i, :, :, :], (h, w, c))
denoised = denoiser(opt, net, noisy)
results[i, :, :, :] = denoised
result_name = str(i) + '.png'
concat_img = np.concatenate((denoised, noisy), axis=1)
imsave(os.path.join(res_img_dir, result_name), concat_img)
denoised = torch.Tensor(denoised).float() / 255.0
noisy = torch.Tensor(noisy).float() / 255.0
mse_loss = loss_criterion(noisy, denoised)
psnr = 10 * math.log10(1 / mse_loss.item())
total_psnr += psnr
print('%.5fs .. [%d/%d] psnr : %.5f, avg_psnr : %.5f' %
(time.time() - start_time, i, n_im, psnr, total_psnr /
(i + 1)))
print("****total avg psnr : %.10f", total_psnr / (n_im))
# save denoised images in a .mat file with dictionary key "results"
res_fn = os.path.join(res_dir, 'results.mat')
res_key = 'results' # Note: do not change this key, the evaluation code will look for this key
savemat(res_fn, {res_key: results})
runtime = 0.0 # seconds / megapixel
cpu_or_gpu = 0 # 0: GPU, 1: CPU
use_metadata = 0 # 0: no use of metadata, 1: metadata used
other = '(optional) any additional description or information'
# prepare and save readme file
readme_fn = os.path.join(res_dir,
'readme.txt') # Note: do not change 'readme.txt'
with open(readme_fn, 'w') as readme_file:
readme_file.write('Runtime (seconds / megapixel): %s\n' % str(runtime))
readme_file.write('CPU[1] / GPU[0]: %s\n' % str(cpu_or_gpu))
readme_file.write('Metadata[1] / No Metadata[0]: %s\n' %
str(use_metadata))
readme_file.write('Other description: %s\n' % str(other))
# compress results directory
res_zip_fn = 'results_dir'
shutil.make_archive(os.path.join(opt.test_result_dir, res_zip_fn), 'zip',
res_dir)
model.cuda()
output = model(images)
pred_tot.append(output.cpu().data.numpy())
brut_tot.append(brut.data.numpy())
true_tot.append(target.data.numpy())
if save:
tosave = {}
name_save = 'prediction' + str(i) + '.mat'
target = target.cpu().data.numpy()
brut = brut.cpu().data.numpy()
img = output.cpu().data.numpy()
tosave['yt'] = target
tosave['y'] = img
tosave['x'] = brut
mio.savemat(save_path + name_save, tosave)
#%%
plt.figure(1)
img = output.cpu().detach().numpy()
handle = plt.subplot(311)
handle.set_title('row image')
plt.imshow(brut[0, 0, :, :], cmap='hot')
handle = plt.subplot(312)
handle.set_title('target image')
plt.imshow(target[0, 0, :, :], cmap='hot')
handle = plt.subplot(313)
handle.set_title('reconstructed image')
def test_roundtrip_zero_dimensions():
stream = BytesIO()
savemat(stream, {'d':np.empty((10, 0))})
d = loadmat(stream)
assert d['d'].shape == (10, 0)
def genLinearAerofoil(m, mW, writeToMat=False, e=0.25, f=0.75):
"""@brief Generate linear model of aerofoil.
@param m Chordwise panels.
@param mW Chordwise panels in wake.
@param delS Non-dim time step.
@param e location of pitch axis aft of LE [0,1], default 0.25.
@param f location of flap hinge aft of LE [0,1], default 0.75.
@param writeToMat write to file in Settings.OutputDir.
