You can apply for a summer research internship with me in Potsdam via the DAAD RISE program here. The application deadline is on Monday, Dec 16. I collected a number of questions from interested students that I’d like to share with anyone who is interested in applying, or just got curious about the title.
This is the abstract for the project (with a few amendments)
Are flares hidden multiplicity indicators?
Around half of the stars in the sky are binaries, systems of two gravitationally bound stars. Some are known because they eclipse, others can be found by the periodic blueshift and redshift of their spectra. Distant binaries can be detected in microlensing events. Binaries with similar masses can be spotted by their overluminosity compared to single stars, and a few systems can even be observed as separate objects, moving in Keplerian orbits in the plane of the sky. But the vast majority of stars that live in multiple systems remains undiscovered.
However, knowing the fraction of binaries is of primary importance for our understanding of star formation. In the case of binary systems with a solar type (spectral type GV) primary and a low mass secondary (MV) star their small mass ratio aggravates detection because the luminosity of the secondary component contributes less than 1% to total luminosity of the system. The MV hides in the light of its brighter companion.
Fortunately, MVs are known to be actively flaring, while GVs flare less often and relatively less energetically. Flares are explosions that occur in the magnetized atmospheres of stars. They can enhance the stellar flux by orders of magnitude within seconds and last up to several hours. Flare signatures can be detected by their characteristic shape in time-resolved optical observations, or light curves. So in principle we can notice the presence of a small secondary star when it brightens up during a flare. And many stars do: More than 100 000 stellar flares were observed in the field of view of the Kepler space mission and thousands are currently being observed by TESS.
Our goal for this project is to find binaries using excess flaring activity. During this RISE project, we will be developing an empirically motivated statistical model of the multiplicity-flaring activity relation in stellar systems in the Kepler field stars. We will then test our model by comparing samples with confirmed binarity status to those without. Eventually, we will use our model to update the base multiplicity probability for any star with known flaring activity, and present a collection of likely unresolved binary candidates in the Kepler field based on our studies for follow-up observations.
And these are the questions that students have asked:
Do we know the mechanism behind the flaring of the lower mass secondary stars?
We believe that the mechanism that drives flares on low mass stars is essentially the same as on the Sun - re-connection of magnetic field lines in the upper atmosphere of the star leads to a sudden relaxation of their topology. This change accelerates particles and produces electromagnetic emission in all wavelength bands from x-ray to radio. But obviously the conditions in the atmospheres of small, cool stars are different from the Sun’s, and we have reason to think that the magnetic fields of these stars must differ in strength and geometry, too. However, the flares that we observe on the Sun and small stars are remarkably similar. So, until we learn otherwise, Sun-like flaring is a good enough model.
Is the flaring periodic, similar to Cepheid variables, or more random in nature?
Large flares seem to follow a Poisson process in time, and to be independent of one another. For small flares, there is evidence that flares are “sympathetic”, that is, the occurrence of a flare makes the occurrence of a small flare shortly after/before more likely.
There are exceptions to this, but they concern exceptional systems - like eccentric close binaries and star-planet systems with Ultra-Hot Jupiters - where we believe that flaring could occur periodically whenever the two objects are closest to each other.
Does it have something to do with the rotation in the binary system?
It depends on whether you mean the rotation of the individual stars or the orbital parameters of the binary system. The answer is in both cases yes, but the former is much more relevant than the latter. Stellar rotation is tightly linked to flaring: All else equal, the faster the star rotates, the more and the stronger the flares you observe become. The relation between the orbit and flaring is more subtle. Very close orbits, where the two stars exchange mass or perturb each others magnetic fields, may directly cause additional flaring (perhaps even periodically in eccentric systems, see above). But we will be looking into well-separated binaries that do not interact. We treat them as single stars that share age and composition, and are gravitationally bound. It is an open question if there are effects on the flaring activity that stem from the formation history of these binaries, though.
What will the statistical model of the flaring activity involve?
Would you like to know more about the methods we are going to use, or about the physical parameters that will go into the model?
In the former case: We’ll definitely use standard Bayesian analysis and fitting methods like maximum likelihood method, MCMC sampling etc. But since we are working with labeled (binary/single/unclear) data, we might also attempt supervised learning methods. There is no big model code base for what we want to study, so we’ll be developing a simple toy model first, and then see how far we can get with this.
In the latter case: Flaring activity and binarity will be a function of stellar mass for sure, but further down the road we might want to include additional information about the star to make better inferences, like rotation, metallicity, or age. Flaring activity as such also has some characteristic behaviors, which we’ll also incorporate, such as a strong power law relation between the occurrence rate and the energy of the flares.
What will the student spend most of their time doing?
It seems to me that the project will be mostly focused on analyzing archival data to determine a suitable model. Does this sound accurate?
I expect the time to be split between analyzing archival data (35%), the development of a simple model that predicts binarity from flaring (20%), reading the literature on flaring activity and stellar multiplicity (20%), research group activities like colloquiua (10%), and discussing science with me and others in the group (10%) … and perhaps 5% exploring the place, getting settled, and figuring out logistics. These are, of course, rather rough estimates, and might change depending on how the project goes. If we find that the model is worth developing further and it becomes a more theoretical work, data analysis will be less important. If we figure out that we can infer a lot about the systems in question from a statistical analysis, we will dig into the archives more.
What will training be like? Will there be much supervision?
This will depend on you. Supervision will be as close as it needs to be so that you can continuously make progress, and won’t get stuck on anything for too long. I will likely be supervising very closely (=daily chats) in the beginning, less in the middle of the internship period, and more closely in the end when wrapping up, drawing conclusions, and making further plans.
I’d prefer to err on the side of closer supervision, but I am happy to let you explore if you decide to do so. This is a learning opportunity for you, and you should get out of it whatever is most valuable to you on the margin. You might want to reflect on what you expect from the internship before you dive into it, and we can figure out a suitable strategy to achieve both our shared and individual goals together.