Repository navigation
Milky Way static science case #626
Description
@akvivas @cbritt4 @vpdebattista @caprastro @yoachim @pmmcgehee @jgizis @knutago @dnidever and all MW-interested co-authors:
Prompted partly by Jason's review of the MW chapter (see github issue #620 ), I would like to develop at least a skeleton "static science" case and figure of merit for Chapter 4, before the WP gets posted to the arXiv if possible.
I have self-assigned this "issue," but all input is of course welcome. Please let me know if you're interested to develop this, and/or if you've already got material in mind or developed!
To start things off, I've come up with the following "pseudocode" for how a science case and associated figure of merit might look. I think all the steps below are already implemented in sims_maf or maf_contrib, it'll just be a case of pulling the pieces together. This may also be a useful stepping-stone on the route to Monte Carlo population studies that we've each discussed at various points.
@yoachim - if you have time to look through this, can you let me know if there are any sims_maf tricks we should be using to accomplish the various loops here? (I'm thinking in particular of the assignment of apparent magnitude to each of the spatial slice-points)
Static-science MW figure of merit: fraction of fields well-measured for photometric metallicity and reddening estimates, for some fiducial object, in a survey covering the inner plane
- Decide what fiducial star to use... how about an object at a 10 Gy main sequence turnoff for solar metallicity?
- Produce absolute magnitudes in {ugrizy} for the fiducial marker of interest (I believe sims_maf has spectral models to be used to do this);
- For each healpix (i.e. pointing in the survey):
3a. place the fiducial star at an appropriate distance (say, along the galactic bar mid-plane?); compute the distance modulus for this star;
3b. modify the apparent magnitude thus produced by reddening (I believe sims_maf has methods to do this);
3c. compute the photometric uncertainty for uncrowded photometry (e.g. sims_maf m52snr), in all filters;
3d. compute the photometric uncertainty due to spatial confusion (e.g. metrics.CrowdingMagUncertMetric() );
3e. Determine how best to estimate the uncertainty due to confusion, from the total set of observations (re: @yoachim 's comment later in this sequence).
3f. Add the two in quadrature to produce photometric error in all 6 filters, accounting for both photometric depth and spatial confusion; - The result should be an array with photometric uncertainty in {ugrizy} for our fiducial star of interest. Now use uncertainty propagation to compute color uncertainties {u-g, r-i, i-z};
- Count the fraction of sight-lines for which the color uncertainties are below the threshold needed for "sufficient" accuracy in parameter determination. (for example, I believe Ivezic et al. 2008's tomography-metallicity SDSS paper suggests 0.05 mag would be sufficient for metallicity determination). This is the figure of merit.
Thoughts welcome!
Will
Activity
drphilmarshall commentedon Apr 26, 2017
willclarkson commentedon Apr 26, 2017
Thanks @drphilmarshall - what I personally have in mind is the uncertainty in age determination for each of multiple spatially-overlapping populations towards the inner Milky Way. Eventually I'd like the community to be able to make a statement like "strategy X will lead to uncertainties in the timing and duration of each star formation epoch in the bulge of Y and Z Gy respectively, and in the number of epochs of N." However I think that's a long way beyond what we can do at the moment with our current understanding of opsim and maf, the figure of merit suggested here is a step in that direction.
But that's my own scientific preference speaking... generically, any science conclusion requiring metallicity, temperature and reddening determination for stellar populations will scale with the fraction of the survey area that allows those parameters to be measured down to the turn-off.
Co-authors: examples from your favorite science areas are welcome!
yoachim commentedon Apr 26, 2017
I think we've got all the tools to do that. The questions I have:
willclarkson commentedon Apr 26, 2017
Thanks Peter (@yoachim) -
What I have in mind is to use model stars along the bar as a reference, yes. I'm trying to come up with a figure of merit that doesn't require a full-blown population simulation, and "fraction of fields that can be well-measured for bar-like populations" seems to be a good place to start. At least, I would expect the results of a full-blown simulation to scale with the results of the simpler suggestion I've made here.
Very interesting point, and indeed the uncertainty in a total set of observations along a particular sight-line (for photometry and astrometry) is something I would expect to be crucial for a very large fraction of the science cases... How does MAF currently compute this?
yoachim commentedon Apr 26, 2017
Right now, we have the Coaddm5Metric which assumes Gaussian errors and combines the depths of multiple images accordingly. Looks like the crowding metric uses the best seeing in the group of observations to compute the crowding uncertainty.
For the astrometry metrics, the depth of each image is used in the computation. The uncertainty in the proper motion is treated as computing the uncertainty of the slope in a linear fit. The proper motion is treated like fitting the amplitude of a sin-wave (where the phase and period are known). I don't think we have any astrometry metrics that take crowding into account, the centroid errors are assumed to be a function of seeing and depth with a floor set by the atmosphere.
willclarkson commentedon Apr 26, 2017
Thanks Peter - it sounds like those of us interested in this issue should take another look at Olsen, Blum & Rigaut 2003, there may be some useful recommendations in that paper. Although at first glance the seeing:confusion interplay may seem like a second-order issue, I can imagine it becoming a main limitation for surveying crowded regions with LSST. Milky Way static science seems as good a place as any to incorporate it into the observing strategy assessment.
pmmcgehee commentedon Apr 26, 2017
willclarkson commentedon Apr 26, 2017
Thanks @pmmcgehee - would you mind writing a couple sentences about the GC-populations case you have in mind here, so that any readers of this issue not on the lsst-milkyway-etc mailing list (and who thus missed your earlier suggestion) can see?
pmmcgehee commentedon Apr 26, 2017
drphilmarshall commentedon Apr 27, 2017
drphilmarshall commentedon May 18, 2017
Hi all - when you make your pull request with the new MW static science case, can you please make sure you refer back to this issue so we can close it out? Thanks!
willclarkson commentedon May 26, 2017
The science case is now specified Section 4.5 in the latest pull request (PR #643) - once @drphilmarshall approves that pull request, it will be in the main paper branch. I'm moving implementation of this science case to the "Version 2.0 Development" milestone.
pmmcgehee commentedon May 26, 2017
[-]Chapter 4 (Milky Way) - Static science case before arXiv posting?[/-][+]Milky Way static science case[/+]drphilmarshall commentedon May 26, 2017
Good plan, thanks Will. I edited the issue title to match, feel free to change it.
willclarkson commentedon May 26, 2017