Dynamical Constraints on Multi-Planet Systems

Dynamical Constraints on Multi-Planet Exoplanetary Systems
http://arxiv.org/abs/1302.5247

Quote :

As a direct result of ongoing efforts to detect more exoplanetary systems, an ever-increasing number of multiple-planet systems are being announced. But how many of these systems are truly what they seem? In many cases, such systems are announced solely on the basis of orbital fits to observational data, and no attempt is made to see whether the proposed orbits are actually dynamically feasible. As a result, it is certain that planetary systems are being announced that involve planets moving on orbits that would be dynamically unstable on timescales of just a few hundred years.
Here, we present the results of dynamical simulations that investigate the orbital stability and evolution of a number of recently discovered exoplanetary systems. These simulations have enabled us to create highly detailed dynamical maps of those systems, allowing us to better constrain the orbits of the planets contained therein. In some cases, our results have even led to the very existence of the planets themselves being called into question.

Of note, HD 142 d (unannounced) 0.3 Jupiter-mass planet at 0.47 AU and e ~ 0.3 is stable.
HU Aqr is highly unstable.
HW Vir is extremely unstable.
NSVS 14256825 is "incredibly unstable."
NN Ser appears stable.

Quote :

The fact that the proposed planetary systems around HW Vir, HU Aqr and NSVS 14256825 are so strikingly similar (featuring an inner planet on a circular orbit and an outer planet with an orbital period almost double that of the innermost, moving on a highly eccentric orbit) suggests that there may be some common property of these systems that is mimicking the presence of planets.

HD 155358 is stable.
HD 204313 is stable only if the b and d planets are in a 2:3 resonance.
HD 200964 is stable and likely trapped in a 4:3 MMR.

Hmmm nice review, any more obviously suspicious systems?

Gliese 667 C (unpublished 6-planet version) doesn't seem to be stable with the best-fit parameters, whether there are any feasible nearby stable configurations I don't know.

HD 142 d isn't unannounced, just tentative and unconfirmed - see Wittenmyer et al. (2012). HD 142 (A) is a fairly fast rotator, and it's not well understood how much rotational velocity affects radial velocity.

Other than that, it is bizarre that so many of the purported planetary systems orbiting post-main sequence eclipsing binaries are multiple - most of what is understood about giant planetary formation of >1.5 solar mass systems points to that they end up small and occasionally multiple or large and lone. Almost all of them being unstable is yet more bizzare.

Looking over the parameters of some of these systems, I wonder now and again if Velicovsky (sp?) was 'right' - except this chaos was happening in other solar systems, not ours.

Well I wait for another paper for gj667C is there any high resonance 9:7 or something between 28.1 and tentative 30.8?

Shellface wrote:

Other than that, it is bizarre that so many of the purported planetary systems orbiting post-main sequence eclipsing binaries are multiple - most of what is understood about giant planetary formation of >1.5 solar mass systems points to that they end up small and occasionally multiple or large and lone. Almost all of them being unstable is yet more bizzare.

There are other oddities too: it looks like the frequency of giant planet systems around these binaries is suspiciously high, far higher than what would be expected for planetary systems around the progenitor stars. The idea that we are not actually detecting planets but the signature of some process intrinsic to the binary stars themselves is quite an appealing one...

I've been wondering about this lately.

The calculations in the OP seem to be made with the view that the distance to the stars in question is accurate. But - how were these distances determined? The only really reliable method is trigometric parallax, which generally takes years. Is Kepler doing parallax work for these stars?

Now...if the stellar distances are being determined via photometry...that can cause some problems. A few years ago, I spent a lot of time going through the old photometric distance catalogues and comparing those distances with Hip parallaxes, ideally Hip parallaxes accurate to withing 5%. Most of the old line photometric distance catalogues were just plain wrong - out by something like 40% - almost half the time. They tended to be within 10% only about one time in four.

The best I found was by a Dr Ed Weis, who did photometric distances to several thousand K and M dwarfs. He used a three color averaging system, and was accurate to within 15% about 75% of the time (where Hip parallaxes with error bars of less than 5% were available). I don't believe the current photometric distance determination methods are any better.

I would also point out that photometrically determined distances to binary stars are especially suspect.

So...what if these stars are say...15% - 20% closer than the given distances - something very possible if the distances were determined photometrically. That would expand the size of the whole system, possibly enough to negate some of the problems cited. Or do I have that backward?

The distance is actually not a factor. These systems were detected by eclipse timing measurements, so what is actually being measured here is the O-C of the eclipse times as the binary star wobbles around the system barycentre (the light taking more time to cross the star's barycentric semi-major axis). That time delay is therefore simply a function of the mass and orbit of the planet, and will be the same regardless of distance.

Incidently, Doppler spectroscopy is also unbound to distance (aside from how it affects the target star's brightness) as it measures simply the Doppler shifting of the spectral lines, again being only a function of the planet's mass and orbit and not related to its distance from Earth.

Drat! I was under the (false) impression, that if the star was closer to us than listed, then in order to make the planets orbit fit, the planet would have to orbit further out from its sun than listed. (In sync with the eclipse timing).

So much for that thought!