A&A 542, A4 (2012) Astronomy

DOI: 10.1051/0004-6361/201218817 &


c ESO 2012 Astrophysics

The TRAPPIST survey of southern transiting planets

I. Thirty eclipses of the ultra-short period planet WASP-43 b,,
M. Gillon1 , A. H. M. J. Triaud2 , J. J. Fortney3 , B.-O. Demory4 , E. Jehin1 , M. Lendl2 , P. Magain1 , P. Kabath5 ,
D. Queloz2 , R. Alonso2 , D. R. Anderson6 , A. Collier Cameron7 , A. Fumel1 , L. Hebb8 , C. Hellier6 , A. Lanotte1 ,
P. F. L. Maxted6 , N. Mowlavi2 , and B. Smalley6

1
Universit de Lige, Alle du 6 aot 17, Sart Tilman, Lige 1, Belgium
e-mail: michael.gillon@ulg.ac.be
2
Observatoire de Genve, Universit de Genve, 51 chemin des Maillettes, 1290 Sauverny, Switzerland
3
Department of Astronomy and Astrophysics, University of California, Santa Cruz, CA 95064, USA
4
Department of Earth, Atmospheric and Planetary Sciences, Department of Physics, Massachusetts Institute of Technology,
77 Massachusetts Ave., Cambridge, MA 02139, USA
5
European Southern Observatory, Alonso de Cordova 3107, 19001 Casilla, Santiago, Chile
6
Astrophysics Group, Keele University, Staordshire ST5 5BG, UK
7
School of Physics and Astronomy, University of St. Andrews, North Haugh, Fife, KY16 9SS, UK
8
Department of Physics and Astronomy, Vanderbilt University, Nashville, TN 37235, USA
Received 13 January 2012 / Accepted 26 March 2012

ABSTRACT

We present twenty-three transit light curves and seven occultation light curves for the ultra-short period planet WASP-43 b, in addition
to eight new measurements of the radial velocity of the star. Thanks to this extensive data set, we improve significantly the parameters
of the system. Notably, the largely improved precision on the stellar density (2.41 0.08  ) combined with constraining the age
to be younger than a Hubble time allows us to break the degeneracy of the stellar solution mentioned in the discovery paper. The
resulting stellar mass and size are 0.717 0.025 M and 0.667 0.011 R . Our deduced physical parameters for the planet are
2.034 0.052 MJup and 1.036 0.019 RJup . Taking into account its level of irradiation, the high density of the planet favors an old
+0.0060
age and a massive core. Our deduced orbital eccentricity, 0.00350.0025 , is consistent with a fully circularized orbit. We detect the
emission of the planet at 2.09 m at better than 11-, the deduced occultation depth being 1560 140 ppm. Our detection of the
occultation at 1.19 m is marginal (790 320 ppm) and more observations are needed to confirm it. We place a 3- upper limit
of 850 ppm on the depth of the occultation at 0.9 m. Together, these results strongly favor a poor redistribution of the heat to the
night-side of the planet, and marginally favor a model with no day-side temperature inversion.
Key words. planetary systems stars: individual: WASP-43 techniques: photometric techniques: radial velocities

1. Introduction orbital obliquities that suggest past violent dynamical perturba-

tions (e.g. Triaud et al. 2010).
There are now more than seven hundreds planets known to or-
bit around other stars than our Sun (Schneider 2011). A signif- Not only the origin of these planets but also their fate raises
icant fraction of them are Jovian-type planets orbiting within many questions. Their relatively large mass and small semi-
0.1 AU of their host stars. Their very existence poses an inter- major axis imply tidal interactions with the host star that should
esting challenge for our theories of planetary formation and evo- lead in most cases to a slow spiral-in of the planet and to a
lution, as such massive planets could not have formed so close transfer of angular momentum to the star (e.g. Barker & Ogilvie
to their star (DAngelo et al. 2010; Lubow & Ida 2010). Long 2009; Jackson et al. 2009; Mastumura et al. 2010), the end result
neglected in favor of disk-driven migration theories (Lin et al. being the planets disruption at its Roche limit (Gu et al. 2003).
1996), the postulate that past dynamical interactions combined The timescale of this final disruption depends mostly on the
with tidal dissipation could have shaped their present orbit has timescale of the migration mechanism, the tidal dissipation ef-
raised a lot of interest recently (e.g. Fabrycky & Tremaine 2007; ficiency of both bodies, and the eciency of angular momentum
Naoz et al. 2011; Wu & Lithwick 2011), partially thanks to the losses from the system due to magnetic braking. These three pa-
discovery that a significant fraction of these planets have high rameters being poorly known, it is desirable to detect and study
in depth extreme hot Jupiters, i.e. massive planets having ex-

Based on data collected with the TRAPPIST and Euler telescopes ceptionally short semi-major axes that could be in the final stages
at ESO La Silla Observatory, Chile, and with the VLT/HAWK-I instru- of their tidal orbital decay.
ment at ESO Paranal Observatory, Chile (program 086.C-0222).

Tables 1 and 4 are available in electronic form at The WASP transit survey detected two extreme examples of
http://www.aanda.org such ultra-short period planets, WASP-18 b (Hellier et al. 2009)
 and WASP-19 b (Hebb et al. 2010), both having an orbital pe-
Photometry is only available at the CDS via anonymous ftp to
cdsarc.u-strasbg.fr (130.79.128.5) or via riod smaller than one day. Thanks to their transiting nature,
http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/A+A/542/A4 these systems were amenable for a thorough characterization
Article published by EDP Sciences A4, page 1 of 15
 A&A 542, A4 (2012)

