Random Variates

When we perform a fit or load and analysis result, the parmeters of our model become distributions in the AnalysisResults object. These are actaully instantiactions of the RandomVaraiates class.

While we have covered most of the functionality of RandomVariates in the AnalysisResults section, we want to highlight a few of the details here.

[2]:
%%capture
import matplotlib.pyplot as plt
from threeML import *

Let’s load back our fit of the line + gaussian from the AnalysisResults section.

[4]:
ar = load_analysis_results("test_mle.fits")

When we display our fit, we can see the parameter paths of the model. What if we want specific information on a parameter(s)?

[5]:
ar.display()
Best fit values:

result unit
parameter
fake.spectrum.main.composite.a_1 1.73 +/- 0.11 1 / (cm2 keV s)
fake.spectrum.main.composite.b_1 (4 +/- 4) x 10^-3 1 / (cm2 keV2 s)
fake.spectrum.main.composite.F_2 (2.9 +/- 0.4) x 10 1 / (cm2 s)
fake.spectrum.main.composite.mu_2 (2.493 +/- 0.013) x 10 keV
fake.spectrum.main.composite.sigma_2 1.02 +/- 0.09 keV

Correlation matrix:

1.00-0.85-0.040.04-0.08
-0.851.00-0.00-0.020.01
-0.04-0.001.000.13-0.16
0.04-0.020.131.00-0.30
-0.080.01-0.16-0.301.00

Values of -log(likelihood) at the minimum:

-log(likelihood)
sim_data 20.581877
total 20.581877

Values of statistical measures:

statistical measures
AIC 52.527390
BIC 60.723868

Let’s take a look at the normalization of the gaussian. To access the parameter, we take the parameter path, and we want to get the variates:

[6]:
norm = ar.get_variates("fake.spectrum.main.composite.F_2")

Now, norm is a RandomVariate.

[7]:
type(norm)
[7]:
threeML.random_variates.RandomVariates

This is essentially a wrapper around numpy NDArray with a few added properties. It is an array of samples. In the MLE case, they are samples from the covariance matrix (this is not at all a marginal distribution, but the parameter “knows” about the entire fit, i.e., it is not a profile) and in the Bayesian case, these are samples from the posterior (this is a marginal).

The output representation for an RV are its 68% equal-tail and HPD uncertainties.

[8]:
norm
[8]:
equal-tail: (2.9 +/- 0.4) x 10, hpd: (2.9 +/- 0.4) x 10

We can access these directly, and to any desired confidence level.

[9]:
norm.equal_tail_interval(cl=0.95)
[9]:
(20.866631191526672, 36.81102188820774)
[10]:
norm.highest_posterior_density_interval(cl=0.5)
[10]:
(26.640851845630955, 31.99016791002738)

As stated above, the RV is made from samples. We can histogram them to show this explicitly.

[11]:
fig, ax = plt.subplots()

ax.hist(norm.samples, bins=50, ec="k", fc="w", lw=1.2)
ax.set_xlabel("norm")
[11]:
Text(0.5, 0, 'norm')
../_images/notebooks_random_variates_19_1.png

We can easily transform the RV through propagation.

[12]:
log_norm = np.log10(norm)
log_norm
[12]:
equal-tail: 1.46 +/- 0.06, hpd: 1.46 -0.05 +0.06
[13]:
fig, ax = plt.subplots()

ax.hist(log_norm.samples, bins=50, ec="k", fc="w", lw=1.2)
ax.set_xlabel("log norm")
[13]:
Text(0.5, 0, 'log norm')
../_images/notebooks_random_variates_22_1.png
.. note:: Some operations will destroy the RV by accessing only its NDArray substructure. For example, using an RV with astropy units will return an array of samples with the given units.