def __init__(self, input_file):
# Load info
self.info = info(input_file)
# Setup likelihood
self.likelihood = likelihood_dict(self.info["likelihoods"])
# DEBUG
from plots import plot_likelihood_pdf
plot_likelihood_pdf(self.likelihood, self.info["params"])
# Setup samples_set and sampler
self.samples = samples_set(self.info["params"])
self.sampler = initialise_sampler(self.likelihood.loglik,
lambda x: 1,
self.samples, self.info["params"],
self.info["sampler"],
self.show_updated_info)
# Room for parallelisation
# n_processes = 1
# n_times_pool = 2
# Fire sampler
status = self.sampler.run()
self.final_report()
return
def show_updated_info(self):
import matplotlib.pyplot as plt
fig = plt.figure()
ax = fig.add_subplot(111)
x = self.samples.params[:self.samples.n_samples][:,0]
import numpy as np
t = np.exp(self.samples.logliks[:self.samples.n_samples])
lims = [(p["prior"]["min"], p["prior"]["max"]) for p in self.info["params"]]
plt.xlim()
plt.plot(x, t, "o")
from plots import plot_likelihood_pdf
plot_likelihood_pdf(self.likelihood, self.info["params"], ax=ax, show=False)
# grf prediction
xpred = np.linspace(sorted(x)[0], sorted(x)[-1], 400)
xpred = np.linspace(lims[0][0], lims[0][1], 400)
# tpred = self.sampler.posterior_expected(xpred)
# spred = np.sqrt(self.sampler.posterior_covariance(xpred))
# plt.plot(xpred, tpred, "-", color="red")
# plt.plot(xpred, tpred+2*spred, ":", color="red")
# plt.plot(xpred, tpred-2*spred, ":", color="red")
plt.show()