that made possible a study of the eects of their tidal interactions eects and to optimize the observational eciency. We gener-
(Brown et al. 2011). Notably, photometric observations of some ally keep the positions of the stars on the chip within a box of
of their occultations made possible not only to probe both plan- a few pixels of side to improve the photometric precision of our
ets dayside emission spectra but also to bring strong constraints TRAPPIST time-series, thanks to a software guiding system
on their orbital eccentricity (Anderson et al. 2010; Gibson et al. deriving regularly astrometric solutions on the science images
2010; Nymeyer et al. 2011), a key parameter to assess their tidal and sending pointing corrections to the mount if needed. It could
history and energy budget. The same WASP survey has recently unfortunately not be used for WASP-43, because the star lies in
announced the discovery of a third ultra-short period Jovian a sky area that is not covered by the used astrometric catalog
planet called WASP-43 b (Hellier et al. 2011b, hereafter H11). (GSC1.1). This translated into slow drifts of the stars on the chip,
Its orbital period is 0.81 d, the only hot Jupiter having a smaller the underlying cause being the imperfection of the telescope po-
period being WASP-19 b (0.79 d). Furthermore, it orbits around lar alignment. The amplitudes of those drifts on the total duration
a very cool K7-type dwarf that has the lowest mass among all of the runs were ranging between 15 and 85 pixels in the right
the stars orbited by a hot Jupiter (0.58 0.05 M , T e = 4400 ascension direction and between 5 and 75 pixels in the declina-
200 K, H11), except for the recently announced M0 dwarf KOI- tion direction. Table 1 presents the logs of the observations. The
254 (0.59 0.06 M , T e = 3820 90 K, Johnson et al. 2012). first of these 20 transits was presented in H11.
Nevertheless, H11 presented another plausible solution for the After a standard pre-reduction (bias, dark, flatfield correc-
stellar mass that is significantly larger, 0.71 0.05 M . This tion), the stellar fluxes were extracted from the images using the
degeneracy translates into a poor knowledge of the physical pa- IRAF/DAOPHOT3 aperture photometry software (Stetson 1987).
rameters of the system. For each transit, several sets of reduction parameters were tested,
With the aim to improve the characterization of this in- and we kept the one giving the most precise photometry for the
teresting ultra-short period planet, we performed an intense stars of similar brightness as WASP-43. After a careful selection
ground-based photometric monitoring of its eclipses (transits of reference stars, dierential photometry was then obtained.
and occultations), complemented with new measurements of the The resulting light curves are shown in Figs. 1 and 2.
radial velocity (RV) of the star. These observations were car-
ried out in the frame of a new photometric survey based on the
60cm robotic telescope TRAPPIST1 (TRAnsiting Planets and 2.2. Euler Gunn-r filter transit photometry
PlanetesImals Small Telescope; Gillon et al. 2011a; Jehin et al. Three transits of WASP-43 b were observed in the Gunn-r filter
2011). The concept of this survey is the intense high-precision (e = 620.4 0.5 nm) with the EulerCAM CCD camera at
photometric monitoring of the transits of southern transiting sys- the 1.2-m Euler Swiss telescope, also located at ESO La Silla
tems, its goals being (i) to improve the determination of the Observatory. EulerCAM is a nitrogen-cooled 4k 4k CCD cam-
physical and orbital parameters of the systems, (ii) to assess the era with a 15 15 field of view (pixel scale = 0.23 ). Here too,
presence of undetected objects in these systems through vari- the telescope was slightly defocused to optimize the photomet-
ability studies of the transit parameters, and (iii) to measure ric precision. The mean exposure time was 85 s. The stars were
or put an upper limit on the very-near-IR thermal emission of kept approximately on the same pixels, thanks to a software
the most highly irradiated planets to constrain their atmospheric guiding system similar to TRAPPISTs but using the UCAC3
properties. We complemented the data acquired in the frame of catalog. The calibration and photometric reduction procedures
this program for the WASP-43 system by high-precision occul- were similar to the ones performed on the TRAPPIST data. The
tation time-series photometry gathered in the near-IR with the logs of these Euler observations are shown in Table 1, while the
VLT/HAWK-I instrument (Pirard et al. 2004; Casali et al. 2006) resulting light curves are visible in Fig. 2. Notice that the first of
in our program 086.C-0222. We present here the results of the these Euler transits was presented in H11.
analysis of this extensive data set. Section 2 below presents our
data. Their analysis is described in Sect. 3. We discuss the ac-
quired results and drawn inferences about the WASP-43 system 2.3. TRAPPIST z filter occcultation photometry
in Sect. 4. Finally, we give our conclusions in Sect. 5.
Five occultations of WASP-43 b were observed with TRAPPIST
in a Sloan z filter (e = 915.9 0.5 nm). Their logs are pre-
sented in Table 1. The mean exposure time was 40 s. The cali-
2. Data
bration and photometric reduction of these occultation data were
2.1. TRAPPIST I+z filter transit photometry similar to the ones of the transits. Figure 3 shows the resulting
light curves with their best-fit models.
We observed 20 transits of WASP-43 b with TRAPPIST and its
thermoelectrically-cooled 2k 2k CCD camera with a field of
view of 22 22 (pixel scale = 0.65 ). All the 20 transits were 2.4. VLT/HAWK-I 1.19 and 2.09 m occultation photometry
observed in an Astrodon I+z filter that has a transmittance
>90% from 750 nm to beyond 1100 nm2 , the red end of the eec- We observed two occultations of WASP-43 b with the cryogenic
tive bandpass being defined by the spectral response of the CCD. near-IR imager HAWK-I at the ESO Very Large Telescope in
This wide red filter minimizes the eects of limb-darkening our program 086.C-0222. HAWK-I is composed of four Hawaii
and dierential atmospheric extinction while maximizing stel- 2RG 2048 2048 pixels detectors (pixel scale = 0.106 ), its to-
lar counts. Its eective wavelength for T e = 4400 200 K tal field of view on the sky being 7.5 7.5 . We choose to
is e = 843.5 1.2 nm. The mean exposure time was 20 s. observe the occultations within the narrow band filters NB2090
The telescope was slightly defocused to minimize pixel-to-pixel
3
IRAF is distributed by the National Optical Astronomy Observatory,
1
See http://www.ati.ulg.ac.be/TRAPPIST which is operated by the Association of Universities for Research
2
See http://www.astrodon.com/products/filters/ in Astronomy, Inc., under cooperative agreement with the National
near-infrared/ Science Foundation.

A4, page 2 of 15
 M. Gillon et al.: Thirty eclipses of the ultra-short period planet WASP-43 b

Fig. 1. Left: WASP-43 b transit photometry (12 first TRAPPIST transits) used in this work, binned per 2 min, period-folded on the best-fit transit
ephemeris deduced from our global MCMC analysis (see Sect. 3.3), and shifted along the y-axis for clarity. The best-fit baseline+transit models
are superimposed on the light curves. The fifth and ninth models (from the top) show some wiggles because of their position-dependent terms.
Right: best-fit residuals for each light curve binned per interval of 2 min.

( = 2.095 m, width = 0.020 m) and NB1190 ( = 1.186 m, The observation of the first occultation (NB2090 filter) took
width = 0.020 m), respectively. The small width of these cos- place on 2010 Dec. 9 from 5h37 to 9h07 UT. Atmospheric con-
mological filters minimizes the eect of dierential extinction. ditions were very good, with a stable seeing and extinction.
Furthermore, they avoid the largest absorption and emission Airmass decreased from 2.1 to 1.05 during the run. Each of the
bands that are present in J and K bands, reducing thus signif- 185 exposures was composed of 17 integrations of 1.7 s each (the
icantly the correlated photometric noise caused by the complex minimum integration time allowed for HAWK-I). We choose to
spatial and temporal variations of the background due to the vari- do not apply a jitter pattern. Indeed, the background contribution
ability of the atmosphere. As this correlated noise is the main in the photometric aperture is small enough to ensure that the
precision limit for ground-based near-IR time-series photome- low-frequency variability of the background cannot cause corre-
try, the use of these two narrow-band filters optimizes the pho- lated noises with an amplitude larger than a few dozens of ppm
tometric quality of the resulting light curves. This is especially in our light curves, making the removal of a background im-
important in the context of the challenging measurement of the age unnecessary. Furthermore, staying on the same pixels dur-
emission of exoplanets. ing the whole run allows minimizing the eects of interpixel

A4, page 3 of 15
 A&A 542, A4 (2012)

Fig. 2. Left: WASP-43 b transit photometry (8 last TRAPPIST transits + the 3 Euler transits) used in this work, binned per 2 min, period-folded
on the best-fit transit ephemeris deduced from our global MCMC analysis (see Sect. 3.3), and shifted along the y-axis for clarity. The best-fit
baseline+transit models are superimposed on the light curves (red = TRAPPIST I + z filter; green = Euler Gunn-r filter). The third, ninth and
tenth models (from the top) show some wiggles because of their position-dependent terms. Right: best-fit residuals for each light curve binned per
interval of 2 min.

