def test_pdos_coord_trans():
coords = {}
# cartesian: first axis is the time axis
coords['cart'] = rand((100, 10, 3))
# cartesian scaled, e.g. Angstrom instead of Bohr
coords['cart2'] = coords['cart'] * 5
# some other coord sys
cell1 = rand((3, 3))
# coord_trans: axis=-1 specifies the "x,y,z"-axis of dimension 3
coords['cell1'] = coord_trans(coords['cart'],
old=np.identity(3),
new=cell1,
axis=-1)
dos = {}
for key, val in coords.items():
dos[key] = pdos(val)
np.testing.assert_array_almost_equal(dos['cart'], dos['cart2'])
assert not tools.array_almost_equal(dos['cart'], dos['cell1'])
def test_pdos_coord_trans():
coords = {}
# cartesian: first axis is the time axis
coords['cart'] = rand((100, 10, 3))
# cartesian scaled, e.g. Angstrom instead of Bohr
coords['cart2'] = coords['cart']*5
# some other coord sys
cell1 = rand((3,3))
# coord_trans: axis=-1 specifies the "x,y,z"-axis of dimension 3
coords['cell1'] = coord_trans(coords['cart'],
old=np.identity(3),
new=cell1,
axis=-1)
dos = {}
for key, val in coords.iteritems():
dos[key] = pdos(val)
np.testing.assert_array_almost_equal(dos['cart'], dos['cart2'])
assert not tools.array_almost_equal(dos['cart'], dos['cell1'])
def test_tools():
x1 = {
'a': 3,
type(1): 'foo',
'b': {
'aa': 1,
'bb': np.array([1, 2]),
'cc': {
'aaa': np.array([1, 2.0]),
'bbb': np.array([2, 4.0])
}
}
}
x2 = copy.deepcopy(x1)
tools.assert_dict_with_all_types_equal(x1, x2)
tools.assert_all_types_equal(x1, x2)
# not equal array
x2['b']['cc']['bbb'] *= 2.0
assert not tools.all_types_equal(x1, x2)
# almost equal array and float
x2 = copy.deepcopy(x1)
x2['b']['cc']['bbb'] += 1e-5
x2['b']['aa'] += 1e-5
tools.assert_all_types_almost_equal(x1, x2)
# sub-dict different (keys don't match)
x2 = copy.deepcopy(x1)
x2['b']['cc']['fff'] = 'extra'
assert not tools.all_types_equal(x1, x2)
# test only some keys of a dict
tools.assert_dict_with_all_types_equal({
'a': 1,
'b': 1
}, {
'a': 1,
'b': 3
},
keys=['a'])
# simple stuff
tools.assert_all_types_equal(1, 1)
tools.assert_all_types_equal(1.0, 1.0)
tools.assert_all_types_equal(1.0, 1)
tools.assert_all_types_equal(1, 1.0)
tools.assert_all_types_equal([1], [1])
tools.assert_all_types_equal([1], [1.0])
tools.assert_all_types_equal('a', 'a')
tools.assert_all_types_almost_equal(1.0, 1.0 + 1e-5)
tools.assert_all_types_almost_equal(np.array([1.0]),
np.array([1.0 + 1e-5]))
assert not tools.all_types_equal(1, 2)
assert not tools.all_types_equal(1.0, 1.1)
assert not tools.all_types_equal([1], [1, 2])
assert not tools.all_types_equal('a', 'b')
# different types not allowed if strict=True
assert not tools.all_types_equal(1.0, 1, strict=True)
# test keys=[...], i.e. ignore some keys in both dicts
x2 = copy.deepcopy(x1)
x2['c'] = 1.0
assert tools.dict_with_all_types_equal(x1, x2, keys=['a', 'b', type(1)])
# fail on same array content but different dimensions
a = np.random.rand(1, 2, 3)
b = a[None, ...]
# this is True and IMHO a numpy bug b/c the dimensions are
# different
assert (a == b).all()
# must catch that here
assert not tools.array_equal(a, b)
a = np.random.rand(1, 2, 3)
b = (a + 1e-8)[None, ...]
assert not (a == b).all() # make sure they are numerically differrent
assert np.allclose(a, b) # True but should be False
assert not tools.array_almost_equal(a, b) # ok
flt = 1.0
np_flt = np.array([1.0 + 1e-9])[0]
assert not tools.all_types_equal(flt, np_flt)
assert not tools.all_types_equal(np_flt, flt)
assert tools.all_types_almost_equal(flt, np_flt)
assert tools.all_types_almost_equal(np_flt, flt)
def test_tools():
x1 = {'a': 3,
type(1): 'foo',
'b': {'aa': 1,
'bb': np.array([1,2]),
'cc': {'aaa': np.array([1,2.0]),
'bbb': np.array([2,4.0])}}}
x2 = copy.deepcopy(x1)
tools.assert_dict_with_all_types_equal(x1, x2)
tools.assert_all_types_equal(x1, x2)
# not equal array
x2['b']['cc']['bbb'] *= 2.0
assert not tools.all_types_equal(x1, x2)
# almost equal array and float
x2 = copy.deepcopy(x1)
x2['b']['cc']['bbb'] += 1e-5
x2['b']['aa'] += 1e-5
tools.assert_all_types_almost_equal(x1, x2)
# sub-dict different (keys don't match)
x2 = copy.deepcopy(x1)
x2['b']['cc']['fff'] = 'extra'
assert not tools.all_types_equal(x1, x2)
# test only some keys of a dict
tools.assert_dict_with_all_types_equal({'a':1,'b':1}, {'a':1, 'b':3},
keys=['a'])
# simple stuff
tools.assert_all_types_equal(1, 1)
tools.assert_all_types_equal(1.0, 1.0)
tools.assert_all_types_equal(1.0, 1)
tools.assert_all_types_equal(1, 1.0)
tools.assert_all_types_equal([1], [1])
tools.assert_all_types_equal([1], [1.0])
tools.assert_all_types_equal('a', 'a')
tools.assert_all_types_almost_equal(1.0, 1.0+1e-5)
tools.assert_all_types_almost_equal(np.array([1.0]), np.array([1.0+1e-5]))
assert not tools.all_types_equal(1, 2)
assert not tools.all_types_equal(1.0, 1.1)
assert not tools.all_types_equal([1], [1,2])
assert not tools.all_types_equal('a', 'b')
# different types not allowed if strict=True
assert not tools.all_types_equal(1.0, 1, strict=True)
# test keys=[...], i.e. ignore some keys in both dicts
x2 = copy.deepcopy(x1)
x2['c'] = 1.0
assert tools.dict_with_all_types_equal(x1, x2, keys=['a','b',type(1)])
# fail on same array content but different dimensions
a = np.random.rand(1,2,3)
b = a[None,...]
# this is True and IMHO a numpy bug b/c the dimensions are
# different
assert (a==b).all()
# must catch that here
assert not tools.array_equal(a,b)
a = np.random.rand(1,2,3)
b = (a + 1e-8)[None,...]
assert not (a==b).all() # make sure they are numerically differrent
assert np.allclose(a,b) # True but should be False
assert not tools.array_almost_equal(a,b) # ok
