def test_first_phase():
from glb import basepath
import os
k = 4
ipt = os.path.join(basepath, 'data/z_5/synfix_5.t01.edges')
ipt_gc = os.path.join(basepath, 'data/z_5_r/synfix_5.t01.gc')
ipt_comm = os.path.join(basepath, 'data/z_5/synfix_5.t01.comm1')
opt = os.path.join(basepath, 'data/z_5_r/s5t01.graph.json')
opt_hubs = os.path.join(basepath, 'data/z_5_r/s5t01.hubs.csv')
gc.gen_gc(ipt, ipt_gc)
g = Graph()
sm.create_graph(ipt_gc, g)
sm.load_cluster_info(g, ipt_comm)
U, H, L, X = core.cdnmf(g.adjmatrix(), k)
ce.getcluster_bycore(g, k, U, H)
ce.getcluster_rest(g, k)
ce.get_hubs(g, k, U, opt_hubs)
nmi = sm.compute_nmi(g)
print '===NMI===\n', nmi
g.dump(opt)
a = g.adjmatrix()
opt = 'data/z_5_r/test.adj'
writer = open(opt, 'w')
for i in range(128 * 128):
writer.write(str(a.item(i)))
if (i + 1) % 128 == 0:
writer.write('\n')
else:
writer.write(',')
def run(lmd):
num_iter = 1
from glb import basepath
iptdir = os.path.join(basepath, 'data/blog')
optdir = os.path.join(basepath, 'data/result/blog')
if not os.path.isdir(optdir):
os.makedirs(optdir)
list_nmi = []
first = True
for i in range(1, 15):
fmt = 'blog_timeSlice_%02d' % i
log.info(fmt + '...')
ipt_edges = os.path.join(iptdir, fmt + '.edges')
path_gc = os.path.join(optdir, fmt+'.gc')
ipt_comm1 = os.path.join(iptdir, fmt+'.comm1')
path_hubs = os.path.join(optdir, fmt+'.hubs.csv')
#gc.gen_gc(ipt_edges, path_gc)
g = Graph()
sm.create_graph(path_gc, g)
idx_clst = sm.load_cluster_info(g, ipt_comm1)
g.dump(os.path.join(optdir,'graph.json'))
k = idx_clst.num_indices()
if first:
lpre = 2.0 ** 32
for i in range(num_iter):
UU, HH, LL, XX = core.cdnmf(g.adjmatrix(), k)
if LL[-1] < lpre:
lpre = LL[-1]
U, H, X = UU, HH, XX
log.info('%02d -- %.4f %.4f CHANGED!' % (i+1, LL[-1], lpre))
else:
log.info('%02d -- %.4f %.4f' % (i+1, LL[-1], lpre))
first = False
else:
X = core.adjust_xpre(gpre, g, X)
lpre = 2.0 ** 32
for i in range(num_iter):
UU, HH, LL, XX = core.cdnmf_dynamic(g.adjmatrix(), k, X, lmd)
if LL[-1] < lpre:
lpre = LL[-1]
U, H, X = UU, HH, XX
log.info('%02d -- %.4f %.4f CHANGED!' % (i+1, LL[-1], lpre))
else:
log.info('%02d -- %.4f %.4f' % (i+1, LL[-1], lpre))
gpre = cp.deepcopy(g)
ce.getcluster_bycore(g, k, U, H)
ce.getcluster_rest(g, k)
ce.get_hubs(g, k, U, path_hubs)
nmi = sm.compute_nmi(g)
list_nmi.append(nmi)
print '===NMI===\n',nmi
print '===RESULT==='
for e in enumerate(list_nmi):
print '{}: {}'.format(e[0] + 1, e[1])
def run(lmd, which, times):
num_iter = 1
iptdir = os.path.join(basepath, 'data/bench/' + which)
optdir = os.path.join(basepath, bd + '/' + which)
if not os.path.isdir(optdir):
os.makedirs(optdir)
list_nmi = []
list_err = []
first = True
list_errmat_expected = []
list_errmat_actual = []
for i in range(1, 11):
fmt = '%s.t%02d' % (which, i)
log.info(fmt + '...')