sensitivity inhomogeneity (i.e. the imperfect flat field). The anal- while the extinction was stable during the first part of the run
ysis of HAWK-I calibration frames showed us that the detector and slightly variable during the second part. No defocus was
is nearly linear up to 1012 kADU. The peak of the target im- applied, the peak of the target being in the linear part of the
age was above 10 kADU in the first images, so a slight defocus detector dynamic in all images. Airmass decreased from 1.37
was applied to keep it below this level during the rest of the run. to 1.03, then increased to 1.13 during the run. Here too, no jitter
The mean full-width at half maximum of the stellar point-spread pattern was applied. Each of the 241 exposures was composed
function (PSF) was 7.3 pixels = 0.77 , its standard deviation for of 17 integrations of 1.7 s.
the whole run being 0.57 pixels = 0.06 . The pointing was care- After a standard calibration of the images (dark-subtraction,
fully selected to avoid cosmetic defects on WASP-43 or on the flatfield correction), a cosmetic correction was applied. This cor-
comparison stars. rection was done independently for each image and based on an
The second occultation (NB1190 filter) was observed on automatic detection of the stars followed by a detection of out-
2011 Jan. 9 from 4h57 to 9h27 UT. The seeing varied strongly lier pixels. This latter was based on a comparison of the value of
during the all run (mean value in our images =0.76 , with a stan- each pixel to the median value of the eight adjacent pixels. For
dard deviation =0.19 , minimum =0.51 , maximum =1.22 ) a pixel within a stellar aperture, a detection threshold of 50-
A4, page 4 of 15
 M. Gillon et al.: Thirty eclipses of the ultra-short period planet WASP-43 b

Fig. 3. Left: WASP-43 b occultation photometry, binned per intervals of 2 min, period-folded on the best-fit transit ephemeris deduced from our
global MCMC analysis (see Sect. 3.3), and shifted along the y-axis for clarity. The best-fit baseline+occultation models are superimposed on the
light curves (blue = TRAPPIST z -filter; green = VLT/HAWK-I NB1190 filter; red = VLT/HAWK-I NB2090 filter). Right: same light curves
divided by their best-fit baseline models. The corresponding best-fit occultation models are superimposed.

was used, while it was 5- for the background pixels. Outlier 2011 Feb. 02 and March 12, the integration time being 30 min
pixels had their value replaced by the median value of the ad- for all of them. RVs were computed from the calibrated spec-
jacent pixels. At this stage, aperture photometry was performed tra by weighted cross-correlation (Baranne et al. 1996) with a
with DAOPHOT for the target and comparison stars, and dier- numerical spectral template. They are shown in Table 2. We an-
ential photometry was obtained for WASP-43. Table 1 presents alyzed these new RVs globally with the RVs presented in H11
the logs of our HAWK-I observations, while Fig. 3 shows the and with our eclipse photometry (see Sect. 3.3).
resulting light curves with their best-fit models.
3. Data analysis
2.5. Euler/CORALIE spectra and radial velocities
Our analysis of the WASP-43 data was divided in two steps.
We gathered eight new spectroscopic measurements of WASP- First, we performed individual analyses of the 30 eclipse light
43 with the CORALIE spectrograph mounted on Euler. curves, determining independently for each light curves the
The spectroscopic measurements were performed between corresponding eclipse and physical parameters. The aim of
A4, page 5 of 15
 A&A 542, A4 (2012)

Table 2. CORALIE radial-velocity measurements for WASP-43. Table 3. Prior PDF used in this work for the quadratic limb-darkening
coecients.
Time RV RV BS
(BJDTDB 2 450 000) (km s1 ) (m s1 ) (km s1 ) Filter u1 u2
5594.848273 3.921 21 0.021 I+z N(0.440, 0.0352 ) N(0.180, 0.0252 )
5604.730390 4.149 19 0.052 Gunn-r N(0.625, 0.0152 ) N(0.115, 0.0102 )
5605.677373 3.862 15 0.044
5626.663879 4.123 18 0.035
5627.670365 3.759 21 0.039
5628.713064 3.059 18 0.061 the transit width (from first to last contact) W;
5629.686781 3.388 17 0.009 the parameter K2 = K 1 e2 P1/3 , K being the RV orbital
5632.715791 3.126 22 0.115 semi-amplitude;
the parameters v sin I cos and v sin I sin , v sin I
Notes. BS = bisector spans.
and being respectively the projected rotational velocity of
the star and the projected angle between the stellar spin axis
this first step was searching for potential variability among the and the planets orbital axis.
eclipse parameters (Sect. 3.2). We then performed a global anal-
ysis of the whole data set, including the RVs, with the aim Uniform or normal prior PDFs can be assumed for the jump and
to obtain the strongest constraints on the system parameters physical parameters of the system. Negative values are not al-
(Sect. 3.3). lowed for dF, dFocc , b , P, T 0 , W and K2 .
Two limb-darkening laws are implemented in our code,
quadratic (two parameters) and non-linear (four parameters).
3.1. Method and models For each photometric filter, values and error bars for the limb-
Our data analysis was based on the most recent version of darkening coecients are interpolated in Claret & Bloemens
our adaptive Markov chain Monte-Carlo (MCMC) algorithm tables (2011) at the beginning of the analysis, basing on input
(see Gillon et al. 2010, and references therein). To summarize, values and error bars for the stellar eective temperature T e ,
MCMC is a Bayesian stochastic simulation algorithm designed metallicity [Fe/H] and gravity log g. For the quadratic law, the
to deduce the posterior probability distribution functions (PDFs) two coecients u1 and u2 are allowed to float in the MCMC,
for the parameters of a given model (e.g. Gregory 2005; Carlin using as jump parameters not these coecients themselves but
& Louis 2008). Our implementation of the algorithm assumes the combinations c1 = 2 u1 + u2 and c2 = u1 2 u2 to mini-
as model for the photometric time-series the eclipse model of mize the correlation of the obtained uncertainties (Holman et al.
Mandel & Agol (2002) multiplied by a baseline model aim- 2006). In this case, the theoretical values and error bars for u1
ing to represent the other astrophysical and instrumental mech- and u2 deduced from Clarets tables can be used in normal prior
anisms able to produce photometric variations. For the RVs, the PDFs. In all our analyses, we assumed a quadratic law and let u1
model is based on Keplerian orbits added to a model for the stel- and u2 float under the control of the normal prior PDFs deduced
lar and instrumental variability. Our global model can include from Claret & Bloemens tables. For the non-standard I + z fil-
any number of planets, transiting or not. For the RVs obtained ter, the modes of the normal PDFs for u1 and u2 were taken as
during a transit, a model of the Rossiter-McLaughlin eect is the averages of the values interpolated from Clarets tables for
also available (Gimnez 2006). Comparison between two mod- the standard filters Ic and z , while the errors were computed
els can be performed based on their Bayes factor, this latter be- as the quadratic sums of the errors for these two filters. The
ing the product of their prior probability ratio multiplied by their prior PDFs deduced for WASP-43 are shown in Table 3. They
marginal likelihood ratio. The marginal likelihood ratio of two were computed for T e = 4400 200 K, log g = 4.5 0.2 and
given models is estimated from the dierence of their Bayesian [Fe/H] = 0.05 0.17 (H11).
Information Criteria (BIC; Schwarz 1978) which are given by At the first step of the MCMC, the timings of the measure-
the formula: ments are passed to the BJDTDB time standard, following the
recommendation of Eastman et al. (2010) that outlined that the
BIC = 2 + k log(N) (1) commonly-used BJDUTC time standard is not practical for high-
precision timing monitoring as it drifts with the addition of one
where k is the number of free parameters of the model, N is the
leap second roughly each year.
number of data points, and 2 is the smallest chi-square found
At each step of the Markov chains, the stellar density is
in the Markov chains. From the BIC derived for two models, the
derivedfrom Keplers third law and the jump parameters dF,
corresponding marginal likelihood ratio is given by eBIC/2 .
b , W, e cos and e sin (Seager & Mallen-Ornelas 2003;
For each planet, the main parameters that can be randomly
Winn 2010). Using as input values the resulting stellar density
perturbed at each step of the Markov chains (called jump param-
and values for T e and [Fe/H] drawn from their prior distribu-
eters) are
tions, a modified version of the stellar mass calibration law de-
the planet/star area ratio dF = (Rp /R )2 , Rp and R being duced by Torres et al. (2010) from well-constrained detached
respectively the radius of the planet and the star; binary systems (see Gillon et al. 2011b, for details) is used to de-
the occultation depth(s) (one per filter) dFocc ; rive the stellar mass. The stellar radius is then derived from the
the parameter b = a cos ip /R which is the transit impact pa- stellar density and mass. At this stage, the physical parameters
rameter in case of a circular orbit, a and ip being respectively of the planet (mass, radius, semi-major axis) are deduced from
the semi-major axis and inclination of the orbit; the jump parameters and stellar mass and radius. Alternatively,
the orbital period P; a value and error for the stellar mass can be imposed at the
the time of minimumlight T 0 (inferior
conjunction); start of the MCMC analysis, In this case, a stellar mass value is
the two parameters e cos and e sin , e being the or- drawn from the corresponding normal distribution at each step
bital eccentricity and being the argument of periastron; of the Markov chains, allowing the code to deduce the other
A4, page 6 of 15
 M. Gillon et al.: Thirty eclipses of the ultra-short period planet WASP-43 b

physical parameters. We preferred here to use this second op- residuals at each step of the chains, thanks to their linear nature
tion, as WASP-43 was potentially lying at the lower edge of the in the baseline functions (Bakos et al. 2009; Gillon et al. 2010).
mass range for which the calibration law of Torres et al. is valid,
0.6 M .
3.2. Individual analysis of the eclipse time-series
If measurements for the rotational period of the star and for
its projected rotational velocity are available, they can be used As mentioned above, we first performed an independent analysis
in addition to the stellar radius values deduced at each step of of the transits and occultations aiming to elect the optimal model
the MCMC to derive a posterior PDF for the inclination of the for each light curve and to assess the variability and robustness
star (Watson et al. 2010; Gillon et al. 2011b). We did not use of the derived parameters. For all these analyses, the orbital pe-
this option here despite that the rotational period of the star riod and eccentricity were kept respectively to 0.813475 days
was determined from WASP photometry to be 15.6 0.4 days and zero (H11), and the normal distribution N(0.58, 0.052) M
(H11), because the V sin I measurement presented by H11 (H11) was used as prior PDF for the stellar mass. For the transit
(4.0 0.4 km s1 ) was presented by these authors as proba- light curves, the jump parameters were dF, b , W and T 0 . For the
bly aected by a systematic error due to additional broadening occultations, the only jump parameter was the occultation depth
of the lines. dFocc , the other system parameters being drawn at each step of
Several chains of 100 000 steps were performed for each the MCMC from normal distributions deduced from the values +
analysis, their convergences being checked using the statistical errors derived in H11.
test of Gelman & Rubin (1992). After election of the best model Table 1 presents the baseline function selected for each light
for a given light curve, a preliminary MCMC analysis was per- curve, the derived factors w , r , CF, and the standard devia-
formed to estimate the need to rescale the photometric errors. tion of the best-fit residuals, unbinned and binned per 120 s.
The standard deviation of the residuals was compared to the These results allow us to assess the photometric precision of the
mean photometric errors, and the resulting ratios w were stored. used instruments. The TRAPPIST data show mean values for w
w represents the under- or overestimation of the white noise of and r very close to 1, the I + z data having w = 1.03 and
each measurement. On its side, the red noise present in the light r = 1.05, while the z data have w = 0.98 and r = 1.11.
curve (i.e. the inability of our model to represent perfectly the This suggests that the photometric errors of each measurement
data) was taken into account as described by Gillon et al. (2010), are well approximated by a basic error budget (photon, read-out,
i.e. a scaling factor r was determined from the standard devia- dark, background, scintillation noises), and that the level of cor-
tions of the binned and unbinned residuals for dierent binning related noise in the data is small. Furthermore, we notice that
intervals ranging from 5 to 120 min, the largest values being kept most TRAPPIST light curves are well modeled by the mini-
as r . At the end, the error bars were multiplied by the correc- mal model, i.e. the sum of an eclipse model and a quadratic
tion factor CF = r w . For the RVs, a jitter noise could be trend in time. Only one TRAPPIST transit light curve requires
added quadratically to the error bars after the election of the best additional terms in x and y, while one occultation light curve ac-
model, to equal the mean error with the standard deviation of the quired when the moon was close to full requires a linear term
best-fit model residuals. In this case, it was unnecessary. in background. The mean photometric errors per 2 min inter-
Our MCMC code can model very complex trends for the vals can also be considered as very good for a 60 cm telescope
photometric and RV time-series, with up to 46 parameters for monitoring a V = 12.4 star: 0.11% and 0.15% in the I + z and
each light curve and 17 parameters for each RV time-series. Our z filters, respectively, which is similar to the mean photometric
strategy here was first to fit a simple orbital/eclipse model and to error of Euler data, 0.12%. Euler data show also small mean val-
analyze the residuals to assess any correlation with the external ues of w = 1.27 and r = 1.05. Their modeling requires PSF
parameters (PSF width, time, position on the chip, line bisector, position terms for 2 out of 3 eclipses, despite the good sampling
etc.), then to use the Bayes factor as indicator to find the optimal of the PSF and the active guiding system keeping the stars nearly
baseline function for each time-series, i.e. the model minimiz- on the same pixels. This could indicate that the flatfields quality
ing the number of parameters and the level of correlated noise in is perfectible.
the best-fit residuals. For ground-based photometric time-series, For the first HAWKI light curve, taken in the NB2090 filter,
we did not use a model simpler than a quadratic polynomial in we notice that we have to account for a ramp eect (the log(t)
time, as several eects (color eects, PSF variations, drift on term) similar to the well-documented sharp variation of the ef-
the chip, etc.) can distort slightly the eclipse shape and can thus fective gain of the Spitzer/IRAC detector at 8 m (e.g. Knutson
lead to systematic errors on the deduced transit parameters. This et al. 2008), and also for a dependance of the measured flux with
is especially important to include a trend in the baseline model the exact position of the PSF center on the chip. This position
for transit light curve with no out-of-transit data before or af- eect could be decreased by spreading the flux on more pixels
ter the transit, and/or with a small amount of out-of-transit data. (defocus), but this would also increase the backgrounds contri-
Having a rather small amount of out-of-transit data is quite com- bution to the noise budget, potentially bringing not only more
mon for ground-based transit photometry, as the target star is white noise but also some correlation of the measured counts
visible under good conditions (low airmass) during a limited du- with the variability of the local thermal structure. The photomet-
ration per night. In such conditions, an eclipse model can have ric quality reached by these NB2090 data (r = 1.09, mean error
enough degrees of freedom to compensate for a small-amplitude of 360 ppm per 2 min time interval), can be judged as excellent.
trend in the light curve, leading to an excellent fit but also to bi- For the NB1190 HAWKI data, we had to include a dependance
ased results and overoptimistic error bars. Allowing the MCMC in the PSF width in the model. This is not surprising, consider-
to twist slightly the light curve with a quadratic polynomial in ing the large variability of the seeing during the run. We also no-
time compensates, at least partially, for this eect. tice for these data that our error budget underestimated strongly
For the RVs, our minimal baseline model is a scalar V repre- the noise of the measurements (w = 2.22), suggesting an un-
senting the systemic velocity of the star. It is worth noticing that accounted noise source. Still, the reached photometric quality
most of the baseline parameters are not jump parameters in the remains excellent (r = 1, mean error of 470 ppm per 2 min
MCMC, they are deduced by least-square minimization from the time interval).
A4, page 7 of 15
 A&A 542, A4 (2012)

