def test_MCMC_edge_1():
n, n_obs = (5, 1000)
basis = edge_basis(n)
tree_prior = None
ct_prior = dists.CountPriors.Uniform()
np.random.seed(123)
for i in range(3):
g = GraphAndBasis(n)
g.SetRandom()
sampler = run_MCMC(g,
n,
n_obs,
tree_prior,
ct_prior,
basis=basis,
seed=123,
fixed_init=True)
print(sampler.res['LIK_'][-1])
median_graph = str_list_to_median_graph(n,
sampler.res['SAMPLES'][500:],
.5)
assert (np.sum(median_graph != g.GetAdjM()) < 3)
assert (only_propose_one_edge_difference(sampler))
assert (lookup_contains_all_params(sampler))
def __init__(self, n, Param, basis=None):
self._n = n
self._Param = Param
if basis is not None:
self._basis = basis
else:
self._basis = edge_basis(n)
def __init__(self, n, Param, alpha=.5, tree_prior=None, basis=None):
# alpha: probability of inclusion
self._n = n
self._Param = Param
self._alpha = alpha
if tree_prior:
self._tree_prior = tree_prior
elif basis is not None:
self._basis = basis
self._tree_prior = None
else:
self._basis = edge_basis(n)
self._tree_prior = None
def __init__(self, n, Param, prob_c, tree_prior=None, basis=None):
# prob_c: distribution of the number of bases i.e. "count"
self._n = n
self._Param = Param
self._prob_c = prob_c # for 0, ... number of basis
if tree_prior:
self._tree_prior = tree_prior
elif basis is not None:
self._basis = basis
self._tree_prior = None
else:
self._basis = edge_basis(n)
self._tree_prior = None
def __init__(self, n, dol=None, tree=None, basis=None):
super().__init__(n, dol)
self._tree = tree
# basis is n_edges x n_basis array
if tree:
cbc = cycle_basis_complete(tree)
self._basis = cbc[:, :(n - 1) * (n - 2) // 2]
self._basis_active = np.linalg.solve(cbc, GF2(
self.GetBinaryL()))[:(n - 1) * (n - 2) // 2]
elif basis is not None:
assert (basis.shape == (n * (n - 1) // 2, n * (n - 1) // 2))
self._basis = basis
self._basis_active = np.linalg.solve(self._basis,
GF2(self.GetBinaryL()))
else:
self._basis = edge_basis(n)
self._basis_active = np.linalg.solve(self._basis,
GF2(self.GetBinaryL()))
def test_MCMC_edge_0():
n, n_obs = (5, 1000)
basis = edge_basis(n)
tree_prior = None
ct_prior = dists.CountPriors.Uniform()
# Empty
g = GraphAndBasis(n)
sampler = run_MCMC(g,
n,
n_obs,
tree_prior,
ct_prior,
basis=basis,
seed=123)
print(sampler.res['LIK_'][-1])
median_graph = str_list_to_median_graph(n, sampler.res['SAMPLES'][500:])
assert ((median_graph == g.GetAdjM()).all())
assert (only_propose_one_edge_difference(sampler))
assert (lookup_contains_all_params(sampler))
# Circle
g = GraphAndBasis(n, circle_graph(n))
sampler = run_MCMC(g,
n,
n_obs,
tree_prior,
ct_prior,
basis=basis,
seed=123)
print(sampler.res['LIK_'][-1])
median_graph = str_list_to_median_graph(n, sampler.res['SAMPLES'][500:],
.75)
assert ((median_graph == g.GetAdjM()).all())
assert (only_propose_one_edge_difference(sampler))
assert (lookup_contains_all_params(sampler))
def cbmcmc(data,
it=1000,
basis='cycle',
treeprior='all',
r=1,
p=.5,
cob_freq=100,
outfile=None,
seed=123,
init=None):
data = np.loadtxt(data, delimiter=",")
_, n = data.shape
# Bases and Tree Prior
if basis == 'edge':
basis = edge_basis(n)
tree_prior = None
cob_freq = None
elif basis == 'cycle':
basis = None
if treeprior == 'all':
tree_prior = dists.TreePriors.Uniform(n)
elif treeprior == 'hub':
tree_prior = dists.TreePriors.Hubs(n)
elif treeprior == 'path':
tree_prior = dists.TreePriors.Paths(n)
else:
raise ValueError("Invalid tree prior")
else:
raise ValueError("Invalid basis")
# Count Prior
if r is None:
ct_prior = dists.CountPriors.TruncatedNB(data.shape[1], p)
else:
ct_prior = dists.CountPriors.TruncatedNB(r, p)
# Prior, Likelihood, and Proposal
prior = dists.Priors.BasisCount(n, GraphAndBasis, ct_prior, tree_prior,
basis)
lik = dists.Likelihoods.GW_LA(data, 3, np.eye(n), GraphAndBasis)
prop = dists.Proposals.BasisWalk(n, GraphAndBasis, tree_prior, cob_freq)
# Run MCMC
print("Starting MCMC...")
sampler = MCMC_Sampler(prior, prop, lik, data, outfile=outfile)
if seed:
np.random.seed(seed)
if init is None:
sampler.run(it)
else:
import pickle
with open(init, 'rb') as handle:
g = pickle.load(handle)
if g._tree is None and tree_prior is not None:
g._tree = tree_prior.Sample()
sampler.run(it, fixed_init=g)
# Saving Results
if outfile is not None:
sampler.save_object()
return sampler