Bandpass integration

The get_emission method implemented by any subclass of Model, including Sky, allows a numpy array of frequencies instead of a single frequency value to define a tophat bandpass and optionally a weights array of the same length to define a custom weight. PySM normalizes the weights array to unit integral then generates the emission at each of the specified frequencies, it multiplies it by the corresponding weight, and integrates them with the Trapezoidal rule. Instead of having generating all the maps as PySM 2, PySM 3 keeps in memory just 1 array for the output map and then creates the maps at a specific frequency one at a time and accomulates it properly into the output array.

sky = pysm.Sky(nside=128, preset_strings=["d1", "s1"])

map_100GHz_delta = sky.get_emission(100 * u.GHz)

bandpass_frequencies = np.linspace(95, 105, 11) * u.GHz

bandpass_frequencies

$[95,~96,~97,~98,~99,~100,~101,~102,~103,~104,~105] \; \mathrm{GHz}$

map_100GHz_tophat = sky.get_emission(bandpass_frequencies)

hp.mollview(map_100GHz_delta[0] - map_100GHz_tophat[0], min=-1, max=1, title="I map difference", unit=map_100GHz_tophat.unit)

bandpass_weights = np.array([2,3,5,9,11,11.5,11,9,5,3,2])

plt.plot(bandpass_frequencies,bandpass_weights);

map_100GHz_bandpass = sky.get_emission(bandpass_frequencies, bandpass_weights)

hp.mollview(map_100GHz_bandpass[0] - map_100GHz_tophat[0], min=-1, max=1, title="I map difference", unit=map_100GHz_tophat.unit)