"""
n = 1 # number of spanwise panels
# infer delS from body discretization
delS = 2.0 / m
# hack
# factor = 4
# mW=factor*mW
# delS = delS/float(factor)
# initialise states and inputs
gam = np.zeros((m * n))
gamW = np.zeros((mW * n))
gamPri = np.zeros((m * n))
chords = np.linspace(0.0, 1.0, m + 1, True)
chordsW = np.linspace(1.0, 1.0 + mW * delS / 2.0, mW + 1, True)
spans = np.linspace(0.0, 2000.0, n + 1, True)
zeta = np.zeros(3 * len(chords) * len(spans))
zetaW = np.zeros(3 * len(chordsW) * len(spans))
zetaPri = np.zeros((3 * len(chords) * len(spans)))
#zetaPri[0::3] = -1.0
nu = np.zeros_like(zetaPri)
nu[0::3] = 0.5
kk = 0
for c in chords:
for s in spans:
zeta[3 * kk] = c
zeta[3 * kk + 1] = s
kk = kk + 1
# end for s
# end for c
kk = 0
for c in chordsW:
for s in spans:
zetaW[3 * kk] = c
zetaW[3 * kk + 1] = s
kk = kk + 1
# end for s
# end for c
# generate model
E, F, G, C, D = genSSuvlm(gam, gamW, gamPri, zeta, zetaW, zetaPri, nu, m,
n, mW, delS, True)
# convert inputs from general kinematics to aerofil DoFs
T = np.zeros((9 * (m + 1) * (n + 1), 5))
for i in range(m + 1):
for j in range(n + 1):
q = i * (n + 1) + j
# alpha, alphaPrime
T[3 * (m + 1) * (n + 1) + 3 * q + 2,
0] = -(zeta[3 * q] + 0.25 / m - e)
T[3 * q + 2, 1] = -(zeta[3 * q] + 0.25 / m - e)
# plunge
T[3 * q + 2, 2] = -1
# beta, betaPrime
if zeta[3 * q] + 0.25 / m > f:
T[3 * (m + 1) * (n + 1) + 3 * q + 2,
3] = -(zeta[3 * q] + 0.25 / m - f)
T[3 * q + 2, 4] = -(zeta[3 * q] + 0.25 / m - f)
G_s = np.dot(G, T)
D_s = np.dot(D, T)
# get coefficients as output
T_coeff = np.zeros((3, 3 * (m + 1) * (n + 1)))
T_coeff[0, 0::3] = 1.0 #drag
T_coeff[1, 2::3] = 1.0 #lift
for i in range(m + 1):
for j in range(n + 1):
q = i * (n + 1) + j
# moment = r*L, +ve nose-up, at quarter chord
T_coeff[2, 3 * q + 2] = -(zeta[3 * q] + 0.25 / m - 0.25)
C_coeff = np.dot(T_coeff, C)
D_coeff = np.dot(T_coeff, D)
D_s_coeff = np.dot(T_coeff, D_s)
if writeToMat == True:
fileName = Settings.OutputDir + 'TESTaerofoil_m' + str(m) + 'mW' + str(
mW) + 'delS' + str(delS)
if e != 0.25:
fileName += 'e' + str(e)
if f != 0.75:
fileName += 'f' + str(f)
savemat(
fileName, {
'E': E,
'F': F,
'G': G,
'C': C,
'D': D,
'm': m,
'mW': mW,
'delS': delS,
'G_s': G_s,
'D_s': D_s,
'C_coeff': C_coeff,
'D_coeff': D_coeff,
'D_s_coeff': D_s_coeff,
'T_coeff': T_coeff
}, True)
# end if
return E, F, G, C, D, delS
def genLinearRectWing(AR,
m,
mW,
n,
e=0.25,
f=0.75,
writeToMat=False,
imageMeth=False):
"""@brief Generate linear model of rectangular wing.
@param AR Aspect ration
@param m Chordwise panels.
@param mW Chordwise panels in wake.
@param n Spanwise panels.
@param delS Non-dim time step.
@param e location of pitch axis aft of LE [0,1], default 0.25.
@param f location of flap hinge aft of LE [0,1], default 0.75.
@param writeToMat write to file in Settings.OutputDir.
@param imageMeth Use image method across xz-plane.
