Saturday, October 31, 2015

The Ghost in the Window

Vinesh Rajpaul and colleagues have forcefully challenged the reality of Alpha Centauri Bb, the Earth-mass Hellworld proposed as a companion to one of the two nearest Sun-like stars. This extrasolar candidate was announced three years ago by a team led by Xavier Dumusque. Even using the exquisitely sensitive HARPS spectrograph, Dumusque and colleagues noted that detecting such a lightweight object stretched the limits of the radial velocity method and required complex data analytic approaches.

The announcement of this planet in a familiar, nearby system generated great excitement in the exoplanet community, along with an initially subdued but slowly growing skepticism. Artie Hatzes first advised caution (2012) and then expressed doubt (2013) about the planet’s existence. A team including Michael Endl (2015) and Christoph Bergmann (2015) began an observational program focused on both stars in the Alpha Centauri binary to return a definitive picture supported by independent data. So far, however, they have not reported any conclusions.

In their October takedown, Rajpaul and colleagues rely on exhaustive statistical analyses instead of new observations. They show that the existing HARPS data previously interpreted as a terrestrial planet on a three-day orbit are just an artifact of the modeling approach used by Dumusque and colleagues. According to their abstract: 

The 3.24 day signal observed in the Alpha Cen B data almost certainly arises from the window function (time sampling) of the original data. We show that when stellar activity signals are removed from the RV variations, other significant peaks in the power spectrum of the window function are coincidentally suppressed, leaving behind a spurious yet apparently-significant 'ghost' of a signal that was present in the window function’s power spectrum ab initio.


In related news, Rodrigo Diaz & colleagues (herafter D15) report that, with more than a decade of HARPS radial velocity measurements in hand, they can confirm only four of the six planets proposed for HD 40307. This system includes a single Sun-like star located a bit farther away than Alpha Centauri, at a distance of 12.8 parsecs (42 light years). Notably, one of the two planets phantomized by D15’s summative analysis (HD 40307 g) was previously imagined as a “potentially habitable Super Earth” – in other words, an object with a minimum mass below 10 Earth masses (10 Mea) orbiting in the system habitable zone (Brasser et al. 2014). By now, however, it seems pretty clear that the maximum planet mass compatible with surface water in the habitable zone is closer to 5 Mea than to 10 Mea. Thus the elision of candidate g, whose minimum mass was supposed to be about 7 Mea, provokes no exobiological regrets.

Despite these two subtractions, HD 40307 persists among the classic examples of compact, low-mass multiplanet systems within 40 parsecs. According to D15, the four robustly detected planets follow circular orbits within 0.25 astronomical units (AU) of the star, and their minimum masses range from 3.6 to 8.7 Mea. A dynamical integration of system elements over half a million years indicated that the true masses of these planets could be at least twice their minimum masses without compromising long-term stability. Similar architectures are quite common in the large sample of Kepler multiplanet systems.

Meanwhile, the Sun’s back yard is getting crowded with ghosts! 


Bergmann C,.Endl M, Hearnshaw JB, Wittenmyer RA, Wright DJ. (2015) Searching for Earth-mass planets around Alpha Centauri: Precise radial velocities from contaminated spectra. International Journal of Astrobiology 14, 173-176. Abstract: 2015IJAsB..14..173B
Brasser R, Ida S, Kokubo E. (2014) A dynamical study on the habitability of terrestrial exoplanets II: The super Earth HD 40307 g. Monthly Notices of the Royal Astronomical Society 440, 3685-3700. Abstract: 2014MNRAS.440.3685B
Diaz RF, Ségransan D, Udry S, Lovis C, Pepe F, Dumusque X, and 19 others. (2015) The HARPS search for southern extra-solar planets XXXVII. Bayesian re-analysis of three systems: New super-Earths, unconfirmed signals, and magnetic cycles. Astronomy & Astrophysics, in press. Abstract: 2015arXiv151006446D
Dumusque X, Pepe F, Lovis C, Ségransan D, Sahlmann J, Benz W, Bouchy F, Mayor M, Queloz D, Santos N, Udry S. (2012) An Earth-mass planet orbiting Alpha Centauri B. Nature 491, 207-211. Abstract:
Endl M, Bergmann C, Hearnshaw J, Barnes SI, Wittenmyer RA, Ramm D, Kilmartin P, Gunn F, Brogt E. (2015) The Mt. John University Observatory search for Earth-mass planets in the habitable zone of Alpha Centauri. International Journal of Astrobiology 14, 305-312. Abstract: 2015IJAsB..14..305E
Hatzes A. (2012) Meet Our Closest Neighbour. Nature 491, 200-201.
Hatzes AP. (2013) Radial velocity detection of Earth-mass planets in the presence of activity noise: The case of Alpha Centauri Bb. Abstract: 2013ApJ...770..133H
Rajpaul V, Aigrain S, Roberts S. (2015) Ghost in the time series: No planet for Alpha Cen B. Monthly Notices of the Royal Astronomical Society, in press. Abstract: 2015arXiv151005598R

