To continue a discussion from Bremerton that I had with Kem, here is the list
of potential improvements to opsim that accumulated during pre-Bremerton analysis:
1) expTime
this problem may have been fixed by now (it should be easy I think); if so,
could this fix be confirmed?
This fix has been done and was released with v 3.3.2. (Kem)
2) m5 (and we need MAF to be able to see zero points)
we need to update the fiducial m5 depths; Lynne and Peter can provide
python code to replace the existing module; we need somehow to
enable MAF to see what zero point values were used (e.g. it would be
good to add them to MAF output page where various input parameters
for proposals are listed. Caveat: it will take us a a week or two to completely
finalize the latest and greatest m5, but the process could be exercised
using values from the overview paper even before then.
*** important: we need to take readout noise into account (see eq. 7 in
the overview paper). ***
3) radians vs. deg
MAF shouldn’t expose the user to radians. Right now, MAF reads values
from opsim db that are in radians. We can either fix this in opsim, or in
MAF, but we need to make a decision soon. This should be easy, too.
Once these three hopefully minor modifications are in, we need to rerun
all simulations that are mentioned in the analysis report for cadence white
paper (which became chapter 2 in WP and will be edited there). Meanwhile,
Peter and Lynne will consolidate all the new metrics from MAF that were
developed over last few weeks and bundle them in a suite of plots/tables
designed to emphasize quantities of relevance to science programs (as
opposed to quantities interesting from engineering point of view).
4) no extra visits (if pair, no 3, 4 etc per night)
Kem reported that up to a week of Francisco’s time should be sufficient
to fix this. It would be good for this to happen sooner rather than later
as this plays a role in how efficient a survey is for NEOs, and we have
a major report to complete, which will require additional opsim runs.
A list of modifications of enigma_1189 aimed at boosting the NEO
completeness will be provided in 1-2 weeks (we need to complete
additional NEO-centric analysis of enigma_1189 first).
If “no extra visits” problem can be fixed within a few weeks, then we
should wait with the rerun of all WP simulations.
5) western bias
We agreed that we should understand better what exactly is happening
(e.g. do we start in west and can’t move away, or do end up there
after scanning around and then get stuck, is alt-az distribution same
for all parts of the sky, same as the survey progresses, etc). Lynne,
Peter and I will do this analysis, but only after a number of more pressing
tasks are completed.
It is likely that addressing this problem will be left for v4 (unless we
discover that the cause is trivial).
6) sky brightness
Implementing Peter’s new sky brightness model would allow us to
explore Stubbs’s suggestion to utilize more twilight time. Twilight time
would also have a major impact on NEO performance because we could
get more so-called “sweet spot” observing. However, given all constraints,
we all agree that the new sky model should be implemented in v4 and
not in v3.
To summarize, everyone understands that tampering with v3 should be
kept at a minimum. At the same time, we have two big projects that
cannot wait for v4: NEO completeness analysis and cadence white paper.
Both should be completed by the end of this year (which implies that
opsim/MAF outputs should be available and re-analyzed by the end of
October - at the latest!).
Just a footnote: the NEO stuff has come to a head and most of the runs/analyses will likely be done by mid-September (Kem).