"""
# infer delS from body discretization
delS = 2.0 / m
# initialise states and inputs
gam = np.zeros((m * n))
gamW = np.zeros((mW * n))
gamPri = np.zeros((m * n))
chords = np.linspace(0.0, 1.0, m + 1, True)
chordsW = np.linspace(1.0, 1.0 + mW * delS / 2.0, mW + 1, True)
if imageMeth:
spans = np.linspace(0.0, AR / 2.0, n + 1, True)
else:
spans = np.linspace(-AR / 2.0, AR / 2.0, n + 1, True)
# end if
zeta = np.zeros(3 * len(chords) * len(spans))
zetaW = np.zeros(3 * len(chordsW) * len(spans))
zetaPri = np.zeros((3 * len(chords) * len(spans)))
zetaPri[0::3] = -0.5
nu = np.zeros_like(zetaPri)
kk = 0
for c in chords:
for s in spans:
zeta[3 * kk] = c
zeta[3 * kk + 1] = s
kk = kk + 1
# end for s
# end for c
kk = 0
for c in chordsW:
for s in spans:
zetaW[3 * kk] = c
zetaW[3 * kk + 1] = s
kk = kk + 1
# end for s
# end for c
# generate model
E, F, G, C, D = genSSuvlm(gam, gamW, gamPri, zeta, zetaW, zetaPri, nu, m,
n, mW, delS, imageMeth)
# convert inputs from general kinematics to aerofil DoFs
T = np.zeros((9 * (m + 1) * (n + 1), 5))
for i in range(m + 1):
for j in range(n + 1):
q = i * (n + 1) + j
# alpha, alphaPrime
T[3 * (m + 1) * (n + 1) + 3 * q + 2,
0] = -(zeta[3 * q] + 0.25 / m - e)
T[3 * q + 2, 1] = -(zeta[3 * q] + 0.25 / m - e)
# plunge
T[3 * q + 2, 2] = -1
# beta, betaPrime
if zeta[3 * q] + 0.25 / m > f:
T[3 * (m + 1) * (n + 1) + 3 * q + 2,
3] = -(zeta[3 * q] + 0.25 / m - f)
T[3 * q + 2, 4] = -(zeta[3 * q] + 0.25 / m - f)
G_s = np.dot(G, T)
D_s = np.dot(D, T)
# get coefficients as output
T_coeff = np.zeros((3, 3 * (m + 1) * (n + 1)))
T_coeff[0, 0::3] = 1.0 #drag
T_coeff[1, 2::3] = 1.0 #lift
for i in range(m + 1):
for j in range(n + 1):
q = i * (n + 1) + j
# moment = r*L, +ve nose-up, at quarter chord
T_coeff[2, 3 * q + 2] = -(zeta[3 * q] + 0.25 / m - 0.25)
C_coeff = np.dot(T_coeff, C)
D_coeff = np.dot(T_coeff, D)
D_s_coeff = np.dot(T_coeff, D_s)
# spanwise lift distribution as output
T_span = np.zeros((n + 1, 3 * (m + 1) * (n + 1)))
for jj in range(n + 1):
T_span[jj, 3 * jj + 2::3 * (n + 1)] = 1.0
if writeToMat == True:
fileName = Settings.OutputDir + 'rectWingAR' + str(AR) + '_m' + str(
m) + 'mW' + str(mW) + 'n' + str(n) + 'delS' + str(delS)
if e != 0.25:
fileName += 'e' + str(e)
if f != 0.75:
fileName += 'f' + str(f)
if imageMeth != False:
fileName += 'half'
savemat(
fileName, {
'E': E,
'F': F,
'G': G,
'C': C,
'D': D,
'm': m,
'mW': mW,
'delS': delS,
'G_s': G_s,
'D_s': D_s,
'C_coeff': C_coeff,
'D_coeff': D_coeff,
'D_s_coeff': D_s_coeff,
'T_coeff': T_coeff,
'T_span': T_span,
'AR': AR,
'm': m,
'mW': mW,
'n': n,
'zeta': zeta
}, True)
# end if
return E, F, G, C, D
def test_save_unicode_field(tmpdir):
filename = os.path.join(str(tmpdir), 'test.mat')
test_dict = {u'a': {u'b': 1, u'c': 'test_str'}}
savemat(filename, test_dict)
def _rt_check_case(name, expected, format):
mat_stream = BytesIO()
savemat(mat_stream, expected, format=format)
mat_stream.seek(0)
_load_check_case(name, [mat_stream], expected)
def Solve_Py(XBINPUT, XBOPTS, VMOPTS, VMINPUT, AELAOPTS):
"""Nonlinear static solver using Python to solve aeroelastic
equation. Assembly of structural matrices is carried out with
Fortran subroutines. Aerodynamics solved using PyAero.UVLM."""
