Remove spectral 3

Author: jonatanvm

src/main.py

@@ -96,5 +96,5 @@ def run_all(files, algorithm, e_mode, n, n_jobs, out=True):
 
     else:
         # run_all(test_files, ALGORITHM_3, 'eigsh', n=100, n_jobs=10)
-        run_all(ptest_files, ALGORITHM_1, 'eigsh', n=1, n_jobs=10)
+        run_all(comp_files, ALGORITHM_4, 'eigsh', n=1, n_jobs=10)
         # run_all(comp_files, ALGORITHM_1, 'eigsh', n=10, n_jobs=10)

src/spectral.py

@@ -2,19 +2,15 @@
 
 import numpy as np
 from scipy.sparse.linalg import eigsh
-from sklearn.cluster import KMeans
 
 from kmeans import k_means
-from laplacian import unnormalized_laplacian, normalized_laplacian
+from laplacian import unnormalized_laplacian
 from read_graph import read_graph
 
 ALGORITHM_1 = 1
 ALGORITHM_2 = 2
 ALGORITHM_3 = 3
 ALGORITHM_4 = 4
-ALGORITHM_5 = 5
-ALGORITHM_6 = 6
-ALGORITHM_7 = 7
 
 
 def spectral_clustering1(graph_src, k_user=None):
@@ -101,40 +97,3 @@ def spectral_clustering2(graph_src, k_user=None):
     print("Cluster sizes: %s" % cluster_sizes)
     return clusters, seed, header
 
-
-def spectral_clustering3(graph_src, k_user=None):
-    print("Reading graph: " + graph_src)
-    start = time()
-    A, D, k, header = read_graph(graph_src)
-    if k_user or k is None:
-        k = k_user
-        header[4] = str(k_user)
-    print("Finished after %.2f seconds" % (time() - start))
-
-    # Calculate laplacian matrix
-    print("Calculating laplacian")
-    start = time()
-    laplacian_matrix = normalized_laplacian(D, A)
-    print("Finished after %.2f seconds" % (time() - start))
-
-    D = None  # Free memory
-    A = None  # Free memory
-
-    # Eigen-decomposition of Laplacian matrix
-    print("Calculating Eigen-decomposition")
-    start = time()
-    e_values, e_vectors = eigsh(laplacian_matrix, k=k, which='SA')
-    print("Finished after %.2f seconds" % (time() - start))
-    laplacian_matrix = None  # Free memory
-    U = np.real(e_vectors)
-    print("Normalizing U")
-    start = time()
-    T = U[:, :k] / np.sum(U[:, :k], axis=0)  # Normalize
-    print("Finished after %.2f seconds" % (time() - start))
-    kmeans = KMeans(n_clusters=k, random_state=0).fit(T[:, :k])
-    cluster_sizes = [0] * k
-    for i in kmeans.labels_:
-        cluster_sizes[i] += 1
-    print("Cluster sizes: %s" % cluster_sizes)
-    print(kmeans.cluster_centers_)
-    return kmeans.labels_