class TestTrivalTransform:
"""
Testing of spherical harmonics transform for 1D fields.
"""
def setup_method(self):
self.l_max = 0
self.m_max = 0
self.n_lat = 1
self.n_lon = 1
self.sht = SHT(self.l_max, self.m_max, self.n_lon, self.n_lat)
def test_transform(self):
"""
Tests the Legendre expansion by transforming a given Legendre polynomial
and ensuring that only one component in the output is set.
"""
z = np.ones((1, 1))
zz = self.sht.synthesize(self.sht.transform(z))
assert np.all(np.isclose(z, zz))
def test_evaluate(self):
"""
Test that evaluating the spectral representation reproduces the spatial
input.
"""
coeffs = np.ones(1)
points = np.random.rand(10, 2)
zz = self.sht.evaluate(coeffs, points)
assert np.all(np.isclose(zz, 1.0))
class TestAddition:
"""
Test addition of SHT coefficient vectors and matrices.
"""
def setup_method(self):
n_lon = 64
n_lat = 64
self.l_max_l = np.random.randint(10, 20)
self.m_max_l = self.l_max_l #np.random.randint(9, self.l_max_l)
self.sht_l = SHT(self.l_max_l, self.m_max_l, n_lon, n_lat)
self.l_max_l_inc = np.random.randint(5, 10)
self.m_max_l_inc = self.l_max_l_inc #np.random.randint(4, self.l_max_l_inc)
self.sht_l_inc = SHT(self.l_max_l_inc, self.m_max_l_inc, n_lon, n_lat)
self.l_max_r = np.random.randint(20, 24)
self.m_max_r = self.l_max_r #np.random.randint(19, self.l_max_r)
self.sht_r = SHT(self.l_max_r, self.m_max_r, n_lon, n_lat)
self.l_max_r_inc = np.random.randint(10, 20)
self.m_max_r_inc = self.l_max_r_inc #np.random.randint(9, self.l_max_r_inc)
self.sht_r_inc = SHT(self.l_max_r_inc, self.m_max_r_inc, n_lon, n_lat)
self.coeff_vector_l = np.random.rand(
self.sht_l.get_n_spectral_coeffs())
self.coeff_vector_l = self.coeff_vector_l + 1j * np.random.rand(
self.sht_l.get_n_spectral_coeffs())
self.coeff_matrix_l = np.random.rand(
self.sht_l_inc.get_n_spectral_coeffs_cmplx(),
self.sht_l.get_n_spectral_coeffs())
self.coeff_matrix_l = self.coeff_matrix_l + 1j * np.random.rand(
self.sht_l_inc.get_n_spectral_coeffs_cmplx(),
self.sht_l.get_n_spectral_coeffs())
self.coeff_vector_r = np.random.rand(
self.sht_r.get_n_spectral_coeffs())
self.coeff_vector_r = self.coeff_vector_r + 1j * np.random.rand(
self.sht_r.get_n_spectral_coeffs())
self.coeff_matrix_r = np.random.rand(
self.sht_r_inc.get_n_spectral_coeffs_cmplx(),
self.sht_r.get_n_spectral_coeffs())
self.coeff_matrix_r = self.coeff_matrix_r + 1j * np.random.rand(
self.sht_r_inc.get_n_spectral_coeffs_cmplx(),
self.sht_r.get_n_spectral_coeffs())
def test_addition(self):
"""
Test addition of coefficient vectors.
"""
sum_2l = SHT.add_coeffs(self.sht_l, self.coeff_vector_l, self.sht_l,
self.coeff_vector_l)
sum_r = SHT.add_coeffs(self.sht_r, self.coeff_vector_r, self.sht_l,
self.coeff_vector_l)
sum_2r = SHT.add_coeffs(self.sht_r, self.coeff_vector_r, self.sht_r,
self.coeff_vector_r)
z_l = self.sht_l.synthesize(self.coeff_vector_l)
z_r = self.sht_r.synthesize(self.coeff_vector_r)
z_2l = self.sht_l.synthesize(sum_2l)
z_2r = self.sht_r.synthesize(sum_2r)
z_rs = self.sht_r.synthesize(sum_r)
coeff_vector_rl = self.sht_l.transform(z_r)
z_rl = self.sht_l.synthesize(coeff_vector_rl)
sum_l = SHT.add_coeffs(self.sht_l, self.coeff_vector_l, self.sht_l,
coeff_vector_rl)
z_ls = self.sht_l.synthesize(sum_l)
assert np.all(np.isclose(2.0 * z_l, z_2l))
assert np.all(np.isclose(2.0 * z_r, z_2r))
assert np.all(np.isclose(z_rs, z_r + z_l))
assert np.all(np.isclose(z_ls, z_l + z_rl))
def test_addition_matrix(self):
"""
Test addition of coefficient vectors.