assert XBOPTS.Solution.value == 112, ('NonlinearStatic requested' +
' with wrong solution code')
# Initialize beam.
XBINPUT, XBOPTS, NumNodes_tot, XBELEM, PosIni, PsiIni, XBNODE, NumDof \
= BeamInit.Static(XBINPUT,XBOPTS)
# Set initial conditions as undef config.
PosDefor = PosIni.copy(order='F')
PsiDefor = PsiIni.copy(order='F')
if XBOPTS.PrintInfo.value == True:
sys.stdout.write('Solve nonlinear static case in Python ... \n')
# Initialise structural eqn tensors.
KglobalFull = np.zeros((NumDof.value, NumDof.value), ct.c_double, 'F')
ks = ct.c_int()
FglobalFull = np.zeros((NumDof.value, NumDof.value), ct.c_double, 'F')
fs = ct.c_int()
DeltaS = np.zeros(NumDof.value, ct.c_double, 'F')
Qglobal = np.zeros(NumDof.value, ct.c_double, 'F')
x = np.zeros(NumDof.value, ct.c_double, 'F')
dxdt = np.zeros(NumDof.value, ct.c_double, 'F')
# Beam Load Step tensors
Force = np.zeros((XBINPUT.NumNodesTot, 6), ct.c_double, 'F')
iForceStep = np.zeros((NumNodes_tot.value, 6), ct.c_double, 'F')
iForceStep_Dof = np.zeros(NumDof.value, ct.c_double, 'F')
# Initialze Aero.
Section = InitSection(VMOPTS, VMINPUT, AELAOPTS.ElasticAxis)
# Declare memory for Aero grid and velocities.
Zeta = np.zeros((Section.shape[0], PosDefor.shape[0], 3), ct.c_double, 'C')
ZetaDot = np.zeros((Section.shape[0], PosDefor.shape[0], 3), ct.c_double,
'C')
# Additional Aero solver variables.
AeroForces = np.zeros((VMOPTS.M.value + 1, VMOPTS.N.value + 1, 3),
ct.c_double, 'C')
Gamma = np.zeros((VMOPTS.M.value, VMOPTS.N.value), ct.c_double, 'C')
# Init external velocities.
Uext = InitSteadyExternalVels(VMOPTS, VMINPUT)
# Create zero triads for motion of reference frame.
VelA_A = np.zeros((3))
OmegaA_A = np.zeros((3))
# Create zero vectors for structural vars not used in static analysis.
PosDotDef = np.zeros_like(PosDefor, ct.c_double, 'F')
PsiDotDef = np.zeros_like(PsiDefor, ct.c_double, 'F')
# Define tecplot stuff.
FileName = Settings.OutputDir + Settings.OutputFileRoot + 'AeroGrid.dat'
Variables = ['X', 'Y', 'Z', 'Gamma']
FileObject = PostProcess.WriteAeroTecHeader(FileName, 'Default', Variables)
# Start Load Loop.
for iLoadStep in range(XBOPTS.MaxIterations.value):
# Reset convergence parameters and loads.
Iter = 0
ResLog10 = 0.0
Force[:, :] = 0.0
AeroForces[:, :, :] = 0.0
oldPos = PosDefor.copy(order='F')
oldPsi = PsiDefor.copy(order='F')
