search

desihub / qlf

Developement of the QLF framework for DESI
BSD 3-Clause "New" or "Revised" License
4 stars 4 forks source link

Fix snr obj #281

Closed felipeaoli closed 6 years ago

felipeaoli commented 6 years ago

In this branch I updated the qasnr/main.py for desispec 0.25, remove obsolete functions, fixed obj_sort and treated bad_object fibers for snr visualization.

felipelm commented 6 years ago

@felipeaoli there are no references for QSO and STAR?

screenshot 2018-11-01 10 47 08

felipeaoli commented 6 years ago

There aren't. For night 20191001+ desispec 0.25, under the keys ['TASKS']['CHECK_SPECTRA']['PARAMS']['FIDSNR_TGT_DARK_REF'] there is list with only two values.

@Srheft, do you know if this behavior is expected in desispec 0.25?

Srheft commented 6 years ago

@felipeaoli the REF values are in transition state. QL can't have correct target type for all the references and all the programs right now. in the night 20191001, the fiber maps still have target names (ELG, QSO, ...)so a REF for DARK program should have 4 targets [so there should be a list of 4 numbers for FIDSNR_TGT]. In the night 20200515, fibermaps divide targets to only two categories 'SCIENCE' and 'STD' so FIDSNR_TGT_DARK would show 2 numbers. If this is making a table not to work in QLF, I'll make a commit for a temporary fix. Please let me know so that PR707 could accommodate this for you. Thanks.

felipeaoli commented 6 years ago

Thank you very much for this clarification. For while, I prevented the table to break but I am just showing the first two values given by the ref list.

Given your reply, is it correct to assume that QSO, elg and lrg will have the same ref value? If this is correct, I can change the table.

Srheft commented 6 years ago

@felipeaoli technically theREF values for QSO and ELG and LRG, ... should be slightly different as they have intrinsically different luminosity range and so the SNR vs. Mag[or flux] fitting parameters would take different values but because they are not "fine-tuned" in the master now, the REFs are the same. Even if there is only SCIENCE and STD target type, each of the DARK, BRIGHT and GRAY programs would have different SCIENCE targets and so their FID_SNR_REF should be fine-tuned to those targets.The next few PRs would hopefully address them and I hope the fiber map format settles by then too. Please let me know if any temp fix in QL would help your development.