Created
July 10, 2018 12:18
-
-
Save ipashchenko/3db8d45d4bc26c3fdb3f27d001697071 to your computer and use it in GitHub Desktop.
Fitting beta_app and other stuff in pymc3
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import pymc3 as pm | |
import numpy as np | |
betas, betas_err, thetas = np.loadtxt("/home/ilya/github/lags_utils.py/data.txt", | |
unpack=True) | |
thetas /= 180/np.pi | |
# Use beta_app and theta from Hovatta | |
with pm.Model() as model: | |
G = pm.Uniform("G", 1, 100) | |
beta = pm.Deterministic("beta", T.sqrt(G**2-1)/G) | |
beta_ = pm.Deterministic("beta_", beta*T.sin(thetas)/(1-beta*T.cos(thetas))) | |
likelihood = pm.Normal("beta_obs", mu=beta_, sd=betas_err, observed=betas) | |
trace = pm.sample(njobs=4, tune=1000) | |
ax1 = pm.traceplot(trace, varnames=[G]) | |
ax2 = pm.plot_posterior(trace, varnames=[G]) | |
# Use only beta_app | |
with pm.Model() as model: | |
G = pm.Uniform("G", 1, 100) | |
beta = pm.Deterministic("beta", T.sqrt(G**2-1)/G) | |
thetas = pm.Uniform("thetas", 0, np.pi/3, shape=len(betas)) | |
beta_ = pm.Deterministic("beta_", beta*T.sin(thetas)/(1-beta*T.cos(thetas))) | |
likelihood = pm.Normal("beta_obs", mu=beta_, sd=betas_err, observed=betas) | |
trace = pm.sample(njobs=4, tune=1000) | |
ax1 = pm.traceplot(trace, varnames=["G", "thetas"]) | |
ax2 = pm.plot_posterior(trace, varnames=["G", "thetas"]) | |
samples = np.hstack((trace.get_values("G").reshape(2000, 1), | |
trace.get_values("thetas"))) | |
fig_corner = corner.corner(samples[::2], labels=['G']+["theta_{}".format(i) for | |
i in range(len(betas_err))], | |
show_titles=True, | |
quantiles=[0.16, 0.50, 0.84]) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment