While reviewing the get_Px_AA function in lya_theory.py, I found these lines of code [right at the start of the function](https://github.com/igmhub/cupix/blob/02c60c5cf842a3efaa77190b0e24b1bf0a90e6c6/cupix/likelihood/lya_theory.py#L551).
zs = np.atleast_1d(zs)
Nz = len(zs)
theta_deg = np.atleast_1d(theta_arcmin) / 60.0
# figure out emulator calls
emu_call, M_AA_of_z, M_tdeg_of_z = self.get_emulator_calls(
zs,
like_params=like_params,
return_M_of_z=True
)
As we discussed in person, one shouldn't need to have an emulator (or emulator calls) to make predictions for Px. This should only be needed when one is working with ForestFlow and varying IGM parameters, but not when doing Kaiser fits or varying Arinyo parameters.
For simplicity, I would have although thought that it would better to call this function one-z at a time, so that we don't make our life harder with arrays of parameter values, or with high-dimensional numpy arrays that need to be squeezed later.
While reviewing the get_Px_AA function in lya_theory.py, I found these lines of code
[right at the start of the function](https://github.com/igmhub/cupix/blob/02c60c5cf842a3efaa77190b0e24b1bf0a90e6c6/cupix/likelihood/lya_theory.py#L551).As we discussed in person, one shouldn't need to have an emulator (or emulator calls) to make predictions for Px. This should only be needed when one is working with ForestFlow and varying IGM parameters, but not when doing Kaiser fits or varying Arinyo parameters.
For simplicity, I would have although thought that it would better to call this function one-z at a time, so that we don't make our life harder with arrays of parameter values, or with high-dimensional numpy arrays that need to be squeezed later.