search

jradavenport / gaia_unresolved

Unresolved binaries from the Gaia CMD?
MIT License
0 stars 0 forks source link

Ask people their thoughts #1

Open jradavenport opened 7 years ago

jradavenport commented 7 years ago

Most literature on unresolved binaries has a quote like "...neglecting the effects of unresolved binary stars". Humbug!

ixkael commented 7 years ago

I think it's definitely possible! One could deconvolve the main sequence and the "binary sequence" simultaneously, and infer the probability of two objects being in a binary given their two absMags and colors. Of course this is specific to unresolved binaries - in order to generalize one would need to model the way objects are resolved or not. This shouldn't be too hard: we could add a simple model for p(resolving a pair of objects | both colors, magnitudes, and distances). If we adopt gaussians or gaussian mixtures for all those distributions then the inference should be easy! If you like it I can put it on paper and make a stan prototype. Tagging @keithhawkins since we just discussed it!

ixkael commented 7 years ago

But maybe deconvolving both sequences is a bit ambitious. Using stellar models is obviously a compelling alternative. Again, I would model p( two stars are in a binary or not and they are resolved or not | both colors, magnitudes, and parallaxes ) as a Gaussian mixture. We would infer its parameter and learn about the binary population while reconstructing/deconvolving the binary sequence.

jradavenport commented 7 years ago

(for clarity) I was thinking "resolving" only within the CMD, not on the sky, so we'd be doing a mixture model of single & binary stars in that CMD.

But you're totally right, the bigger fish is doing the detection of binaries over all size scales, and thus resolving 2 stars in either the CMD or on the sky.

I'm not sure what the right "space" for doing the mixture model is - if that makes sense? Points don't move linearly in the CMD between the single-star main sequence and the equal-mass binary main sequence. Also, there's good reason to believe secondary stars don't follow a gaussian mass distribution... so my brain is having a hard time (after admittedly only a few min of naïve pondering) figuring out what the right dimensions to work in.

jradavenport commented 7 years ago

I've been thinking about this as a "forward modeling" exercise. Trying different binary fractions and binary mass ratio distributions to simulate the CMD we should see (incl. uncertainties), and then seeing which is consistent with the observations.

This seems like the most brute-force approach though

davidwhogg commented 7 years ago

Oddly, @andersdot and I have strong posterior beliefs about a bunch of unresolved binary stars in TGAS. Maybe we should say or do something about that...

jradavenport commented 7 years ago

I hope you do! And let me know what those beliefs are :)

On May 25, 2017, at 5:58 AM, David W. Hogg notifications@github.com wrote:

Oddly, @andersdot and I have strong posterior beliefs about a bunch of unresolved binary stars in TGAS. Maybe we should say or do something about that...

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub, or mute the thread.

davidwhogg commented 7 years ago

Will discuss with her next week...

ixkael commented 7 years ago

Tagging @adrn and @smoh