-1

-2

0 50 100 150 200 250

Epoch

-1

-2

0 50 100 150 200 250

Epoch

Fig. 4. Correlation diagrams for the transit parameters deduced from the Fig. 5. Top: observed minus calculated transit timings obtained from the
individual MCMC analysis of the 23 transit light curves. Top left: transit individual analysis of the transit light curves as a function of the transit
depth vs. transit impact parameter. Top right: transit duration vs. transit epoch. The filled black and open red symbols correspond, respectively,
impact parameter. Bottom left: transit duration vs. transit depth. Bottom to the TRAPPIST and Euler light curves. Bottom: same but deduced
right: TTV (observed minus calculated transit timing) vs. transit dura- from the global analysis of all transits.
tion. The filled black and open red symbols correspond, respectively, to
the TRAPPIST and Euler light curves.

such low-amplitude structures, so we do not reject this hypothe-

Table 4 presents for each eclipse light curve the values and sis. On their side, unocculted spots should have a negligible im-
errors deduced for the jump parameters. Several points can be pact on the measured transit depths, considering the low ampli-
noticed. tude of the rotational photometric modulation detected in WASP
data (6 1 mmag). Still, they could alter the slope of the pho-
The emission of the planet is clearly (10-) detected at tometric baseline and make it more complex. We represent this
2.09 m. At 1.19 m, it is barely detected (2.3-). It is baseline as an analytical function of several external parameters
not detected in any of the z -band light curves. in our MCMC simulations, but the unavoidable inaccuracy of the
The transit shape parameters b , W, and dF, show scatters chosen baseline model can also bias the derived results. As de-
slightly larger than their average error, the corresponding ra- scribed in Sect. 2.1, no pointing corrections were applied during
tio being, respectively, 1.24, 1.26, and 1.54. Furthermore, the TRAPPIST runs because of a stellar catalog problem. Even
these parameters show significant correlation, as can be seen if the resulting drifts did not introduce clear correlations of the
in Fig. 4. measured fluxes with PSF positions, they could have slightly af-
Fitting a transit ephemeris by linear regression with the fected the shape of the transits. These considerations reinforce
measured transit timings shown in Table 4, we ob- the interest of performing global analysis of extensive data sets
tained T (N) = 2 455 528.868289( 0.000072) + N (i.e. many eclipses) in order to minimize systematic errors and
0.81347728( 0.00000060) BJDTDB . Table 4 shows (last to reach high accuracies on the derived parameters.
column) the resulting transit timing residuals (observed mi-
nus computed, OC). Figure 5 (upper panel) shows them as
a function of the transit epochs. The scatter of these timing 3.3. Global analysis of photometry and radial velocities
residuals is 2.1 times larger than the mean error, 18 s. No
clear pattern is visible in the OC diagram. The global MCMC analysis of our data set was divided in three
steps. For each step of the
analysis,the MCMC jump parameters
The apparent variability of the four transit parameters is present were dF, W, b , T 0 , P, e cos , e sin , three dFocc (for the
in both TRAPPIST and Euler results. The correlation of the de- z , NB1190, and NB2090 filters), K2 , and the limb-darkening co-
rived transit parameters is not consistent with actual variations ecients c1 and c2 for the I + z and Gunn-r filters. No model for
of these parameters and favors biases from instrumental and/or the Rossiter-McLaughlin was included in the global model, as no
astrophysical origin. This apparent variability of the transit pa- RV was obtained at the transit phase. For each light curve, we as-
rameters could be explained by the variability of the star itself. sumed the same baseline model than for the individual analysis.
Indeed, WASP-43 is a spotted star (H11), and occulted spots The assumed baseline for the RVs was a simple scalar (systemic
(and faculae) can alter the observed transit shape and bias the velocity V ).
measured transit parameters (e.g. Huber et al. 2010; Berta et al. In a first step, we performed a chain of 100 000 steps with
2011). We do not detect clear spot signatures in our light curves, the aim to redetermine the scaling factors of the photometric er-
but our photometric precision could be not high enough to detect rors. The deduced values are shown in Table 1. It can be noticed
A4, page 8 of 15
 M. Gillon et al.: Thirty eclipses of the ultra-short period planet WASP-43 b

that the mean r for the 20 transits observed by TRAPPIST

in I + z filter is now 1.52, for 1.05 after the individual analysis
of the lightcurves. We notice the same tendency for the Euler
Gunn-r transits: r goes up from 1.05 to 1.42. Considering
the apparent variability of the transit parameters deduced from
the individual analysis of the light curves and their correlations,
our interpretation of this increase of the r is that a part of the
correlated noise (from astrophysical source or not) of a transit
light curve can be swallowed in the transit+baseline model,
leading to a good fit in terms of merit function but also to biased
derived parameters. By relying on the assumption that all tran-
sits share the same profile, the global analysis allows to better
separate the actual transit signal from the correlated noises of
similar frequencies, leading to larger r values. It is also worth
noticing that the r for the transits, 1.50, is larger than the one
for the occultations, 1.12, which is consistent with the crossing
of spots by the planet during transit, but can also be explained
by the much larger number of transits than occultations.
In a second step, we used the updated error scaling factors
and performed a MCMC of 100 000 steps to derive the marginal-
ized PDF for the stellar density  . This PDF was used as input
to interpolate stellar tracks. As in Hebb et al. (2009), we used
1/3
 , mostly for practical reasons. Other inputs were the stellar
eective and metallicity derived by H11, T e = 4400 200 K Fig. 6. Main panel: modifield Herztsprung-Russell diagram showing the
and [Fe/H] = 0.05 0.17 dex. Those parameters were as- posterior PDF of 1/3 as a function of T e after interpolating with the

