Additional Planets at Gliese 581

So Vogt et al.'s latest paper analyses only the updated HARPS measurements. So I decided to play around with the data in Systemic myself, to have a look for myself what is going on and also to investigate the combined HARPS+HIRES datasets.

(You need to extract the HARPS velocities from the source of the Forveille et al. paper on arXiv, the actual paper has them going off the page and therefore unselectable)

For the HARPS dataset only, I find that planet g vanishes in an eccentric fit and appears in the circular fit. The HARPS data do not allow to distinguish between an eccentric planet d and the existence of planet g.

The case of the combined HARPS+HIRES data is more interesting.

For an eccentric fit, the eccentricity of the innermost planet becomes close to zero (as would be expected for the result of tidal circularisation), and the eccentricity of planet d is increased. The next most significant period after fitting the four planets is at 49.8937 days with a false alarm probability of 4.05%, too high to claim a detection.

For a circular fit, the next most significant period after fitting the four planets is at 50.0101 days with a false alarm probability of 0.789%.

In neither the eccentric nor the circular case is there anything significant at a period of 32 days. Combine the datasets and planet g disappears. This is the opposite of the combined HARPS+HIRES data with the previous HARPS dataset, where it seemed to take the combination of datasets to get planet g to appear.

Big mess with this star at all. Around 50 days still very habitable would be but will be it stable for long time?

It would also be dangerously close to the d planet.

One further point.

Lunar month = 29.53 days
Year = 365.25 days

Combine the two as an alias:

1/(1/29.53 - 1/365.25) = 32.1275 days

Which is basically the same as the claimed orbital period of g.

It must be coincidence

Why do you say that? are you familiar with the way aliases related to the motion of the Solar System conspire to inject artificial signals into RV data?

Lunar month is relevant for scheduling observing time: there are some observations that need very dark skies, for which you don't want to have the moon shining lots of light around. Exoplanet RV measurements are not as sensitive to this so end up getting scheduled in other times, which leads to gaps in the data on the period of the lunar month. These gaps can result in a periodicity in the data that may masquerade as an apparent additional planet.

You get similar gaps on the period of a year because of requirements on the star being above the horizon at night. The declination of -7.7 degrees means Gliese 581 is not circumpolar at the La Silla Observatory.

The fact that this planet seems to have a period that is exactly matching the period obtained by combining these two observational cycles means caution is needed!

For now, I'm in the "no planet g" camp.

Lazarus wrote:

For the HARPS dataset only, I find that planet g vanishes in an eccentric fit and appears in the circular fit. The HARPS data do not allow to distinguish between an eccentric planet d and the existence of planet g

In your opinion, is there a way to lift this degeneracy without a significant increase of RV data precision? For example, changing phase coverage, the cadence of data, increasing the number or timespan of observations could help in a near future to solve the problem, or we have to wait a dedicated campaign from some Laser-comb spectrographs?

Lazarus wrote:

The case of the combined HARPS+HIRES data is more interesting.

For an eccentric fit, the eccentricity of the innermost planet becomes close to zero (as would be expected for the result of tidal circularisation), and the eccentricity of planet d is increased. The next most significant period after fitting the four planets is at 49.8937 days with a false alarm probability of 4.05%, too high to claim a detection.

For a circular fit, the next most significant period after fitting the four planets is at 50.0101 days with a false alarm probability of 0.789%

Under some aspects, a planet in a 50 day period is probably safer than a 32 day one, if there is a greenhouse effect to help. The mass of this putative planet is still on the Super-Earth regime?

Lazarus wrote:

In neither the eccentric nor the circular case is there anything significant at a period of 32 days. Combine the datasets and planet g disappears. This is the opposite of the combined HARPS+HIRES data with the previous HARPS dataset, where it seemed to take the combination of datasets to get planet g to appear

By now Vogt has two additional years of unpublished HIRES data, plus observations from the Carnegie Planet Finder Spectrograph, which also helped in the confirmation of GJ667Cc. But I guess that before announcing any new controversial discovery they want to be very sure this time.
It will be also interesting to re-analyze the 240 HARPS data points with the HARPS-TERRA code, since apparently it provides higher accuracy for M dwarfs.

