def benchmark_ssvd(ctx, timer): expr.set_random_seed() DIM = (1280, 1280) #A = expr.randn(*DIM, dtype=np.float64) A = np.random.randn(*DIM) A = expr.from_numpy(A) t1 = datetime.now() U,S,VT = svd(A) t2 = datetime.now() cost_time = millis(t1, t2) print "total cost time:%s ms" % (cost_time)
def benchmark_pca(ctx, timer): expr.set_random_seed() DIM = (1280, 512) data = np.random.randn(*DIM) A = expr.from_numpy(data) #A = expr.randn(*DIM, dtype=np.float64) t1 = datetime.now() m = PCA(N_COMPONENTS) m.fit(A) t2 = datetime.now() cost_time = millis(t1, t2) print "total cost time:%s ms" % (cost_time)
def test_pca(self):
expr.set_random_seed()
FLAGS.opt_parakeet_gen = 0
data = np.random.randn(*DIM)
A = expr.from_numpy(data, tile_hint=util.calc_tile_hint(DIM, axis=0))
m = PCA(N_COMPONENTS)
m2 = SK_PCA(N_COMPONENTS)
m.fit(A)
m2.fit(data)
print m2.components_ - m.components_
assert np.allclose(absolute(m.components_), absolute(m2.components_))
def test_ssvd(self):
expr.set_random_seed()
ctx = blob_ctx.get()
# Create a sparse matrix.
A = expr.randn(*DIM, tile_hint = (int(DIM[0]/ctx.num_workers), DIM[1]),
dtype=np.float64)
U,S,VT = svd(A)
U2,S2,VT2 = linalg.svd(A.glom(), full_matrices=0)
assert np.allclose(absolute(U.glom()), absolute(U2))
assert np.allclose(absolute(S), absolute(S2))
assert np.allclose(absolute(VT), absolute(VT2))