Project Dust Storm

Can we observe dust moving? In other words, can we see the proper motions of the ISM? It should be moving around relative to us. There is some known about the shell of ISM around the Sun, but mystery remains about the small-scale structure of the ISM throughout the Milky Way.

There are 2 ways to approach this problem:

1. Direct observation

  • requires high-res molecular (radio/mm) maps over time, or high-res IR emission maps
  • most current maps available in literature probably not sufficient over large ares, and certainly not stable over time.
  • most IR surveys do not have the required resolution, but future missions would
  • radio/mm probably has the resolution and stability needed to directly track the motions of the near-field ISM
  • gives some info about density structure, but lots of info about kinematics and dynamical structure of the ISM

2. Variations of extinction, E(B-V)

  • probably easier to detect in the blue optical, can possibly be done from large surveys determining the extinction of distant stars shining through near-field ISM that is moving
  • gives less specific detail on the motions of the ISM relative to the Sun
  • does give info on the density structure of the ISM on potentially very small scales
  • people have started to compare dust extinction maps, e.g. Planck vs WISE which may be a way forward

Hack Day Idea - Kepler FFI's

These images have been largely ignored, though the opinion of the community has been that if the right question were posed they would be a gold mine.

I don’t have the perfect question, but I have a few mediocre ones:

  1. Can we find transients in the Kepler FFI’s?
  2. Can we find new eclipsing binaries or pulsators (use supersmoother, like w/ S82 work) using the FFI’s?
  3. Can we find stars with long term variability (e.g. LPVs, YSOs) using the FFI’s?

Images are here

These transients could be from Novae/SuperNovae, or flares, or asteroids (less likely given 30min cadence)

A first-pass was done as an REU project years ago

And this is the only real result in the literature using FFI’s I’m aware of:

There are 53 of these images, w/ 30min exposures each, and an area of ~100 sq deg. While this pales in comparison to the data from the Kepler targeted objects, it is actually not a shabby time series catalog. Stripe 82 had an area of ~300 sq deg and 70-90 exposures…

Update: Ben Montet is on it!

How many eclipsing binaries in an open cluster?

This is a simple question I’ve been interested in for a long time: Given excellent statistics about the stars we see in an open cluster, how many eclipsing binaries should we observe in a given dataset (e.g. Kepler)?

Given excellent observations of a cluster, we can constrain the properties of binaries. The parameters to fit would be the total binary fraction and the mass ratio distribution (maybe could be described using a slope or something).

An older cluster’s dynamical/tidal history through the Galaxy should affect it’s binary population. This might be an interesting way to constrain that history versus field stars. Or versus young clusters.

Other Cluster / EB ideas: It would be nice to have a definitive “Catalog of Variable Stars in M67” based on every available time domain dataset.

  • K2
  • Catalina
  • several one-off datasets

Seems like a great hack-day idea too, doing really crappy forward or Monte Carlo model of what you should see, taking in best guesses for everything

Recover Differential Rotation using ML?

We showed in Aigrain et al. (2015) that, so far, nobody has an accurate way to recover differential rotation from simulated light curves.

The most promising approaches revolved around studying the power spectrum or periodogram. The main peak (due to rotation) would be broadened or split.

However, this signature can also be caused by starspot decay and evolution, or the appearance of many spots, even without any differential rotation in our model (alpha).

I hypothesized that the periodogram peak structure might be able to recover some type of combined parameter that includes many of these degenerate stellar properties. Perhaps this would be enough information to be useful to theorists?

My attempts to find such a “starspot evolution parameter” based on the peak broadening in the periodogram has not shown any correlation with simulated properties from the Aigrain model light curves. I also tried a simple Random Forest regression of the peak width (and width relative to the period) versus every simulated property for all 1000 light curves. Nada

So the question/idea is: can we invert the problem, do machine learning on the power spectrum or periodogram itself to try and learn what the features of interest are? A giant amount of simulated light curves (even just using a super crappy evolving sine-curve code) would be helpful to tease out results….

How many halo stars are in Kepler?

There should be a small number of stars in the Kepler (and K2) data that belong to the halo, just based on their dynamics.

It might be interesting to find them!

These would have systematically older ages, and would be useful for stretching our gyrochronology methods.

It looks like Sebastian Lepine has already written a K2 proposal on this.