# Calculate aero loads.
if hasattr(XBINPUT, 'ForcedVel'):
CoincidentGrid(PosDefor,
PsiDefor,
Section,
XBINPUT.ForcedVel[0, :3],
XBINPUT.ForcedVel[0, 3:],
PosDotDef,
PsiDotDef,
XBINPUT,
Zeta,
ZetaDot,
ctrlSurf=VMINPUT.ctrlSurf)
else:
CoincidentGrid(PosDefor,
PsiDefor,
Section,
VelA_A,
OmegaA_A,
PosDotDef,
PsiDotDef,
XBINPUT,
Zeta,
ZetaDot,
ctrlSurf=VMINPUT.ctrlSurf)
if hasattr(XBINPUT, 'ForcedVel'):
ZetaStar, GammaStar = InitSteadyWake(VMOPTS, VMINPUT, Zeta,
XBINPUT.ForcedVel[0, :3])
else:
ZetaStar, GammaStar = InitSteadyWake(VMOPTS, VMINPUT, Zeta)
# Define AICs here for debgugging - Rob 16/08/2016
AIC = np.zeros(
(VMOPTS.M.value * VMOPTS.N.value, VMOPTS.M.value * VMOPTS.N.value),
ct.c_double, 'C')
BIC = np.zeros(
(VMOPTS.M.value * VMOPTS.N.value, VMOPTS.M.value * VMOPTS.N.value),
ct.c_double, 'C')
# Solve for AeroForces.
UVLMLib.Cpp_Solver_VLM(Zeta, ZetaDot, Uext, ZetaStar, VMOPTS,
AeroForces, Gamma, GammaStar, AIC, BIC)
AeroForces[:, :, :] = AELAOPTS.AirDensity * AeroForces[:, :, :]
# Write solution to tecplot file.
PostProcess.WriteUVLMtoTec(FileObject,
Zeta,
ZetaStar,
Gamma,
GammaStar,
iLoadStep,
XBOPTS.NumLoadSteps.value,
iLoadStep * 1.0,
Text=True)
# Map AeroForces to beam.
CoincidentGridForce(XBINPUT, PsiDefor, Section, AeroForces, Force)
# Add gravity loads.
AddGravityLoads(Force, XBINPUT, XBELEM, AELAOPTS, PsiDefor, VMINPUT.c)
# Apply factor corresponding to force step.
if iLoadStep < XBOPTS.NumLoadSteps.value:
iForceStep = (Force + XBINPUT.ForceStatic)*float( (iLoadStep+1) ) / \
XBOPTS.NumLoadSteps.value
else:
# continue at full loading until equilibrium
iForceStep = Force + XBINPUT.ForceStatic
if XBOPTS.PrintInfo.value == True:
sys.stdout.write(' iLoad: %-10d\n' % (iLoadStep + 1))
sys.stdout.write(' SubIter DeltaF DeltaX ResLog10\n')
# Start Newton Iteration.
while ((ResLog10 > np.log10(XBOPTS.MinDelta.value))
& (Iter < XBOPTS.MaxIterations.value)):
Iter += 1
if XBOPTS.PrintInfo.value == True:
sys.stdout.write(' %-7d ' % (Iter))
# Set structural eqn tensors to zero
KglobalFull[:, :] = 0.0
ks = ct.c_int()
FglobalFull[:, :] = 0.0
fs = ct.c_int()
Qglobal[:] = 0.0
# Assemble matrices for static problem
BeamLib.Cbeam3_Asbly_Static(XBINPUT, NumNodes_tot, XBELEM, XBNODE,
PosIni, PsiIni, PosDefor, PsiDefor,
iForceStep, NumDof, ks, KglobalFull,
fs, FglobalFull, Qglobal, XBOPTS)
# Get state vector from current deformation.
PosDot = np.zeros((NumNodes_tot.value, 3), ct.c_double, 'F')
PsiDot = np.zeros((XBINPUT.NumElems, Settings.MaxElNod, 3),
ct.c_double, 'F')
BeamLib.Cbeam_Solv_Disp2State(NumNodes_tot, NumDof, XBINPUT,
XBNODE, PosDefor, PsiDefor, PosDot,
PsiDot, x, dxdt)
# Get forces on unconstrained nodes.
BeamLib.f_fem_m2v(
ct.byref(NumNodes_tot), ct.byref(ct.c_int(6)),
iForceStep.ctypes.data_as(ct.POINTER(ct.c_double)),
ct.byref(NumDof),
iForceStep_Dof.ctypes.data_as(ct.POINTER(ct.c_double)),
XBNODE.Vdof.ctypes.data_as(ct.POINTER(ct.c_int)))
# Calculate \Delta RHS.
Qglobal = Qglobal - np.dot(FglobalFull, iForceStep_Dof)
