def test_normalize(self):
df_target = bu.col_matrix(v=dict(x=1, y=1))
df_true = bu.col_matrix(v=dict(x=1/np.sqrt(2), y=1/np.sqrt(2)))
df_res = bu.normalize(df_target)
self.assertTrue(df_true.equals(df_res))
def setUp(self):
self.df_basis = \
bu.col_matrix(
v1 = dict(x=1, y=1),
v2 = dict(x=1, y=-1)
)
self.df_basis_ill = \
bu.col_matrix(
v1 = dict(x=1, y=1),
v2 = dict(x=1, y=1 - 1e-9)
)
self.df_b1 = bu.col_matrix(
v = dict(x=1., y=1.)
).sort_values("rowname").reset_index(drop=True)
self.df_b1_ill = bu.col_matrix(
v = dict(x=1., y=1.),
w = dict(x=1., y=1.)
).sort_values("rowname").reset_index(drop=True)
self.df_x1 = bu.col_matrix(
v = dict(v1=1., v2=0.)
).sort_values("rowname").reset_index(drop=True)
def test_add_col(self):
df_ful = bu.col_matrix(
rho = dict(M=1, L=-3),
U = dict(L=1, T=-1),
D = dict(L=1)
).sort_values("rowname") \
.reset_index(drop=True)
df_sub = bu.col_matrix(rho=dict(M=1, L=-3), U=dict(L=1, T=-1))
df_res = bu.add_col(df_sub, D=dict(L=1)) \
.sort_values("rowname") \
.reset_index(drop=True)
pd.testing.assert_frame_equal(df_ful,
df_res,
check_exact=False,
check_dtype=False,
check_column_type=False)
with self.assertRaises(ValueError):
bu.add_col(pd.DataFrame(), D=dict(M=1))
with self.assertWarns(Warning):
bu.add_col(df_sub, D=[0, 1, 0])
def test_complete(self):
df_dim = bu.col_matrix(
rho = dict(M=1, L=-3),
U = dict(L=1, T=-1),
D = dict(L=1),
mu = dict(M=1, L=-1, T=-1),
eps = dict(L=1)
)
df = bu.col_matrix(Re=dict(rho=1, U=1, D=1, mu=-1))
df_res = bu.complete(df, df_dim)
df_true = bu.col_matrix(Re=dict(rho=1.0, U=1.0, D=1.0, mu=-1.0, eps=0.0))
# Original weights not scrambled
self.assertTrue(
df_res[["rowname", "Re"]]
.sort_values("rowname")
.reset_index(drop=True)
.equals(
df_true
.sort_values("rowname")
.reset_index(drop=True)
)
)
# Given basis dimensionless
self.assertTrue(np.all(
np.abs(bu.inner(df_dim, df_res).drop("rowname", axis=1).values) < 1e-6
))
def setUp(self):
self.df_dim = bu.col_matrix(
x = dict(M=1),
y = dict(M=1)
)
self.df_pi = bu.col_matrix(
pi0 = dict(x=-1 / np.sqrt(2), y=+1 / np.sqrt(2))
)
def test_col_matrix(self):
df_dim = bu.col_matrix(
rho=dict(M=1, L=-3),
U=dict(L=1, T=-1),
D=dict(L=1),
mu=dict(M=1, L=-1, T=-1),
eps=dict(L=1)).sort_values("rowname").reset_index(drop=True)
self.assertTrue(self.df_dim[df_dim.columns].equals(df_dim))
def test_nondim(self):
df_dim = bu.row_matrix(L = dict(u=1, v=1))
df = bu.col_matrix(q = dict(L=1))
df_res = bu.nondim(df, df_dim) \
.sort_values("rowname") \
.reset_index(drop=True)
df_true = bu.col_matrix(q = dict(u=-0.5, v=-0.5)) \
.sort_values("rowname") \
.reset_index(drop=True)
pd.testing.assert_frame_equal(
df_res,
df_true,
check_exact=False
)
with self.assertRaises(ValueError):
bu.nondim(pd.DataFrame(), df_dim)
with self.assertRaises(ValueError):
bu.nondim(df, pd.DataFrame())
with self.assertRaises(ValueError):
bu.nondim(bu.col_matrix(q = dict(T=1)), df_dim)
def test_inner(self):
df = bu.col_matrix(
v = dict(x=+1, y=+1),
w = dict(x=-1, y=+1)
)
# Full weights
df_w1 = bu.col_matrix(z = dict(v=1, w=1))
df_r1 = bu.inner(df, df_w1)
df_t1 = bu.col_matrix(z = dict(x=0, y=2))
# Missing weights
df_w2 = bu.col_matrix(z = dict(v=1))
df_r2 = bu.inner(df, df_w2)
df_t2 = bu.col_matrix(z = dict(x=+1, y=+1))
pd.testing.assert_frame_equal(
df_r1,
df_t1,
check_exact=False
)
pd.testing.assert_frame_equal(
df_r2,
df_t2,
check_exact=False
)
with self.assertRaises(ValueError):
bu.inner(pd.DataFrame(), df_w1)
with self.assertRaises(ValueError):
bu.inner(df, pd.DataFrame())
with self.assertRaises(ValueError):
bu.inner(bu.col_matrix(w=dict(x=1, y=1)), df_w1)
def test_rowname(self):
df_tmp = bu.col_matrix(x=dict(a=1), rowname="foo")
df_ref = pd.DataFrame({"foo": ["a"], "x": [1]})
self.assertTrue(df_ref.equals(df_tmp))
def setUp(self):
self.df = bu.col_matrix(v = dict(x=+1, y=+1))
self.df_v1 = bu.col_matrix(w = dict(x=+1, y=-1))
self.df_v2 = bu.col_matrix(w = dict(x=+1, y=+1))
def test_rank(self):
df_low = bu.col_matrix(v1=dict(x=1, y=1), v2=dict(x=2, y=2))
df_full = bu.col_matrix(v1=dict(x=1, y=1), v2=dict(x=1, y=0))
self.assertTrue(bu.rank(df_low) == 1)
self.assertTrue(bu.rank(df_full) == 2)