sumed distributed as a Gaussian. For each 1/3  value from the Geneva stellar evolution tracks for stars on the main sequence. Masses
MCMC, a random Gaussian value was drawn for stellar param- are indicated and correspond to the bold tracks. The solid, dashed and
eters. That point was placed among the Geneva stellar evolu- dotted lines correspond to the 1-, 2- and 3- contours, respectively.
tion tracks (Mowlavi et al. 2012) that are available in a grid fine Secondary panels: marginalised PDF of 1/3  and T e . In both panels,
enough to allow an MCMC to wander without computational there are three histograms: dotted: input distribution; gray: output dis-
systematic eects; a mass and age were interpolated. If the point tribution; black: output distribution for ages <12 Gyr.
fell o the tracks, it was discarded. Because of the way stars be-
have at these small masses, within the (1/3 , T e , [Fe/H]) space,
the parameter space over which the tracks were interpolated cov- analysis, we kept fixed the parameters T 0 and P to the values
ered stellar ages well above a Hubble time. A restriction on the shown in Table 5, and we added a timing oset as jump param-
age was therefore added. Only taking ages <12 Gyr, we obtained eter for each transit. The orbit was assumed to be circular. The
the distribution as shown in Fig. 6, refining the stellar parame- other jump parameters were dF, W, b , and the limb-darkening
ters to T e = 4520 120 K and [Fe/H] = 0.01 0.12 dex. coecients c1 and c2 for both filters. The resulting transit tim-
The stellar mass is estimated to M = 0.717 0.025 M . ings and their errors are shown in Table 6. Fitting a transit
The output distribution is very close to Gaussian as displayed ephemeris by linear regression with these new derived transit
timings, we obtained T (N) = 2 455 528.868227( 0.000078) +
in Fig. 6. In addition the output 1/3
 distribution has not been N 0.81347764( 0.00000065) BJDTDB . Table 6 shows (last
aected. Thus, our strongly improved precision on the stellar column) the resulting transit timing residuals (observed minus
density combined with a constraint on the age (to be younger computed, OC). Figure 5 (lower panel) shows them as a func-
than a Hubble time) allows us to reject the main stellar solution tion of the transit epochs. The scatter of these timing residuals
proposed by H11 (M = 0.58 0.05 M ). Because the allowed is now 1.8 times larger than the mean error, 19 s. Comparing the
(1/3
 , T e , [Fe/H]) space within the tracks still covers quite a errors on the timings derived from individual and global analysis
large area, it is impossible at the moment to constrain the stellar of the transit photometry, one can notice that the better constraint
age. The process is described in Triaud (2011, Ph.D. Thesis). on the transit shape for the global analysis improves the error of
We then performed a third MCMC step, using as prior distri- a few timings, but that most of them have a slightly larger error
butions for M , T e , and [Fe/H], the normal distributions match- because of the better separation of the actual transit signal from
ing the results of our stellar modeling, i.e. N(0.717, 0.0252) M , the red noise (see above).
N(4520, 1202) K, and N(0.01, 0.122) dex, respectively. The
best-fit photometry and RV models are shown respectively in
Figs. 7 and 8, while the derived parameters and 1- error bars 4. Discussion
are shown in Table 5.
4.1. The physical parameters of WASP-43 b

Global analysis of the 23 transits with free timings Comparing our results for the WASP-43 system with the ones
presented in H11, we notice that our derived parameters agree
As a complement to our global analysis of the whole data set, we well with the second solution mentioned in H11, while being sig-
performed a global analysis of the 23 transit light curves alone. nificantly more precise. WASP-43 b is thus a Jupiter-size planet,
The goal here was to benefit from the strong constraint brought twice more massive than our Jupiter, orbiting at only 0.015 AU
on the transit shape by the 23 transits to derive more accurate from a 0.72 M main-sequence K-dwarf. Despite having the
transit timings and to assess the periodicity of the transit. In this smallest orbital distance among the hot Jupiters, WASP-43 b is
A4, page 9 of 15
 A&A 542, A4 (2012)

Fig. 7. Left: TRAPPIST I + z (top) and Euler Gunn-r (bottom) transit photometry period-folded on the best-fit transit ephemeris from the global
MCMC analysis, corrected for the baseline and binned per 2 min intervals, with the best-fit transit models over-imposed. Right: same for the
occultation photometry obtained by TRAPPIST in z filter (top), and by VLT/HAWK-I in NB1190 (middle) and NB2090 (bottom) filters, except
that the data points are binned per 5 min intervals.

2.5

2.0
Radius [RJup]

1.5

1.0

Fig. 8. Euler/CORALIE RV measurements period-folded on the best-fit

transit ephemeris from the global MCMC analysis, corrected for the 0.5
systemic RV, with the best-fit orbital model overimposed.

0.0
1011 1010 109 108 107 106 105
Incident flux [erg s-1 cm-2]
far from being the most irradiated known exoplanet, because of
the small size and temperature of its host star. Assuming a solar- Fig. 9. Planetary radii as a function of incident flux. WASP-43 b is
twin host star, its incident flux 9.6 108 erg s1 cm2 would shown as a black square. gray filled circles are Kepler planetary candi-
correspond to a P = 2.63 d orbit (a = 0.0374 AU), i.e. to a rather dates (see Demory & Seager 2011). Transiting giant planets previously
typical hot Jupiter. With a radius of 1.04 RJup , WASP-43 b lies published, and mostly from ground-based surveys, are shown as gray
toward the bottom of the envelope described by the published triangles. The relevant parameters Rp , Rs , T e and a have been drawn
planets in the Rp vs. incident flux plane (Fig. 9). Taking into from http://www.inscience.ch/transits on August 29, 2011.
account its level of irradiation, the high density of the planet fa-
vors an old age and a massive core under the planetary structure
models of Fortney et al. (2010). Our results are consistent with a be a planet exceptionally close to its final tidal disruption, unlike
circular orbit, and we can put a 3- upper limit <0.03 to the or- WASP-19 b that orbits at only 1.2 aR (Hellier et al. 2011a).
bital eccentricity. Despite its very short period, WASP-43 b has
a semi-major axis approximatively twice larger than its Roche
limit aR , which is typical for hot Jupiters (Ford & Rasio 2006; 4.2. The atmospheric properties of WASP-43 b
Mastumura et al. 2010). As noted by these authors, the inner We have modeled the atmosphere of WASP-43 b using the meth-
edge of the distribution of most hot Jupiters at 2 aR favors ods described in Fortney et al. (2005, 2008) and Fortney &
migrational mechanisms based on the scattering of planets on Marley (2007). In Fig. 10, we compare the planet-to-star flux
much wider orbit and the subsequent tidal shortening and circu- ratio data to three atmosphere models. The coldest model (or-
larization of their orbits. Thus, WASP-43 b does not appear to ange) uses a dayside incident flux decreased by 1/2 to simulate
A4, page 10 of 15
 M. Gillon et al.: Thirty eclipses of the ultra-short period planet WASP-43 b

