We have a released a set of v5.0 simulations, in response to early commissioning feedback.
In all v5.0 simulations (except the “two_snap” simulation), exposure times per visit are 1x30s in grizy and 1x38s in u. This is in accordance tests with single-snap visits from early commissioning with LSSTCam.
The default Deep Drilling Strategy has also been updated to the “ocean” strategy, where most sequences in the DDF are much shorter (a handful of exposures in two or three out of the ugrizy bands, alternating with the remaining bands within a night or two) but much more frequent; for each DDF, there is also a season (~year) where the sequences are much longer (over one hundred visits in each sequence) while remaining relatively frequent. The total number of visits per DDF is approximately the same as in previous simulations, but the visits are distributed over time quite differently.
The slewtime has been increased, attempting to incorporate some early indications from the TMA performance … we anticipate that this will change as we move further from commissioning into operations, but by incorporating a slower-than-previous baseline speed we hope to preserve a better match with average survey performance.
Slides describing the general changes with v5.0 were presented at the RCW.
Opsim Databases for these runs are available to download from USDF.
Notebooks summarizing metric performance
Full MAF outputs
The v5.0 simulations include:
- baseline simulations - there is both a baseline_v5.0.0_10yrs and baseline_v5.0.1_10yrs, reflecting small software changes in rubin_scheduler, which should not statistically change the output simulation
- baseline_production simulations - these should be very similar to the baseline, but are run with a configuration more like what we need to actually use for the summit. This is essentially a code reorganization, however there are some small additional changes such as adding azimuth constraints at the end of the night (consistent with summit performance requirements at present).
- four-roll - rolling cadence but with four cycles of rolling instead of three, so no uniform seasons
- two-snap - what if we did 2x15s visits … primarily for comparison to previous simulations
- ddf_dither - this tests different sizes of dither offsets. These dithers are only changed between nights.
- ddf_all_dither - this tests different sizes of dither offsets, including dithering within each night. Note that we almost certainly have to adopt this method (especially with the deep sequences), and we will almost certainly need at least a range of 0.2 degrees in dither sizes due to scattered light.
- variations on the DDF ocean strategy -
- ddf_acc (an accordion strategy added on top of the ocean shallow sequences)
- ddf_acc_short ( accordion but making the shallow seasons shorter than in ddf_acc)
- ddf_acc_early ( like the accordion strategy, but moving all of the deep seasons as early as possible - this strategy would acquire the bulk of the DDF visits within the first half of the LSST).
- variations on template acquisition - production_templates_y1 is the most relevant here. This adds a template tier during y1, configured to attempt to match template requirements we saw during commissioning.