New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Optimize for MWS-specific tiles in minisv3 #601
Conversation
Obsconditions and priorities are set assuming that MWS-specific miniSv3 tiles will be marked POOR (for regular dark and bright MWS observations) and TWILIGHT (for tiles with very bright stars).
The idea is as follows. For miniSV3, define all the "MWS" observations as equivalent to
and remove With these changes, I don't think there would be any need to adjust priorities except for the small tweak |
For reference, I see the following mix of targets on the 65008 miniSV3 tile Sarah already made for us by hand. First line is skys. You might need to scroll right to see the number and priority columns
|
With tweaks in this PR, I get the following (without manual intervention):
|
Accounts for SV0_BGS and SV0_MWS having different obsconditions.
@geordie666, I would appreciate your input on this desitarget PR for streamlined production of MWS minisv3 tiles. A more detailed explanation is in the github thread for the PR (above, if you're reading this on github already). @Srheft may be interested too. Main changes:
Some basic diagnostics of the outputs from this PR for MWS minisv3 tiles can be found here: You can click through to specific tiles. For example: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks, @apcooper. This PR looks good to me and the obsconditions
hack is entirely defensible.
I've included some minor requested changes. Most important is to record the new MINI_SV_MWS_FAINT
target class on the wiki. If you don't want to address my other suggestion about making a _cmx_calc_priority()
function, that's fine, I can deal with that in a later PR.
I ran this branch end-to-end, so I can confirm it works (through the end of the targeting/MTL stage). In the HEALPixel I ran, there were about ~12% new targets that were purely MINI_SV_MWS_FAINT
and after running MTL, for MINI_SV_MWS_FAINT
targets (without a zcat
), PRIORITY
was set to the UNOBS
priority (5
) and NUMOBS_MORE
was 1
.
Spins off special case calc_priority() logic for cmx into a separate function for readability.
Forgot priority and nobs in previous commit
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I've made the changes you requested, @geordie666. Thanks.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This looks good, I tested it end-to-end again and everything appears to work as expected.
Feel free to self-merge this whenever you're ready.
Merged, thanks for the checks! |
I thought I would document work in progress on this as a PR rather than email threads.
It was mentioned that, ideally, fiberassignment files for minisv3 observations should be produced using only reproducible
desitarget
andfiberassign
code, without the manual tweaking that was involved up to now (e.g. Sarah editing out target classes from certain tiles, changing priorities etc.).MWS is already OK on this for SV1, but I've only just caught up with the equivalent procedure for SV0/minisv. Thanks to Sarah and Adam for their help.
Now my goal is to create sky-ready
fiberassign
files for the MWS SV0 tiles proposed at the Tucson meeting for observation in miniSV3 (https://desi.lbl.gov/trac/wiki/MilkyWayWG/MiniSV3). MWS also defineVERYBRIGHT
counterparts for some of those tiles.Within the constraint of not hacking anything, my goal is to maximize
WD
andMWS_CLUSTER
targets in these fields and fill the bulk of the remaining fibers withSV0_MWS
. The VERYBRIGHTtiles should maximize fibers on the
MWS_CLUSTER_VERYBRIGHTtargets and fill with
BACKUPtargets and/or
SV0_MWS`.I don't think this is possible with the current
cmx_targets.yaml
. Tiles with obsconditionsBRIGHT
are going to be dominated bySV0_BGS
(which have higher priority thanSV0_MWS
) and theMINI_SV*
classes (which have very permissive obsconditions).The only way I can see to get what MWS prefers without hacking outside desitarget is to abuse the condition definitions a little in SV0 (assuming we can get away with this, because tiles are still picked manually at the telescope).
That's the point of this PR. It also includes a few other fixes that might be desirable in any case, but those would be easy to cherry-pick if needs be.