def from_outputs(cls, options, **kw):
column_names, chains, metadata, comments, final_metadata = output_module.input_from_options(
options)
chains = np.vstack(chains).T
chains = dict(zip(column_names, chains))
chain_data = {"Chains": chains}
return cls(chain_data, **kw)
def from_outputs(cls, options, **kw):
burn = kw.pop("burn", 0)
thin = kw.pop("thin", 1)
column_names, chains, metadata, comments, final_metadata = output_module.input_from_options(
options)
if burn == 0:
pass
elif burn < 1:
for i, chain in enumerate(chains):
print("Burning fraction %f of chain %d, which is %d samples" %
(burn, i, int(burn * len(chain[:, 0]))))
chains = [
chain[int(burn * len(chain[:, 0])):, :] for chain in chains
]
else:
burn = int(burn)
chains = [chain[burn:, :] for chain in chains]
#In this case all the chains are assumed to be from a single
#run. So we should concatenate them all for a single
chains = np.vstack(chains).T
chains = dict(list(zip(column_names, chains)))
chain_data = {"Chains": chains}
return cls(chain_data, **kw)
def load_ini(self, inputs):
output_options = dict(inputs.items('output'))
filename = output_options['filename']
self.name = filename
sampler = inputs.get("runtime", "sampler")
for key,val in inputs.items(sampler):
self.sampler_options[key]=str(val)
self.colnames, self.data, self.metadata, self.comments, self.final_metadata = \
output_module.input_from_options(output_options)