"""
sum_2l = SHT.add_coeffs(self.sht_l_inc, self.sht_l,
self.coeff_matrix_l, self.sht_l_inc,
self.sht_l, self.coeff_matrix_l)
z_l = synthesize_matrix(self.sht_l_inc, self.sht_l,
self.coeff_matrix_l)
z_2l = synthesize_matrix(self.sht_l_inc, self.sht_l, sum_2l)
assert np.all(np.isclose(2.0 * z_l, z_2l))
z_r = synthesize_matrix(self.sht_r_inc, self.sht_r,
self.coeff_matrix_r)
sum_rl = SHT.add_coeffs(self.sht_r_inc, self.sht_r,
self.coeff_matrix_r, self.sht_l_inc,
self.sht_l, self.coeff_matrix_l)
z_rs = synthesize_matrix(self.sht_r_inc, self.sht_r, sum_rl)
assert np.all(np.isclose(z_r + z_l, z_rs))
coeff_matrix_rl = transform_matrix(self.sht_l_inc, self.sht_l, z_r)
sum_lr = SHT.add_coeffs(self.sht_l_inc, self.sht_l,
self.coeff_matrix_l, self.sht_l_inc,
self.sht_l, coeff_matrix_rl)
z_ls = synthesize_matrix(self.sht_l_inc, self.sht_l, sum_lr)
z_rl = synthesize_matrix(self.sht_l_inc, self.sht_l, coeff_matrix_rl)
assert np.all(np.isclose(z_ls, z_rl + z_l))
class TestSHT:
"""
Testing of spherical harmonics transform for non-degenrate 2D fields.
"""
def setup_method(self):
self.l_max = np.random.randint(20, 100)
self.m_max = np.random.randint(10, self.l_max)
self.m_max = self.l_max #np.random.randint(10, self.l_max)
self.n_lat = 4 * self.l_max
self.n_lon = 4 * self.l_max
self.sht = SHT(self.l_max, self.m_max, self.n_lon, self.n_lat)
self.lat_grid = np.linspace(0, np.pi, self.n_lat + 1)
self.lat_grid = 0.5 * (self.lat_grid[1:] + self.lat_grid[:-1])
self.lon_grid = np.linspace(0, 2.0 * np.pi, self.n_lon + 1)[:-1]
def test_latitude_grid(self):
"""
Test that the SHT class returns the expected latitude grid (Gauss-Legendre).
"""
lat_grid_ref = self.lat_grid
lat_grid = self.sht.get_latitude_grid()
assert np.all(np.isclose(lat_grid_ref, lat_grid))
def test_colatitude_grid(self):
"""
Test that the SHT class returns the expected co-latitude grid (Gauss-Legendre).
"""
lat_grid_ref = np.cos(self.lat_grid)
lat_grid = self.sht.get_colatitude_grid()
assert np.all(np.isclose(lat_grid_ref, lat_grid))
def test_spatial_to_spectral(self):
"""
Test transform from spatial to spectral representation by transforming
spherical harmonics and ensuring that the result contains only one
significant component.
"""
l = np.random.randint(1, self.l_max)
m_max = min(self.m_max, l)
m_max = l
m = np.random.randint(-m_max, m_max)
xx, yy = np.meshgrid(self.lat_grid, self.lon_grid, indexing="xy")
zz = sph_harm(m, l, yy, xx)
coeffs = self.sht.transform(zz.real)
assert np.sum(np.abs(coeffs) > 1e-6) == 1
def test_inverse_transform(self):
"""
Test transforming from spatial field to spectral field and back to ensure
that the input field is recovered.
"""
l = np.random.randint(1, self.l_max)
m_max = min(self.m_max, l)
m_max = l
m = np.random.randint(-m_max, m_max)
xx, yy = np.meshgrid(self.lon_grid, self.lat_grid, indexing="ij")
zz = sph_harm(m, l, xx, yy)
coeffs = self.sht.synthesize(self.sht.transform(zz.real).ravel())
assert np.all(np.isclose(zz.real, coeffs))
def test_inverse_transform_cmplx(self):
"""
Test transforming from spatial field to spectral field and back to ensure
that the input field is recovered.
"""
l = np.random.randint(1, self.l_max)
m_max = l #min(self.m_max, l)
m = np.random.randint(-m_max, m_max)
xx, yy = np.meshgrid(self.lon_grid, self.lat_grid, indexing="ij")
zz = sph_harm(m, l, xx, yy)
coeffs = self.sht.transform_cmplx(zz)
zz_rec = self.sht.synthesize_cmplx(2.0 * coeffs)
assert np.all(np.isclose(2.0 * zz, zz_rec))
def test_evaluate(self):
"""
Test that evaluating the spectral representation reproduces the spatial
input.
"""
l = np.random.randint(1, self.l_max)
m_max = min(self.m_max, l)
m_max = l
m = np.random.randint(-m_max, m_max)
xx, yy = np.meshgrid(self.lat_grid, self.lon_grid, indexing="xy")
zz_ref = sph_harm(m, l, yy, xx)
coeffs = self.sht.transform(zz_ref.real)
points = np.stack([yy.ravel(), xx.ravel()], axis=-1)
zz = self.sht.evaluate(coeffs, points)
assert np.all(np.isclose(zz, zz_ref.real.ravel()))