hispecdrp.primitives.nonlinear_correction.calc_nonlin_coefs_Brandt

Functions

calc_A_matrix(diffs, diffs_pow, Cov, sig, gain)

Calculate the matrix elements needed to compute a nonlinearity correction

get_nonlin_coefs(data_list, sig, gain, ...)

Use many ramps to fit for nonlinearity correction coefficients

Classes

Covar(read_times[, singlereads])

class Covar holding read and photon noise components of alpha and beta and the time intervals between the resultant midpoints

Nonlin_Result()

class Covar(read_times, singlereads=False)[source]

Bases: object

class Covar holding read and photon noise components of alpha and beta and the time intervals between the resultant midpoints

class Nonlin_Result[source]

Bases: object

calc_A_matrix(diffs, diffs_pow, Cov, sig, gain, countrateguess=None, diffs2use=None, rescale=True, dn_scale=10)[source]

Calculate the matrix elements needed to compute a nonlinearity correction

Parameters:
  • diffs (ndarray, shape (ndiffs, npix)) – Resultant differences.

  • diff_pow (ndarray, shape (order+1, ndiffs, npix)) – Templates for nonlinearity fit.

  • Cov (Covar) – Class holding the covariance matrix information.

  • sig (ndarray, shape (npix,)) – Read noise.

  • gain (ndarray, shape (npix,)) – Gain.

  • countrateguess (ndarray, shape (npix,), default None) – Count rates to be used to estimate the covariance matrix. If None, the average difference is used, replacing negative means with zeros.

  • diffs2use (ndarray of bool, shape (ndiffs, npix), default None) – Boolean mask of whether to use each resultant difference for each pixel.

  • rescale (bool, default True) – Scale the covariance matrix internally to avoid possible overflow/underflow problems for long ramps. Slightly increases computational cost.

  • dn_scale (int) – How often to keep tallies of the scale factor.

Returns:

A_all (ndarray, shape (order, order)) – order is the first dimension of diffs_pow

get_nonlin_coefs(data_list, sig, gain, order, intercepts, nsamples_est=5, d2u_list=None, satval=60000.0, use_legendre=True, calc_cond=False, last_pars=None, rescale_domain=True)[source]

Use many ramps to fit for nonlinearity correction coefficients

Parameters:
  • data_list (list of ndarrays) – list of 2D ndarrays for ramps, each of shape (nreads, npixels)

  • sig (ndarray) – 1D array for read noise, shape (npixels)

  • gain (ndarray) – 1D array for read noise, shape (npixels)

  • order (int) – order of polynomial to fit for the nonlinearity correction

  • intercepts (ndarray) – bias levels to anchor nonlinearity fit, shape (npixels)

  • nsamples_est (int, default 5) – number of initial samples whose median provides an estimate of the count rate for the total slope target and for getting a baseline to reject egregious jumps.

  • d2u_list (list of ndarrays or None, default None) – list of uint8 ndarrays, same length/shapes as data_list, =1 for the read differences that we should use. If None, use all reads. Arrays will be further flagged in this routine for saturation and for egregious jumps

  • satval (float or ndarray, default 6e4) – saturation value. If an ndarray, should be of shape (npixels)

  • use_legendre (boolean, default True) – fit in the Legendre polynomial basis?

  • calc_cond (boolean, default False) – compute the condition number of the matrix for each pixel?

  • last_pars (Nonlin_Result or None, default None) –

    output of a previous run of get_nonlin_coefs to set the

    covariance matrix from the previously estimated count rates

  • rescale_domain (boolean, default True) – rescale the domain to [-1, 1] for better numerical behavior?

Returns:

Nonlin_Result – result structure with lots of information