Skip to content

Fisher Forecast

See Fisher Forecast for what the forecast computes and the command-line interface.

cmb_lensing.fisher_forecast.forecast

forecast(nside, theta_pix, noise_level, is_sampled, param_ground, delensed_covariance, pol=DEFAULT_POL, qe_filter=DEFAULT_QE_FILTER, phi_noise=None, include_bb=True, l_knee=0, beam_fwhm=0, l_cutoff=DEFAULT_ELL_CUTOFF, camb_lmax=CAMB_LMAX)

Gaussian Fisher matrix for the sampled LCDM parameters on an nside x nside box.

Parameters:

Name Type Description Default
nside, theta_pix, noise_level

the box (pixels per side, arcmin, uK-arcmin in temperature; the polarization noise is sqrt(2) higher)

required
is_sampled

{param_name: bool}; False parameters are held FIXED

required
param_ground

{param_name: float} fiducial point, all five

required
delensed_covariance

a merged delensed_covariance.npz measured at this box, noise level and pol (required): the field blocks' delensed signal at theta_0 and theta_0 +/- h, and the steps h

required
pol

"I" (temperature), "P" (polarization) or "IP" (both): the data

DEFAULT_POL
qe_filter

"unlensed" (default) or "lensed": the spectra in the filter of the quadratic-estimate N0 used as the phi noise. Its response always uses the unlensed spectra. The estimator is load_sim's: TT for "I", EB for "P" and "IP"

DEFAULT_QE_FILTER
phi_noise

a merged map_joint_phi_noise.npz / sampler_phi_noise.npz whose nphi replaces the QE N0

None
include_bb

for "P" / "IP", whether the delensed BB block is counted

True
l_knee, beam_fwhm

1/f knee (0 = white, as every sampler entry point) and beam FWHM

required
camb_lmax

how far CAMB is run (default CAMB_LMAX, the simulations' support)

CAMB_LMAX

Returns:

Type Description

(fisher, names) - the n_sampled x n_sampled matrix and the sampled names, both in

constants.PARAM_ORDER.

Raises:

Type Description
ValueError

if delensed_covariance is None.

Source code in cmb_lensing/fisher_forecast.py
def forecast(nside, theta_pix, noise_level, is_sampled, param_ground, delensed_covariance,
             pol = DEFAULT_POL, qe_filter = DEFAULT_QE_FILTER, phi_noise = None,
             include_bb = True, l_knee = 0, beam_fwhm = 0, l_cutoff = DEFAULT_ELL_CUTOFF,
             camb_lmax = CAMB_LMAX):
    """Gaussian Fisher matrix for the sampled LCDM parameters on an nside x nside box.

    Args:
        nside, theta_pix, noise_level: the box (pixels per side, arcmin, uK-arcmin in
                      temperature; the polarization noise is sqrt(2) higher)
        is_sampled:   {param_name: bool}; False parameters are held FIXED
        param_ground: {param_name: float} fiducial point, all five
        delensed_covariance: a merged delensed_covariance.npz measured at this box, noise
                      level and `pol` (required): the field blocks' delensed signal at
                      theta_0 and theta_0 +/- h, and the steps h
        pol:          "I" (temperature), "P" (polarization) or "IP" (both): the data
        qe_filter:    "unlensed" (default) or "lensed": the spectra in the filter of the
                      quadratic-estimate N0 used as the phi noise. Its response always uses
                      the unlensed spectra. The estimator is load_sim's: TT for "I", EB for
                      "P" and "IP"
        phi_noise:    a merged map_joint_phi_noise.npz / sampler_phi_noise.npz whose `nphi`
                      replaces the QE N0
        include_bb:   for "P" / "IP", whether the delensed BB block is counted
        l_knee, beam_fwhm: 1/f knee (0 = white, as every sampler entry point) and beam FWHM
        camb_lmax:    how far CAMB is run (default CAMB_LMAX, the simulations' support)

    Returns:
        (fisher, names) - the n_sampled x n_sampled matrix and the sampled names, both in
        constants.PARAM_ORDER.

    Raises:
        ValueError: if delensed_covariance is None.
    """
    names, steps, blocks_fid, blocks_plus, blocks_minus = covariance_stencil(
        nside, theta_pix, noise_level, is_sampled, param_ground, delensed_covariance,
        pol = pol, qe_filter = qe_filter, phi_noise = phi_noise, include_bb = include_bb,
        l_knee = l_knee, beam_fwhm = beam_fwhm, l_cutoff = l_cutoff, camb_lmax = camb_lmax)
    return fisher_from_blocks(blocks_fid, blocks_plus, blocks_minus, steps, nside), names