# Calculate \Delta State Vector.
DeltaS = -np.dot(np.linalg.inv(KglobalFull), Qglobal)
if XBOPTS.PrintInfo.value == True:
sys.stdout.write('%-10.4e %-10.4e ' %
(max(abs(Qglobal)), max(abs(DeltaS))))
# Update Solution.
BeamLib.Cbeam3_Solv_Update_Static(XBINPUT, NumNodes_tot, XBELEM,
XBNODE, NumDof, DeltaS, PosIni,
PsiIni, PosDefor, PsiDefor)
# Record residual at first iteration.
if (Iter == 1):
Res0_Qglobal = max(abs(Qglobal)) + 1.e-16
Res0_DeltaX = max(abs(DeltaS)) + 1.e-16
# Update residual and compute log10
Res_Qglobal = max(abs(Qglobal)) + 1.e-16
Res_DeltaX = max(abs(DeltaS)) + 1.e-16
ResLog10 = max([
np.log10(Res_Qglobal / Res0_Qglobal),
np.log10(Res_DeltaX / Res0_DeltaX)
])
if XBOPTS.PrintInfo.value == True:
sys.stdout.write('%8.4f\n' % (ResLog10))
# Stop the solution.
if (ResLog10 > 10.):
sys.stderr.write(' STOP\n')
sys.stderr.write(' The max residual is %e\n' % (ResLog10))
exit(1)
elif Res_DeltaX < 1.e-14:
break
# END Newton iteration
# After incremental loading continue until equilibrium reached
if iLoadStep >= XBOPTS.NumLoadSteps.value:
Pos_error = PosDefor - oldPos
Psi_error = PsiDefor - oldPsi
if( (np.linalg.norm(Pos_error)<=XBOPTS.MinDelta) & \
(np.linalg.norm(Psi_error)<=XBOPTS.MinDelta) ):
break
# END Load step loop
if XBOPTS.PrintInfo.value == True:
sys.stdout.write(' ... done\n')
# Write deformed configuration to file.
ofile = Settings.OutputDir + Settings.OutputFileRoot + '_SOL112_def.dat'
if XBOPTS.PrintInfo.value == True:
sys.stdout.write('Writing file %s ... ' % (ofile))
fp = open(ofile, 'w')
fp.write('TITLE="Non-linear static solution: deformed geometry"\n')
fp.write('VARIABLES="iElem" "iNode" "Px" "Py" "Pz" "Rx" "Ry" "Rz"\n')
fp.close()
if XBOPTS.PrintInfo.value == True:
sys.stdout.write('done\n')
WriteMode = 'a'
BeamIO.OutputElems(XBINPUT.NumElems, NumNodes_tot.value, XBELEM, PosDefor,
PsiDefor, ofile, WriteMode)
# Print deformed configuration.
if XBOPTS.PrintInfo.value == True:
sys.stdout.write('--------------------------------------\n')
sys.stdout.write('NONLINEAR STATIC SOLUTION\n')
sys.stdout.write('%10s %10s %10s\n' % ('X', 'Y', 'Z'))
for inodi in range(NumNodes_tot.value):
sys.stdout.write(' ')
for inodj in range(3):
sys.stdout.write('%12.5e' % (PosDefor[inodi, inodj]))
sys.stdout.write('\n')
sys.stdout.write('--------------------------------------\n')
# Close Tecplot ascii FileObject.
PostProcess.CloseAeroTecFile(FileObject)
if True:
# print and save deformed wing total force coefficients (may include gravity
# and other applied static loads). Coefficients in wind-axes.
cF = np.zeros((3))
for i in range(VMOPTS.M.value + 1):
for j in range(VMOPTS.N.value + 1):
cF += AeroForces[i, j, :]
Calpha = Psi2TransMat(np.array([VMINPUT.alpha, 0.0, 0.0]))
cF = np.dot(Calpha, cF)
cF = cF / (0.5 * AELAOPTS.AirDensity * np.linalg.norm(VMINPUT.U_infty)
**2.0 * VMINPUT.c * XBINPUT.BeamLength)
print("Reference condition total force coefficients: {}".format(cF))
fileName = Settings.OutputDir + Settings.OutputFileRoot + 'refCf'
savemat(fileName, {'cF': cF})
# Return solution
return PosDefor, PsiDefor, Zeta, ZetaStar, Gamma, GammaStar, iForceStep, NumNodes_tot, NumDof, XBELEM, XBNODE, PosIni, PsiIni, Uext
plt.show()
# save res to mat-file
savemat('turn_result.mat', {
'X': X,
'Y': Y,
'psi': psi,
'r': r,
'vx': vx,
'vy': vy,
'delta': delta,
'ddelta': ddelta,
'alphaf': alphaf,
'alphar': alphar,
'kappaf': kappaf,
'kappar': kappar,
'omegaf': omegaf,
'omegar': omegar,
'Fyf': Fyf,
'Fyr': Fyr,
'Fxf': Fxf,
'Fxr': Fxr,
'Twf': Twf,
'Twr': Twr,
'FX': FX,
'FY': FY,
'MZ': MZ,
't': t
},
appendmat=False)
def test_single_object():
stream = BytesIO()
savemat(stream, {'A': np.array(1, dtype=object)})
if f != 0.75:
fileName += 'f' + str(f)
if imageMeth != False:
fileName += 'half'
savemat(
fileName, {
'E': E,
'F': F,
'G': G,
'C': C,
'D': D,
'm': m,
'mW': mW,
'delS': delS,
'G_s': G_s,
'D_s': D_s,
'C_coeff': C_coeff,
'D_coeff': D_coeff,
'D_s_coeff': D_s_coeff,
'T_coeff': T_coeff,
'T_span': T_span,
'AR': span / chord,
'm': m,
'mW': mW,
'n': n,
'zeta': zeta
}, True)
if runLinear == True:
nT = 4001 # number of time steps
u = np.zeros((nT, G_s.shape[1])) # inputs
def setUp(self):
self.prior = np.array([1, 0, 0])
self.transmat = np.matrix([[0, 1, 0], [0, 0, 1], [0, 0, 1]])
self.mu = np.zeros((2, 3, 2))
self.mu[:, :, 0] = np.array([[1, 2, 3], [1, 2, 3]])
self.mu[:, :, 1] = np.array([[4, 5, 6], [4, 5, 6]])
self.Sigma = np.zeros((2, 2, 3, 2))
for i in range(3):
self.Sigma[:, :, i, 0] = np.diag(np.ones((2, )) * 0.01)
self.Sigma[:, :, i, 1] = np.diag(np.ones((2, )) * 0.01)
self.mixmat = np.array([[.5, .5], [.5, .5], [.5, .5]])
try:
with open('MhmmEM2DTestGaussInit.cache', 'rb') as f:
cache = load(f)
self.obs = cache['obs']
self.prior0 = cache['prior0']
self.transmat0 = cache['transmat0']
self.mu0 = cache['mu0']
self.Sigma0 = cache['Sigma0']
self.mixmat0 = cache['mixmat0']
except:
self.obs, hidden = mhmm_sample(T=4,
numex=100,
initial_prob=self.prior,
transmat=self.transmat,
mu=self.mu,
Sigma=self.Sigma,
mixmat=self.mixmat)
self.prior0, _ = mk_stochastic(np.random.rand(3))
self.transmat0, _ = mk_stochastic(np.random.rand(3, 3))
O = self.obs.shape[0]
M = 2
Q = 3
mu0, Sigma0, weights0 = mixgauss_init(Q * M,
self.obs,
cov_type='diag')
self.mu0 = np.transpose(np.reshape(mu0, (O, M, Q)), (0, 2, 1))
self.Sigma0 = np.transpose(np.reshape(Sigma0, (O, O, M, Q)),
(0, 1, 3, 2))
self.mixmat0, _ = mk_stochastic(np.random.rand(Q, M))
cache = {
'obs': self.obs,
'prior0': self.prior0,
'transmat0': self.transmat0,
'mu0': self.mu0,
'Sigma0': self.Sigma0,
'mixmat0': self.mixmat0
}
with open('MhmmEM2DTestGaussInit.cache', 'wb') as f:
dump(cache, f)
savemat('MhmmEM2DTestGaussInit.mat', cache)
