adding the use of psd from preprocessing to use as mapmaking weights #1224
Conversation
… Also a repeated NmatWhite class is removed
… Also a repeated NmatWhite class is removed
…obs/sotodlib into chervias/weights_with_psd
|
Ok, this is ready for review now |
sotodlib/mapmaking/demod_mapmaker.py
Outdated
| @@ -64,7 +64,7 @@ def __init__(self, signals=[], noise_model=None, dtype=np.float32, verbose=False | |||
| self.ncomp = len(comps) | |||
| self.singlestream = singlestream | |||
|
|
|||
| def add_obs(self, id, obs, noise_model=None, split_labels=None): | |||
| def add_obs(self, id, obs, noise_model=None, split_labels=None, use_psd=True, wn_label = 'preprocess.noiseQ_mapmaking.white_noise'): | |||
There was a problem hiding this comment.
Wrap that new arg onto next line. Remove spaces around "=".
I see this arg, as well as use_psd, aren't included in the docstring. These args are documented elsewhere ... so either copy paste that here, or else reduce duplication and just refer the reader somehow ... "the use_psd and wn_label args are passed through to add_obs; see that docstring".
sotodlib/mapmaking/noise_model.py
Outdated
| def check_ready(self): | ||
| return self.ivar is not None |
There was a problem hiding this comment.
This is at odds with the (undocumented) point of base class Nmat.check_ready, which is designed to raise and exception if the ready condition is not met. So then calling self.check_ready(), in self.write() for example, has no consequence.
I think what you're trying to do might better be achieved by removing this redefinition of check_ready, and instead making ready a computed attribute:
@property
def ready(self):
return self.ivar is not None
There was a problem hiding this comment.
computed atribute means re-calculated every time I ask for it? That's what I need. If I ask the first time, it should be False since ivar is None, the second time I ask it should be True since ivar will be != None
There was a problem hiding this comment.
ok I think I have answered my own question
This is not ready yet, I haven't tested either on real data, but I guess people can look at the code in the meantime.