def test_baseembed_er(self):
n_components = 4
embed = BaseEmbed(n_components=n_components)
n = 10
M = 20
A = er_nm(n, M) + 5
embed._reduce_dim(A)
self.assertEqual(embed.latent_left_.shape, (n, n_components))
self.assertTrue(embed.latent_right_ is None)
def test_algorithms(self):
embed = BaseEmbed(n_components=self.n, algorithm="full")
embed._reduce_dim(self.A)
self.assertEqual(embed.latent_left_.shape, (self.n, self.n))
self.assertEqual(embed.latent_right_.shape, (self.n, self.n))
# When algoritm != 'full', cannot decompose to all dimensions
embed = BaseEmbed(n_components=self.n, algorithm="truncated")
with self.assertRaises(ValueError):
embed._reduce_dim(self.A)
embed = BaseEmbed(n_components=self.n, algorithm="randomized")
with self.assertRaises(ValueError):
embed._reduce_dim(self.A)
def test_baseembed(self):
embed = BaseEmbed(n_components=None)
n = 10
M = 20
A = er_nm(n, M) + 5
embed._reduce_dim(A)