def main(args):
set_quiet(args.quiet)
if args.corpus:
unigrams = count_unigrams(args.corpus)
vocab, outfname = decide_vocab(unigrams, args.cutoff,
args.vocab, args.want)
if args.rewrite:
rewrite_corpus(args.corpus, vocab, outfname)
else:
extract_stat(args.corpus, vocab, outfname, args.window)
if args.stat:
assert(args.m is not None and args.kappa is not None)
if args.no_matlab:
C = canon()
C.set_params(args.m, args.kappa)
C.get_stat(args.stat)
C.start_logging()
C.approx_cca()
C.end_logging()
C.write_result()
else:
call_matlab(args.stat, args.m, args.kappa)
if args.clean: clean()
def main(args):
set_quiet(args.quiet)
if args.corpus:
unigrams = count_unigrams(args.corpus)
vocab, outfname = decide_vocab(unigrams, args.cutoff, args.vocab,
args.want)
if args.rewrite:
rewrite_corpus(args.corpus, vocab, outfname)
else:
extract_stat(args.corpus, vocab, outfname, args.window)
if args.stat:
assert (args.m is not None and args.kappa is not None)
if args.no_matlab:
C = canon()
C.set_params(args.m, args.kappa)
C.get_stat(args.stat)
C.start_logging()
C.approx_cca()
C.end_logging()
C.write_result()
else:
call_matlab(args.stat, args.m, args.kappa)
if args.clean: clean()
('dog', '<?><+1>'): 2,
('the', 'cat<-1>'): 1,
('<?>', 'cat<-1>'): 1,
('the', 'cat<+1>'): 2,
('cat', '<?><+1>'): 1
}
stat = check()
# Check if the result of python sparsesvd agrees with the result of Matlab.
m = 2
kappa = 1
C = canon()
C.set_params(m, kappa)
C.get_stat(stat)
C.start_logging()
C.approx_cca()
C.end_logging()
C.write_result()
outdirname = call_matlab(stat, m, kappa)
sv_matlab = map(lambda line: float(line.split()[0]),
open(os.path.join(outdirname, 'sv')).readlines())
for i in range(len(C.sv)):
assert (abs(C.sv[i] - sv_matlab[i]) < 1e-10)
sys.stderr.write('Correctness of statistics and svd calculations verified.\n')
sys.stderr.write('Cleaning.\n')
clean()
('the','<?><-1>'): 2,
('dog','<?><+1>'): 2,
('the','cat<-1>'): 1,
('<?>','cat<-1>'): 1,
('the','cat<+1>'): 2,
('cat','<?><+1>'): 1
}
stat = check()
# Check if the result of python sparsesvd agrees with the result of Matlab.
m = 2
kappa = 1
C = canon()
C.set_params(m, kappa)
C.get_stat(stat)
C.start_logging()
C.approx_cca()
C.end_logging()
C.write_result()
outdirname = call_matlab(stat, m, kappa)
sv_matlab = map(lambda line: float(line.split()[0]),
open(os.path.join(outdirname, 'sv')).readlines())
for i in range(len(C.sv)): assert(abs(C.sv[i] - sv_matlab[i]) < 1e-10)
sys.stderr.write('Correctness of statistics and svd calculations verified.\n')
sys.stderr.write('Cleaning.\n')
clean()