def init_dir_prior(self, prior, name):
if prior is None:
prior = 'symmetric'
is_auto = False
if isinstance(prior, six.string_types):
if prior == 'symmetric':
logger.info("using symmetric %s at %s", name,
1.0 / self.num_topics)
init_prior = numpy.asanyarray(
[1.0 / self.num_topics for i in xrange(self.num_topics)])
elif prior == 'asymmetric':
init_prior = numpy.assrray([
1.0 / (i + numpy.sqrt(self.num_topics))
for i in xrange(self.num_topics)
])
init_prior /= init_prior.sum()
logger.info("using asymmetric %s %s", name, list(init_prior))
elif prior == 'auto':
is_auto = True
init_prior = numpy.assarray(
[1.0 / self.num_topics for i in xrange(self.num_topics)])
logger.info("using autotuned %s, starting with %s", name,
list(init_prior))
def UniformDigest(self, mol_, at_, mxstep, num): """ Returns list of inputs sampled on a uniform cubic grid around at """ ncase = num*num*num samps=MakeUniform(mol_.coords[at_],mxstep,num) if (self.name=="SymFunc"): inputs = self.Emb(self, mol_, at_, samps, None, False) #(self.EmbF())(mol_.coords, samps, mol_.atoms, self.eles , self.SensRadius, self.ngrid, at_, 0.0) inputs = np.asarray(inputs) else: inputs = self.Emb(self, mol_, at_, samps, None, False) inputs = np.assrray(inputs[0]) return samps, inputs