Sunday, October 4, 2015

A Different Picture of HD 219134

Figure 1. Schematic view of the six-planet architecture proposed by Vogt and colleagues for HD 219134, a small Sun-like star of spectral class K3 at a distance of only 6.53 parsecs (21 light years). Compare this model with the one proposed by Motalebi and colleagues.
A few months ago we read about the discovery of HD 219134, a remarkable new multiplanet system in the Sun’s back yard (Motalebi et al. 2015; hereafter M15). Its architecture is characterized by a long-period gas giant encircling several low-mass planets. All these objects were detected through radial velocity variations, and the innermost might also be observable in transit. In a postscript to the discovery paper, the authors noted that another group led by Stephen Vogt had also detected planetary signals from this star.

Vogts team just published a preprint detailing their findings (hereafter V15). As I anticipated, some of the new material complements and some of it conflicts with the picture presented by M15. We still see a mixed-mass architecture with a pronounced division between inner and outer systems (Figure 1). As in M15, the inner system contains only low-mass planets, all orbiting starward of the habitable zone (which for this star extends from 0.5 to 0.9 AU; see Kastings et al. 2014). The outer system contains a single gas giant whose mass places it in the “sub-Jovian desert” (Colon et al. 2015) and whose orbit places it outside the habitable zone. Both teams agree on the basic characteristics of planets b and c, and both also agree that some kind of smallish planet is present in a period just shy of 47 days. 

Table 1. Two views of the planetary system around HD 219134
Columns headed by V present findings from Vogt et al. 2015; columns headed by M present findings from Motalebi et al. 2015. Abbreviations: a = orbital semimajor axis; e = orbital eccentricity.
Nonetheless, as Table 1 shows, many results in V15 and M15 are in conflict. Here are the most salient disagreements:
  • Although both groups present essentially the same picture of the two innermost planets, their results on the outer planets differ dramatically. Both groups identify a planet with a period just shy of 47 days, but for V15 this is planet e, with a minimum mass in excess of 20 Mea, while for M15 it is planet d, with a minimum mass around 9 Mea. Vogt et al. do propose a planet d of approximately 9 Mea, but they allow it a period of only 22.8 days – about one-half the period of planet d in M15, and about one-half that of planet e in V15. 
  • The most substantial mismatch in planet characteristics involves the cool gas giant. Whereas M15 indicate a sub-Saturn mass object (planet e) with a period of about 1200 days and a semimajor axis of 2.14 AU, V15 present an object slightly more massive than Saturn (planet g) with a period of about 2200 days and a semimajor axis of 3.11 AU.
  • Vogt’s group finds two planets with periods of about 23 and 94 days, respectively, whereas M15 find no evidence for more than three planets inside 0.4 AU. Notably, planets d-f in V15 have period ratios that approximate a Laplace resonance of 1:2:4, while the period of planet g in V15 is almost twice that of planet e in M15. As an amateur who’s been following this stuff for years, I have to wonder if there’s a problem with aliasing in one or both analyses.
  • The planets in V15 have much more circular orbits than those in M15. Vogt’s group reports that while fitting the data for their inner planets (b-f), they kept the orbital eccentricities fixed at zero to ensure a dynamically stable configuration. However, they don’t extend this explanation to planet g, which in their model has an eccentricity of only 0.06, very similar Saturn’s (0.056). The corresponding value in M15 is 0.27. Nevertheless, M15 caution that the high eccentricities they found for planet d as well as planet e might stem from “un-perfect modelling of the outer signal,” and thus might be exaggerated. (One can only hope that “un-perfect” becomes “imperfect” on publication, unless one’s principal interest is amusement!)
  • M15 provide wide confidence intervals for the estimated minimum mass of each planet candidate, while V15 provide very narrow intervals (not shown). Nevertheless, the two teams’ estimates for planets b and c do not overlap; Vogt’s team finds a somewhat smaller mass for planet b and a somewhat larger mass for planet c. Much more substantial is the mismatch for the gas giant: the candidate in V15 (planet g) is 74% more massive than the candidate in M15 (planet e).
  • V15 present a system of planets whose aggregated minimum mass (156 Mea) is twice that of the system described by M15 (78 Mea). Nevertheless, both groups assume essentially the same mass for the host star (0.78 Msol in M15, 0.79 Msol in V15).  
plus ça change . . . 

