SUNYSB Hack Day output

[cwwalter]

A place to put hack day output.

[humnaawan]

Some plots based on Dask files:

[sowmyakth]

Ellipticities for different star galaxy separation value of psfMag - cmodelMag

[ellongley]

Sky locals ! Got setup at NERSC

[humnaawan]

An attempt to calculate the 2pt correlation for the (super*) extended objects.

Undithered: Tot objects: 11288832 Tot (super) extended: 11573

Dithered: Tot objects: 12311932 Tot (super) extended: 13932

Randoms are not ok: Undithered:

Dithered:

But the 2pt function differs when using the same terrible random catalog:

• generally extended objects: psfMag - cmodelMag > 0.145. Defining super extended as psfMag - cmodelMag > 3 just to narrow down the number of objects.

[SimonKrughoff]

@humnaawan @ellongley @sowmyakth you already may know this, but it is important to filter the data on certain flags. An important one is isPrimary. That flag will make sure you are not double counting where the patches overlap.

I know the flags weren't available last week, so maybe that's still an issue, but I wanted to mention it before I forgot.

[humnaawan]

@SimonKrughoff yes, @fjaviersanchez pointed this out. the Dask files I used didn't have the flags (one of the possible reasons why undithered w(\theta) is much stronger on the small scales) but @cwwalter has mentioned that they will be updated. Hopefully, we can then get more reliable diagnostics.

[drphilmarshall]

@sowmyakth Interesting measured ellipticity distributions! Am I right in remembering that a sawtooth distribution results from a population of thin circles at random inclinations in 3D? Puzzling out the CatSim universe from the DC1 data :-)

This makes me wonder: @cwwalter where should new additions to our science wishlists for DC2 be issued? There could be a lot of little requests for DC2 enhancements coming out of investigations like these. Thanks for making them possible!

[cwwalter]

This makes me wonder: @cwwalter where should new additions to our science wishlists for DC2 be issued? There could be a lot of little requests for DC2 enhancements coming out of investigations like these. Thanks for making them possible!

I'll probably be setting up a DC2 repo like this soon. But, I wanted to think about it and if there should be stuff in both of them and if there should really just be one etc...

[ellongley]

@fjaviersanchez Is this a bit what you meant by "2D histogram with the median SNR (flux/fluxerr) as a function of position in the sky" I selected the stars (is_primary==True and extendedness ==0) using the butler, and plotted the median flux/fluxerr in a bin for stars and extended objects? Thank you so much !

[fjaviersanchez]

@ellongley Good job! It looks kind of high for the median (I expected it to be ~20 or 30). Are you sure that you are plotting the median and not accumulating the SNR in each pixel -- i.e. plt.hist2d(ra,dec,weights=SNR) (which is also interesting)?

[ellongley]

yes it is the median! thank you !

[cwwalter]

Is this one patch?

[cwwalter]

yes it is the median! thank you !

Maybe a 1d distribution of the the values would be useful?

[fjaviersanchez]

@ellongley It is kind of weird that you get such high values but let's make more tests. What if you use less bins (like 15 x 15)? This is the distribution of SNR for stars that I get for one randomly selected patch (I am trimming the histogram at 100 but there are very few entries with SNR higher than that and the max SNR of any object in that patch is ~5300):

I get a median value of ~18.9 for the entire patch so I expected the median in your plot to be close to that too.
Could you please send me the code that you are using to compute the median and also produce a plot with the mean across the sky?

@cwwalter I think it's just one patch (judging by the size of it).