DM-30535: Update the docs, including API changes #58
Conversation
fred3m
force-pushed
the
tickets/DM-30535
branch
3 times, most recently
from
a935587
to
48f3d0c
Compare
| @@ -442,7 +448,7 @@ def updateBlendRecords(blendData, catalog, modelPsf, observedPsf, redistributeIm | |||
| contains those columns. | |||
| """ | |||
| # We import here to avoid a circular dependency | |||
There was a problem hiding this comment.
This is an artifact of the current state of the code and is not worth fixing. Before the scarlet models were exported, the only place to make measurements on the deconvolved models was in the ScarletDeblendTask. But these are band-dependent measurements and the output from ScarletDeblendTask is now a single monochromatic catalog. DM-34558 will create a sub-task to make these (and potentially other) measurements, but until that is created there is really no appropriate place for these measurements to live.
| blend : `scarlet.lite.LiteBlend` | ||
| The full scarlet model for the blend. | ||
| """ | ||
| # Import here to prevent circular import |
There was a problem hiding this comment.
As above, surely you can refactor these modules to avoid potential circular imports?
There was a problem hiding this comment.
In general I agree and most of the refactoring in this ticket was to clean things like this up. However in this case I think that it would add an unnecessary module containing only a single function.
fred3m
force-pushed
the
tickets/DM-30535
branch
2 times, most recently
from
c3af6c7
to
9c30e37
Compare
| While it is outside of the discussion in this tutorial, we have observed reddening residuals in edge-on galaxies and it is currently an active area of research to better model galactic (and intergalactic) dust using *scarlet*. | ||
|
|
||
| Because the *scarlet* model is multi-band, and the observations are not required (or recommended) to be matched to a common PSF, the model itself must exist in a space that can be convolved into the observed seeing in each band, which we will refer to as *model space*. | ||
| In order to be sufficiently well sampled we currently use a PSF for the model space that is a circular Gaussian with ``σ=0.8`` pixels, ensuring that our model image is well-sampled but smaller than the observed PSF in each observed band. |
There was a problem hiding this comment.
Well, yes, but FFT convolution requires zero-padding the image, which I imagine can get quite memory intensive for a large blend without using sparse matrices. Direct convolution doesn't need padding or even a full-sized kernel image, amongst other things.
| In that case setting :attr:`~lsst.meas.extensions.scarlet.ScarletDeblendConfig.setSpectra` to ``False`` is recommended. | ||
|
|
||
| .. note:: | ||
| An interesting fact about optimization in scarlet is that the best initialization (in terms of 𝛘²) will not necessarily result in the fastest runtime or even the most optimal fit. |
There was a problem hiding this comment.
Ok, then perhaps just mention that this is optimizer-dependent and not some quirk of other aspects of Scarlet.
| .. image:: images/scarletModel3.png | ||
|
|
||
| While much improved, the models still aren't perfect. | ||
| Notice that the *scarlet* model for source 0 has stray flux to the upper left and lower right that (fortunately) doesn't matter in the flux re-apportioned model, however sources 1 and 2 have their flux slightly truncated due the the presence of their neighbor in both the *scarlet* models and flux re-apportioned models. |
There was a problem hiding this comment.
Ok, then again I would note that this seems to be the optimizer's fault, rather than a problem with model selection or anything else the user could try to fix.
| In this case the main assumption is that an astrophysical source can be modeled as a collection of components, where each component can be described by a single SED that is roughly constant but can vary in magnitude over its spatial extent (morphology). | ||
| This assumption can be justified physically by considering the structure of a galaxy. | ||
| In general the populations of stars in the bulge and disk are different, as well as perhaps additional populations of star forming regions, each with its own distinct SED. | ||
| However, the actual colors in each population are not that different, which is why parametric bulge-disk models (Sersic models, Spergel models, Gaussian mixture models, etc.) have been successfully used in deblending and photometric measurement. |
There was a problem hiding this comment.
Ok, it sounded ambiguous so I suggest changing it to "the range of colors within each population is not that large", and potentially add something like "e.g. star-forming regions are blue, and bulges tend to be red".
fred3m
force-pushed
the
tickets/DM-30535
branch
from
9c30e37
to
9aeff6f
Compare
mwittgen
changed the title
No description provided.