Table 5. Median and 1- limits of the posterior marginalized PDFs Table 6. Median and 1- errors of the posterior PDFs deduced for the
obtained for the WASP-43 system derived from our global MCMC timings of the transits from their global analysis.
analysis.
Epoch Filter T tr OC
MCMC Jump parameters [BJDTDB -2 450 000] [min]
+0.024
Planet/star area ratio (Rp /R )2 [%] 2.5420.025 +0.00047
11 I+z 5537.816730.00046 0.36 0.68
b = a cos ip /R [R ] 0.656 0.010
+0.0055 22 I+z 5546.76494 0.00022 0.30 0.32
Transit width W [h] 1.20890.0050
+0.00015
T 0 2 450 000 [BJDTDB ] 5726.54336 0.00012 27 I+z 5550.832190.00016 0.10 0.23
Orbital period P [d] 0.81347753 0.00000071 38 I+z 5559.78033 0.00021 0.07 0.30
RV K2 [m s1 d1/3 ] +5.1
511.55.0 38 Gunn-r +0.00016
5559.780780.00017 0.58 0.24
+0.022
e cos 0.0200.023 43 I+z 5563.84771 0.00022 0.08 0.32
+0.066
e sin 0.0250.064 49 I+z 5568.72836 0.00015 0.39 0.22
+190
dFocc,z [ppm] 210130 , <850 (3-) 59 I+z +0.00016
5576.863800.00015 0.57 0.23
dFocc, NB1190 790+320
310 , <1700 (3-) 65 I+z 5581.74410 0.00020 0.25 0.29
dFocc, NB2090 1560 140 70 I+z +0.00021
5585.812050.00022 0.56 0.32
c1I+z 0.983 0.050
76 Gunn-r 5590.69259 0.00018 0.09 0.26
c2I+z 0.065 0.060
87 Gunn-r +0.00026
5599.640430.00025 0.51 0.37
c1r 1.363 0.047
c2r 0.401 0.051 97 I+z 5607.77517 0.00014 0.56 0.20
+0.00019
Deduced stellar parameters 124 I+z 5629.739950.00018 0.71 0.27
u1I+z 0.406 0.026 140 I+z 5642.75450 0.00015 0.86 0.22
u2I+z 0.171 0.024 141 I+z 5643.56872 0.00028 0.21 0.40
+0.00024
u1r 0.625 0.024 152 I+z 5652.515860.00025 1.39 0.36
u2r 0.112 0.020 168 I+z +0.00019
5665.532290.00018 0.26 0.27
V [km s1 ] 3.5950 0.0040 173 I+z 5669.59976 0.00023 0.14 0.33
+0.079
Density  [ ] 2.4100.075 189 I+z +0.00022
5682.615840.00023 0.49 0.33
+0.011
Surface gravity log g [cgs] 4.6450.010 200 I+z +0.00022
5691.563830.00023 0.11 0.33
Mass M [M ] 0.717 0.025 +0.00014
+0.010
211 I+z 5700.512370.00015 0.52 0.22
Radius R [R ] 0.6670.011
243 I+z 5726.54399 0.00035 1.00 0.50
T e [K]a 4520 120
[Fe/H] [dex]a 0.01 012 Notes. The last column shows the dierence (and its error) be-
Deduced planet parameters tween the measured timing and the one deduced from the best-
RV K [ m s1 ] +5.5
547.95.4 fitting transit ephemeris computed by linear regression, T (N) =
Rp /R +0.00076
0.159450.00077 2 455 528.868227(0.000078) + N 0.81347764(0.00000065) BJDTDB .
+0.0089
btr [R ] 0.65800.0095
+0.012
boc [R ] 0.6550.013
+0.00084
T oc 2 450 000 [BJDTDB ] 5726.950690.00078
Orbital semi-major axis a [AU] 0.01526 0.00018
(see, e.g., Hansen 2008). The red model features a dayside tem-
perature inversion due to the strong optical opacity of TiO and
Roche limit aR [AU] 0.00768 0.00016
+0.030
VO gases (Hubeny et al. 2003; Fortney et al. 2006). The blue
a/aR 1.9860.029 model is run in the same manner as the red model, but TiO and
+0.053
a/R 4.9180.051 VO opacity are removed.
Orbital inclination ip [deg] 82.33 0.20 The blue and red models are constructed to maximize the
+0.0060
Orbital eccentricity e 0.00350.0025 , <0.0298 (3-) emission from the dayside of the planet. Figure 10 shows that
Argument of periastron [deg] 32+115
34 only such bright models could credibly match the data points.
Equilibrium temperature T eq [K]b 1440+4039 While a dayside with no temperature inversion is slightly favored
Density p [Jup ] +0.084
1.8260.078 by the 1.2 m data point, it is dicult to come to a firm con-
+0.063
Density p [g/cm3 ] 1.3770.059 clusion. Day-side emission measurements from Warm Spitzer
+0.013 will help to better constrain the atmosphere as well. They will
Surface gravity log gp [cgs] 3.6720.012
Mass Mp [MJup ] +0.052
2.0340.051 also allow getting an even smaller upper-limit on the orbital
Radius Rp [RJup ] 1.036 0.019 eccentricity.
Fortney et al. model fits to near infrared photometry of other
Notes. (a) From stellar evolution modeling (Sect. 3.3). (b) Assuming A = transiting planets (e.g., Croll et al. 2010) have generally favored
0 and F = 1. inecient temperature homogenization between the day and
night hemispheres, although the warm dayside of WASP-43 b
appears to be at the most inecient end of this continuum. The
the loss of half of the absorbed flux to the night side of the planet. apparent eeciency of temperature homogenization is expected
Clearly the planet is much warmer than this model. In blue and to vary with wavelength. In particular, near infrared bands, at
red are two models where the dayside incident flux is increased minima in water vapor opacity, are generally expected to probe
by a factor of 4/3 to simulate zero redistribution of absorbed flux deeper into atmospheres than the Spitzer bandpasses.
A4, page 11 of 15
 A&A 542, A4 (2012)

1000.0

Detectable Mass [MEarth]

100.0

10.0

1.0

0.1
1.5 2.0 2.5 3.0
Orbital Period [days]

