Closed jbkalmbach closed 7 years ago
@drphilmarshall @rbiswas4 What do you think about merging this so that we have all the updates we added during hack week in the master branch and continuing on with new pull requests from here?
Good idea. I'll take a look.
All those sound good. I'll make the changes.
@rbiswas4 These are ready for a look. Thanks.
@drphilmarshall @rbiswas4
Ok. I think I've addressed most of the requests from the last update. The only thing is that instead of splitting up the error model notebook as you proposed Rahul, I explicitly mention in the README that the depth_curve notebook is a good intro to the Monitor and that the simple_error_model is a sample of the analysis we are doing with it. If I really missed anything else though let me know. Thanks.
The only thing is that instead of splitting up the error model notebook as you proposed Rahul, I explicitly mention in the README that the depth_curve notebook is a good intro to the Monitor and that the simple_error_model is a sample of the analysis we are doing with it.
Sounds good enough! Let us go forward.
Looks good, Bryce :-) However, I wrote previously:
Last thing: you talk about comparisons between DM and OpSim values - but I guess its really PhoSim+DM and OpSim that are being compared. Can you add some discussion of this point please? We don't want the reader going away thinking that there are errors in the DM measurements when there may not be... What do you think?
I still think it's best to address this, even in one sentence, in the error model example notebook - and it certainly should be in the DESC Note on the Twinkles 1 error model. I think we want a caveat like the following in the conclusions:
"There are two possible sources of discrepancy, either 1) the PhoSim realization of the OpSim observation, or 2) the DM processing of the PhoSim realization. If we simply want to predict DM measurements given OpSim inputs, the error model shown here (and its descendants) will enable that. However, when looking to develop an accurate inference scheme for real data, we will need to isolate step 2) by comparing with PhoSim 'inputs'. This kind of investigation will also be enabled by the Monitor functionality demonstrated here."
What do you think?
Ah, ok. I will add that in. I was interpreting your original comment as more just making sure to clearly label what the results coming out of the database were. I understand now what you meant and will add that paragraph into the conclusions. I also think that definitely helps explain where we are currently at with the error model as shown in the notebook. Thanks.
Here to follow along development of simple error model during DESC Hack Week.