All this gives me a sad sense of déjà vu. We’ve seen many similar cases of conflicting astronomical findings over the past decade, and they typically involve observations of stars that, like HD 219134, are located within 10 parsecs. The most visible of these controversies relate to Gliese 581 and Gliese 667C. Both were once proposed as rich multiplanet systems, each with six low-mass planets and at least one candidate for the coveted status of “habitable Super Earth.” Yet current data indicate that Gliese 581 has only three small planets, all hellishly hot, while Gliese 667C has only two.

This pair of systems experienced their abrupt population shrinkage in 2014, when Paul Robertson and colleagues demonstrated that the radial velocity analyses for both stars were compromised by correlated noise related to stellar activity cycles. In each case, the analytic breakthrough came along with a determination of the stellar rotation period. What have exoplanetary astronomers learned since then?

Well, M15 provide a rotational period of 42.3 days for HD 219134 and argue that none of their periodicities are related to this value. V15 assume (but did not measure) a rotational period of only ~20 days for the star, and they also argue that none of their periodicities are related to it.  However, the shorter period presented by V15 appears inconsistent with the stellar age they estimate: 12.46 billion years, which would make the star older than the spiral arm where it currently resides. Stellar spin normally slows down with age (Meibom et al. 2015). As a point of reference, our Sun rotates in about 25 days, consistent with its measured age of 4.55 billion years.

Both groups might be wrong, but it seems unlikely that both could be right.

Although I’ve often written about astronomical controversies, I wish they would just go away. To break the present impasse, I look forward to a concerted, cooperative, and international effort to obtain an authoritative perspective on HD 219134. Even though no habitable Super Earths are at stake in this case, the proposed system architecture – a cluster of likely gas dwarfs inside 0.5 AU and an undersized gas giant outside 2 AU – has huge significance for theories of system evolution. In addition, as reported by M15, HD 219134 b appears to transit its parent star, a claim with far-reaching implications that should be followed up as soon as possible.

A fascinating and difficult case indeed!

Colon KD, Morehead RC, Ford EB. (2015) Vetting Kepler planet candidates in the sub-Jovian desert with multiband photometry. Monthly Notices of the Royal Astronomical Society 452, 3001-3009. Abstract: 2015MNRAS.452.3001C 
Kasting JF, Kopparapu R, Ramirez RM, Harman CE. (2014) Remote life-detection criteria, habitable zone boundaries, and the frequency of Earth-like planets around M and late K stars. Proceedings of the National Academy of Sciences 111, 12641-12646. Abstract: 
Motalebi F, Udry S, Gillon M, Lovis C, Ségransan D, Buchhave LA, Demory BO, Malavolta L, Dressing CD, Sasselov D, et al. (2015) The HARPS-N Rocky Planet Search I. HD 219134 b: A transiting rocky planet in a multi-planet system at 6.5 pc from the Sun. Astronomy & Astrophysics, in press. Abstract: 2015arXiv150708532M  
Meibom S, Barnes S, Platais I, Gilliland R, Latham D, Mathieu R. (2015) A spin-down clock for cool stars from observations of a 2.5-billion-year-old cluster. Nature 517, 589-591. Abstract:
Robertson P, Mahadevan S, Endl M, & Roy A. (2014) Stellar activity masquerading as planets in the habitable zone of the M dwarf Gliese 581. Science 345, 440-444. Abstract: 
Robertson P, Mahadevan S. (2014) Disentangling planets and stellar activity for Gliese 667C. Astrophysical Journal Letters 793, L24. Abstract:   
Vogt SS, Burt J, Meschiari S, Butler RP, Henry GW, Wang S, Holden B, Gapp C, Hanson R, Arriagada P, Keiser S, Teske J, Laughlin G. A six-planet system orbiting HD 219134. Astrophysical Journal, in press. Abstract: 2015arXiv150907912V