search

cosmodesi / pypower

power spectrum and window function estimation
BSD 3-Clause "New" or "Revised" License
17 stars 6 forks source link

document how default ells are chosen for window calculation #10

Closed pmcdonal closed 2 years ago

pmcdonal commented 2 years ago

This is almost a "just send Arnaud an email asking" (or try to look up in papers) thing, instead of an "issue," but probably others would benefit if it was written in the documentation. Off the top of my head, I don't know how the ells needed for the window are chosen. I.e., the meaning of "the list of projections is set to be able to compute window convolution of theory power spectrum multipoles of orders power_ref.ells".

adematti commented 2 years ago

I have added some explanations in 8e0f40f460a7370b92f0936782768df00298bc2c, see: https://github.com/cosmodesi/pypower/blob/f6f5e2636b26902cd87e23c72c2dd354d6e843d3/pypower/smooth_window.py#L521

pmcdonal commented 2 years ago

Thanks! I might write "maximum order yielded by the sum of any theory and output multipole" and/or refer to Eq. ~C.8 in my paper with Florian, if no better place. I know what you mean by "product"... the product in that equation... but to the user it may be puzzling when it is the sum of the ells. (I wonder if this limit exact, or a result of approximations in these equations?)

adematti commented 2 years ago

I changed this. What do you mean by "this limit is exact"? (Sometimes people truncate the window function multipoles after e.g. ell = 4, but I find no good reason no to use all window function multipoles requested by theory, typically ell = 8 or 9).

pmcdonal commented 2 years ago

I had in my head something like "does l=10 really exactly not matter for the 4,4 window"... but I realize since you ask that this doesn't necessarily make much sense to ask, because the object carrying this l doesn't even appear in the more exact calculation.