Does taking into account red noise eliminate the evidence for the d planet?

The impact of red noise in radial velocity planet searches: Only three planets orbiting GJ581?
http://arxiv.org/abs/1209.3154

Quote :

We perform a detailed analysis of the latest HARPS and Keck radial velocity data for the planet-hosting red dwarf GJ581, which attracted a lot of attention in recent time. We show that these data contain important correlated noise component ("red noise") with the correlation timescale of the order of 10 days. This red noise imposes a lot of misleading effects while we work in the traditional white-noise model. To eliminate these misleading effects, we propose a maximum-likelihood algorithm equipped by an extended model of the noise structure. We treat the red noise as a Gaussian random process with exponentially decaying correlation function.
Using this method we prove that: (i) planets b and c do exist in this system, since they can be independently detected in the HARPS and Keck data, and regardless of the assumed noise models; (ii) planet e can also be confirmed independently by the both datasets, although to reveal it in the Keck data it is mandatory to take the red noise into account; (iii) the recently announced putative planets f and g are likely just illusions of the red noise; (iv) the reality of the planet candidate GJ581 d is questionable, because it cannot be detected from the Keck data, and its statistical significance in the HARPS data (as well as in the combined dataset) drops to a marginal level of $\sim 2\sigma$, when the red noise is taken into account.
Therefore, the current data for GJ581 really support existence of no more than four (or maybe even only three) orbiting exoplanets. The planet candidate GJ581 d requests serious observational verification.

Before long, there aren't going to be any planets left at GJ 581

Sirius_Alpha wrote:

Does taking into account red noise eliminate the evidence for the d planet?

The impact of red noise in radial velocity planet searches: Only three planets orbiting GJ581?
http://arxiv.org/abs/1209.3154

Quote :

We perform a detailed analysis of the latest HARPS and Keck radial velocity data for the planet-hosting red dwarf GJ581, which attracted a lot of attention in recent time. We show that these data contain important correlated noise component ("red noise") with the correlation timescale of the order of 10 days. This red noise imposes a lot of misleading effects while we work in the traditional white-noise model. To eliminate these misleading effects, we propose a maximum-likelihood algorithm equipped by an extended model of the noise structure. We treat the red noise as a Gaussian random process with exponentially decaying correlation function.
Using this method we prove that: (i) planets b and c do exist in this system, since they can be independently detected in the HARPS and Keck data, and regardless of the assumed noise models; (ii) planet e can also be confirmed independently by the both datasets, although to reveal it in the Keck data it is mandatory to take the red noise into account; (iii) the recently announced putative planets f and g are likely just illusions of the red noise; (iv) the reality of the planet candidate GJ581 d is questionable, because it cannot be detected from the Keck data, and its statistical significance in the HARPS data (as well as in the combined dataset) drops to a marginal level of $\sim 2\sigma$, when the red noise is taken into account.
Therefore, the current data for GJ581 really support existence of no more than four (or maybe even only three) orbiting exoplanets. The planet candidate GJ581 d requests serious observational verification.

Before long, there aren't going to be any planets left at GJ 581

Perhaps two astromety-validated planets with 9.5 and 4.9 Jupiter masses only and inclination (i) of 0.3 and 0.2 respectively. And none so far has detected'em! *lol*

Got to wonder about some of these low-mass, multi-planet RV systems. At this level of detection these effects become really important.

This analysis doesn't particularly rule out planet d, the signal is there but at too low a significance to claim a detection.

On the other hand we know based on the Kepler data that densely packed planetary systems are common...

It's somewhat intriguing that apparently planet d hasn't been indipendently confirmed by HIRES after 11 years and 121 data points. Maybe a future enlarged Keck dataset will shed light on the question.

