A_L matches expectation at low L, but rises at higher L

[ajvanengelen]

• symlens version: most recent, I think
• Python version: 3.8
• Operating System: linux

Description

I'm computing some new A_L curves and trying to match expectations --- found that the result rose faster than expected. I wrote this test case script that just sets $f=1$ and $F=1$. At low ell I get what I expect according to the definition, based on the inverse of the area of the annulus from l_min to l_max. But the result rises with L.

What I Did

from __future__ import print_function
from orphics import maps,io,cosmology,stats
from pixell import enmap
import numpy as np
from symlens import qe
import matplotlib.pyplot as plt

shape,wcs = maps.rect_geometry(width_deg=20.,px_res_arcmin=1.5,proj='car')
theory = cosmology.default_theory()
modlmap = enmap.modlmap(shape,wcs)

#Set both f and F to just be 1
f_ones = 1  #enmap.ones(modlmap.shape, wcs)
F_ones = 1  #enmap.ones(modlmap.shape, wcs)

lmin, lmax = 100, 3500
Lmax = 3000
xmask = maps.mask_kspace(shape,wcs,lmin=lmin,lmax=lmax)
ymask = maps.mask_kspace(shape,wcs,lmin=lmin,lmax=lmax)

feed_dict = {}

#get A_L
A_L_ones = qe.A_l_custom(shape, wcs, feed_dict, f_ones, F_ones, xmask=xmask, ymask=ymask)

#We expect A_L to be L**2 times the inverse of ((1/(2pi)**2) times the area of a disk of outer radius lmax and inner radius lmin))
expected_A_L = modlmap**2 * 1/( 1 / (2 * np.pi)**2 * (np.pi * (lmax**2 - lmin**2)))

#set up binninga
bin_edges = np.arange(20,Lmax,20)
ells = np.arange(20,Lmax,1)
binner = stats.bin2D(modlmap,bin_edges)

cents, A_L_ones_1d = binner.bin(A_L_ones)
cents, expected_A_L_1d = binner.bin(expected_A_L)

plt.figure()
plt.plot(cents, A_L_ones_1d / expected_A_L_1d, label = 'binned')
indmax = 100
plt.plot(modlmap[0, :indmax], (A_L_ones / expected_A_L)[0, :indmax], label = 'slice on x axis')
plt.legend()
plt.ylabel('obtained / expected')
plt.xlabel('L')


plt.savefig('../output/test_ones.png')

[ajvanengelen]

Update: I just tried setting both of f_ones and F_ones to sympy.Integer(1), leading to

type(f_ones)
Out[7]: sympy.core.numbers.One

This doesn't change the results. Any ideas about what might be going on here?