BUG: Fixed wavenumber definition and amplitude scaling in Eisenstein & Hu model

[jucordero]

Description

The Eisenstein & Hu approximation for the linear matter power spectrum now assumes the input wavenumber vector to be in units of Mpc instead of Mpc/h (See ADR-2)

The overall amplitude was also missing factors of h. This has now been fixed such that multiplying eisenstein_hu by the square of growth_function_carroll gives a power spectrum that closely matches CAMB and CLASS.

This addresses issue #442

from astropy.cosmology import Planck18
from matplotlib import pyplot as plt
import numpy as np
from skypy.power_spectrum import camb, eisenstein_hu, growth_function_carroll

k = np.logspace(-3, -1, 50)
z, A_s, n_s = 0, 2.2E-9, 0.97

p_camb = camb(k, z, Planck18, A_s, n_s)
p_eh = eisenstein_hu(k, A_s, n_s, Planck18)
d = growth_function_carroll(z, Planck18)

plt.plot(k, p_camb, label='CAMB')
plt.plot(k, p_eh * d * d, label='Eisenstein-Hu')
plt.loglog()
plt.legend()
plt.show()

Returns a power spectrum which has the correct wavenumbers and amplitude

Checklist

• Follow the Contributor Guidelines
• Write unit tests
• Write documentation strings
• Assign someone from your working team to review this pull request
• Assign someone from the infrastructure team to review this pull request

[rrjbca]

Some suggestions for improved clarity. Could you also confirm where the updated numerical values for test_eisenstein_hu came from? Did you re-calculate them using cosmosis for the correct wavenumbers?

Review does not request any changes

[rrjbca]

Looks great thanks @jucordero @Lucia-Fonseca @ntessore for working on this

[Lucia-Fonseca]

Looks good to me too!