For now the putative planets f and g are pretty much ruled out.
Planet f orbital period shifting from 433, 399, 191 or 450 days based on the analysis method and datasets available, believed to be a more firm detection than g in the combined KECK + HARPS dataset of Vogt et al. (2010), then apparently disappearing in the enlarged 240 points HARPS datasets (2011); planet g whose existence is based on the orbital property of the now not-so-much confirmed planet d, at 32 or 36 days orbital period, but that can shift even to around 50 days when Keck data is added (see Lazarus'analysis). Are we just trying to catch ghosts?

The apparent properties of these hopeful planets seems to be dependent on the analyzed datasets, which doesn't seem very promising. Correlated noise interactions between the datasets could explain the changing behaviour of the detections, as the paper brilliantly set forth.

Galzi wrote:

It's somewhat intriguing that apparently planet d hasn't been indipendently confirmed by HIRES after 11 years and 121 data points. Maybe a future enlarged Keck dataset will shed light on the question.

The conclusion of the Vogt et al. paper (the one claiming that the new HARPS dataset shows evidence of planet g) does mention that they have more data:

Quote :

Nevertheless, over the past year, we have continued to observe GJ 581, obtaining an-other observing season’s worth of Keck and Magellan PFS RVs. We are also making further improvements to our data reduction pipelines with the goal of eventually compiling a data set sufficient to lift this degeneracy.

(the degeneracy being that between 4 eccentric planets and 5 circular ones)

Certainly I don't think we have heard the last from this system!

Maybe gj 581 d will be back to 82 days

Another study found no traces of planets f and g

http://arxiv.org/abs/1208.0814

Using the HARPS dataset, a stable 4 planets solution is found with an eccentric orbit for planet d. Even in this analysis, when taking into account HIRES data, planet d is rejected for insufficient confidence.

That paper doesn't seem to reference the Forveille et al. (2011) paper which contains the extended HARPS dataset, instead they are using the older Mayor et al. (2009) dataset.

A search for transits of GJ 581\lowercase{e} and characterization of the host star variability using MOST space telescope photometry
http://arxiv.org/abs/1211.0577

No transits of b, but a photometric signal at about the same period. No detection of e. If it transits, it has a radius less than 1.62 R_e (I feel we've discussed this before but I'm not sure where)

GJ 581 e probably has radius around 1.3 earth so no wonder they did not detect transit

Another paper on how many planets are there, the same results.

Counting the number of planets around GJ 581. False positive rate of Bayesian signal detection methods
http://arxiv.org/abs/1211.1280

A debris disk around the planet-hosting M-star GJ 581 spatially resolved with Herschel

http://www.aanda.org/index.php?option=com_article&access=doi&doi=10.1051/0004-6361/201220325&Itemid=129

A resolved disc? Sweet... never was particularly comfortable with the uncertainties in the modelling of disc locations from the spectrum of the excess radiation...

Now on arXiv.
http://arxiv.org/abs/1211.4898

Quote :

This disk complements our view of this remarkable system known to host at least four low mass, close-in planets. These planets cannot perturb sufficiently the modeled cold disk to trigger destructive collisions between planetesimals over the age of the star, but a hypothetical outer planet, for example a Neptune mass planet with an orbital radius of 5 AU and a moderate eccentricity, could replenish the system with dust. Alternatively, the self-stirring mechanism could operate for this old star causing sufficient dynamical excitation to produce the observed dust.

So maybe getting a handle of the 3D orientation of the system:

Quote :

First, we note that our determination of the inclination of the disk relative to the plane of the sky is 30° < i < 70° (face on disk is i = 0°). This is mostly constrained by the 160 μm image and is fairly insensitive to the masks used for the background sources. If the disk mid-plane and the orbits of the planets are coplanar, this range of inclination makes the masses of the planets of GJ 581 no more than ∼ 1.6 times their measured minimum masses by radial velocity and, interestingly, ensures the long-term stability of the orbits in this system as shown in dynamical studies by Beust et al. (2008) and Mayor et al. (2009).

And also hints that there should be outer planets, even if "Gliese 581 f" turned out to be a systematic error rather than a true planet.