Gaussain noise of 'rot' estimator

[catketchup]

Hi,

I was wondering if I would like to get the gaussian noise of 'rot' estimator, do I still need to use 'N_l_from_A_l_optimal' or just use the norm itself? I notice that for rot estimator, the norm equals the gaussian noise in https://arxiv.org/abs/0902.4466.

Thank you!

[msyriac]

You always need to use 'N_l_from_A_l_optimal' because the normalization convention might not be such that N_L=A_L, as is written in some papers. The function applies the appropriate conversion (quickly). It is only correct if your estimator is a minimum variance estimator. Otherwise, you have to use the slower 'N_l_from_A_l'

[catketchup]

I see. I notice, for 'rot' estimator, the gaussian noise given by 'N_l_from_A_l_optimal' is several order of magnitude larger than that in the Fig2 of https://arxiv.org/abs/0902.4466. I am confused where might go wrong.