ipt_edges = os.path.join(iptdir, fmt + '.edges')
path_gc = os.path.join(optdir, fmt + '.gc')
ipt_comm1 = os.path.join(iptdir, fmt + '.comm1')
path_hubs = os.path.join(optdir, fmt + '.hubs.csv')
#gc.gen_gc(ipt_edges, path_gc)
g = Graph()
#sm.create_graph(path_gc, g)
sm.create_graph(ipt_edges, g)
idx_clst = sm.load_cluster_info(g, ipt_comm1)
k = idx_clst.num_indices()
log.info('num clusters: %d' % k)
if first:
lpre = 2.0**32
for i in range(num_iter):
UU, HH, LL, XX = core.cdnmf(g.adjmatrix(), k)
if LL[-1] < lpre:
lpre = LL[-1]
U, H, X = UU, HH, XX
log.info('%02d -- %.4f %.4f CHANGED!' %
(i + 1, LL[-1], lpre))
else:
log.info('%02d -- %.4f %.4f' % (i + 1, LL[-1], lpre))
first = False
else:
X = core.adjust_xpre(gpre, g, X)
lpre = 2.0**32
for i in range(num_iter):
UU, HH, LL, XX = core.cdnmf_dynamic(g.adjmatrix(), k, X, lmd)
if LL[-1] < lpre:
lpre = LL[-1]
U, H, X = UU, HH, XX
log.info('%02d -- %.4f %.4f CHANGED!' %
(i + 1, LL[-1], lpre))
else:
log.info('%02d -- %.4f %.4f' % (i + 1, LL[-1], lpre))
gpre = cp.deepcopy(g)
'''
ce.getcluster_bycore(g, k, U, H)
ce.getcluster_rest(g, k)
#ce.get_hubs(g, k, U, path_hubs)
'''
getcluster_byx(g, X)
nmi = sm.compute_nmi(g)
list_nmi.append(nmi)
e, a = sm.get_errmat(g, k)
list_errmat_expected.append(e)
list_errmat_actual.append(a)
list_err.append(sm.compute_error(g, k))
print '===NMI===\n', len(list_nmi), nmi
d = {}
d['errmat_expected'] = sm.cell(list_errmat_expected)
d['errmat_actual'] = sm.cell(list_errmat_actual)
sio.savemat(os.path.join(optdir, 'lmd{}_times{}.mat'.format(lmd, times)),
d)
print '===RESULT==='
for e in enumerate(list_nmi):
print '{}: {}'.format(e[0] + 1, e[1])
return list_nmi, list_err
def run(filename, lmd, times):
num_iter = 1
if filename != 'syn_T_10_z_4_nC_3_bS_128_aD_16.mat':
return None, None
ipt = os.path.join(basepath, 'data/syn/' + filename)
optdir = os.path.join(basepath, 'data/result/syn/{}'.format(filename[:-4]))
if not os.path.isdir(optdir):
os.makedirs(optdir)
list_nmi = []
list_err = []
first = True
list_errmat_expected = []
list_errmat_actual = []
As, list_clsts = load_mat(ipt)
for idx in range(10):
log.info('%d ...' % idx)
A = As[idx]
list_clst = list_clsts[idx]
g = Graph()
idx_clst = sm.load_graph(g, A, list_clst)
k = idx_clst.num_indices()
log.info('num clusters: %d' % k)
if first:
lpre = 2.0**32
for i in range(num_iter):
UU, HH, LL, XX = core.cdnmf(g.adjmatrix(), k)
if LL[-1] < lpre:
lpre = LL[-1]
U, H, X = UU, HH, XX
log.info('%02d -- %.4f %.4f CHANGED!' %
(i + 1, LL[-1], lpre))
else:
log.info('%02d -- %.4f %.4f' % (i + 1, LL[-1], lpre))
first = False
else:
X = core.adjust_xpre(gpre, g, X)
lpre = 2.0**32
for i in range(num_iter):
UU, HH, LL, XX = core.cdnmf_dynamic(g.adjmatrix(), k, X, lmd)
if LL[-1] < lpre:
lpre = LL[-1]
U, H, X = UU, HH, XX
log.info('%02d -- %.4f %.4f CHANGED!' %
(i + 1, LL[-1], lpre))
else:
log.info('%02d -- %.4f %.4f' % (i + 1, LL[-1], lpre))
gpre = cp.deepcopy(g)
'''
ce.getcluster_bycore(g, k, U, H)
ce.getcluster_rest(g, k)
#ce.get_hubs(g, k, U, path_hubs)
'''
getcluster_byx(g, X)
nmi = sm.compute_nmi(g)
list_nmi.append(nmi)
e, a = sm.get_errmat(g, k)
list_errmat_expected.append(e)
list_errmat_actual.append(a)
list_err.append(sm.compute_error(g, k))
print '===NMI===\n', len(list_nmi), nmi
d = {}
d['errmat_expected'] = sm.cell(list_errmat_expected)
d['errmat_actual'] = sm.cell(list_errmat_actual)
sio.savemat(os.path.join(optdir, 'lmd{}_times{}.mat'.format(lmd, times)),
d)
print '===RESULT==='
for e in enumerate(list_nmi):
print '{}: {}'.format(e[0] + 1, e[1])
return list_nmi, list_err