Advanced Parameter Study

In this example, we’ll follow the same procedure as in the Simple Parameter Study example, but add a few dimensions and take advantage of some advanced features. It’ll be advantageous to write the following as a script (i.e., not in an interactive Python session) so that we can execute it in parallel. Call it ares_2d_grid.py:

import ares
import numpy as np

As in Simple Parameter Study, we’ll save the redshift, 21-cm brightness temperature, and spin temperature at the redshifts corresponding to extrema in the global signal (which we refer to as turning points B, C, and D):

fields = ['z', 'dTb', 'Ts']
redshifts = ['B', 'C', 'D']

and now, initialize a ModelGrid instance:

base_kwargs = \
{
 'inline_analysis': [fields, redshifts],
}

mg = ares.inference.ModelGrid(**base_kwargs)

Let’s again survey a 2-D swath of parameter space, varying the X-ray normalization parameter and now Tmin, the minimum virial temperature of star-forming halos:

mg.set_axes(fX=np.linspace(0.1, 0.5, 3), Tmin=np.logspace(3, 4, 3))

This is a case where load-balancing is very helpful. The Tmin dimension of this parameter space requires some significant overhead at the outset of each calculation - but once this has been done once, we can simply store that information and use it again and again. Because of this, load-balancing over the Tmin dimension of a model grid makes lots of sense:

mg.LoadBalance(method=1)

Finally, to run the thing:

mg.run('advanced_param_study')

To run this as a script, back in the terminal invoke the script with mpirun (here, with 4 cores)

mpirun -np 4 python ares_2d_grid.py

All the usual analysis routines still apply.