Fig. 10. Model planet-to-star flux ratios compare to the three data Fig. 11. Detectivity domain for a putative WASP-43 c planet, assum-
points. The data are green diamonds with 1 error bars shown. The ing ec = 0 (black) and ec = 0.05 (red). The solid curves delimit the
orange model assumes planet-wide redistribution of absorbed flux. The mass-period region where planets yield maximum TTV on WASP-43 b
red and blue models assume no redistribution of absorbed flux, to maxi- above 114 s (3- detection based on the present data). The dotted curves
mize the day-side temperature. The red model includes gaseous TiO and show the 1- threshold. Nearly horizontal solid, dashed and dotted
VO and has a temperature inversion. (See the figure inset.) The blue and lines shows RV detection limits for RV semi-amplitude K = 5, 10 and
orange models have TiO and VO opacity removed, and do not have a 15 m s1 respectively.
temperature inversion. For each model, filled circles are model fluxes
averaged over each bandpass. The data slightly favor a model with no
day-side temperature inversion. we have significantly improved the parameters of the system.
Notably, our strongly improved precision on the stellar density
(2.41 0.08  ) combined with a very reasonable constraint
4.3. Transit timings on its age (to be younger than a Hubble time) allowed us to
Dynamical constraints can be placed on short orbits companion break the degeneracy of the stellar solution mentioned by H11.
planets from the 23 transits obtained in this study. The linear fit The resulting stellar mass and size are 0.717 0.025 M and
to the transit timings described above yields a reduced 2 of 4.6 0.667 0.011 R . Our deduced physical parameters for the
and the rms of its residuals is 38 s. Two transits (epochs 140 planet are 2.034 0.052 MJup and 1.036 0.019 RJup . Taking
and 152) have a OC dierent from zero at the 4- level. The into account its level of irradiation, the high density of the planet
most plausible explanation for the significant scatter observed favors an old age and a massive core. Our deduced orbital eccen-
in the transit timings is systematic errors on the derived timings tricity, 0.0035+0.0060
0.0025, is consistent with a fully circularized orbit.
due to the influence of correlated noise. Another potential expla- The parameters deduced from the individual analysis of the
nation is asymmetries in the transit light curves caused by the 23 transit light curves show some extra scatter that we attribute
crossing of one or several star spots by the planet. In such cases, to the correlated noise of our data and, possibly, to the crossing
the fitted transit profile is shifted with time, producing timing of spots during some transits. This conclusion is based on the
variations. correlation observed among the transit parameters. These results
We also explored the detectability domain of a second reinforce the interest of performing global analysis of extensive
planet in the WASP-43 system. To this end, we followed Agol data sets in order to minimize systematic errors and to reach high
et al. (2005) and used the Mercury n-body integrator package accuracies on the derived parameters.
(Chambers 1999). We simulated 3-body systems including a sec- We detected the emission of the planet at 2.09 m at better
ond companion with orbital periods ranging between 1.3 and than 11-, the deduced occultation depth being 1560 140 ppm.
50 days, masses from 0.1 M to 2.0 MJup and an orbital ec- Our detection of the occultation at 1.19 m is marginal (790
centricities of ec = 0 and ec = 0.05. For each simulation, we 320 ppm) and more observations would be needed to confirm
computed the rms of the computed transits timings variations. it. We place a 3- upper limit of 850 ppm on the depth of the
The results are shown in Fig. 11 where the computed 1- (red) occultation at 0.9 m. Together, these results strongly favor a
and 3- (black) detection thresholds are plotted for each point in poor redistribution of the heat from the dayside to the nightside
the mass-period plane. We also added RV detectability threshold of the planet, and marginally favor a model with no day-side
curves based on 5 (solid), 10 (dash) and 15 (dot) m s1 semi- temperature inversion.
amplitudes (K).
Based on the present set of timings, the timings variations Acknowledgements. TRAPPIST is a project funded by the Belgian Fund
caused by a 5 Earth mass companion in 2:1 resonance would for Scientific Research (Fond National de la Recherche Scientifique, F.R.S-
have been detected with 3- confidence, while unseen with ra- FNRS) under grant FRFC 2.5.594.09.F, with the participation of the Swiss
National Science Fundation (SNF). M. Gillon and E. Jehin are FNRS Research
dial velocities alone. Associates. We are grateful to ESO La Silla and Paranal stas for their continu-
ous support. We thank the anonymous referee for his valuable suggestions.

5. Conclusions
In this work we have presented 23 transit light curves and References
7 occultation light curves for the ultra-short period planet Agol, E., Steen, J., Sari, R., & Clarkson, W. 2005, MNRAS, 359, 567
WASP-43 b. We have also presented 8 new measurements of Anderson, D. R., Gillon, M., Maxted, P., et al. 2010, A&A, 513, L3
the radial velocity of the star. Thanks to this extensive data set, Bakos, G. ., Torres, G., Pl, A., et al. 2009, ApJ, 710, 1724

A4, page 12 of 15
 M. Gillon et al.: Thirty eclipses of the ultra-short period planet WASP-43 b

Baranne, A., Queloz, D., Mayor, M., et al. 1996, A&AS, 119, 373 Hansen, B. M. S. 2008, ApJS, 179, 484
Barker, A. J., & Ogilvie, G. I. 2009, MNRAS, 395, 2268 Hebb, L., Collier Cameron, A., Loeillet, B., et al. 2009, ApJ, 693, 1920
Berta, Z. K., Charbonneau, D., Bean, J., et al. 2011, ApJ, 736, 12 Hebb, L., Collier Cameron, A., Triaud, A. H. M. J., et al. 2010, ApJ, 708, 224
Brown, D. J. A., Collier Cameron, A., Hall, C., et al. 2011, MNRAS, 415, 605 Hellier, C., Anderson, D. R., Collier Cameron, A., et al. 2009, Nature, 460, 1098
Carlin, B. P., & Louis, T. A. 2008, Bayesian Methods for Data Analysis, third Hellier, C., Anderson, D. R., Collier Cameron, A., et al. 2011a, ApJ, 730, L31
edition (Chapman & Hall/CRC) Hellier, C., Anderson, D. R., Collier Cameron, A., et al. 2011b, A&A, 535, L7
Casali, M., Pirard, J.-F., Kissler-Patig, M., et al. 2006, SPIE, 6269, 29 Holman, M. J., Winn, J. N., Latham, D. W., et al. 2006, ApJ, 652, 1715
Chambers J. E. 1999, MNRAS, 304, 793 Hubeny, I., Burrows, A., & Sudarsky, D. 2003, ApJ, 594, 1011
Claret, A., & Bloemen, S. 2011, A&A, 529, A75 Huber, H. F., Czesla, S., Wolter, U., & Schmitt, J. H. M. M. 2010, A&A, 514,
Croll, B., Jayawardhana, R., Fortney, J. J., Lafrenire, D., & Albert, L. 2010, A39
ApJ, 718, 920 Jackson, B., Barnes, R., & Greenberg, R. 2009, ApJ, 698, 1357
DAngelo, G., Durisen, R. H., & Lissauer, J. J. 2010, Exoplanets, ed. S. Seager Jehin, E., Gillon, M., Queloz, D., et al. 2011, The Messenger, 145, 2
(University of Arizona Press), 319 Johnson, J. A., Gazak, J. Z., Apps, K., et al. 2012, AJ, 143, 111
Demory, B.-O., & Seager, S. 2011, ApJS, 197, 12 Knutson, H. A., Charbonneau, D., Allen, L. E., et al. 2008, ApJ, 673, 526
Eastman, J., Siverd, R., & Gaudi, B. S. 2010, PASP, 122, 935 Lubow, S. H., & Ida, S. 2010, Exoplanets, ed. S. Seager (University of Arizona
Fabrycky, D. C., & Tremaine, S. 2007, ApJ, 669, 1298 Press), 347
Ford, E. B., & Rasio, F. A. 2006, ApJ, 638, L45 Lin, D. N. C., Bodenheimer, P., & Richardson, D. C. 1996, Nature, 380, 606
Fortney, J. J., & Marley, M. S. 2007, ApJ, 666, L45 Mandel, K., & Agol, E. 2002, ApJ, 580, 181
Fortney, J. J., Marley, M. S., Lodders, K., Saumon, D., & Freedman, R. S. 2005, Mastumura, S., Peale, S. J., & Rasio, F. R. 2010, ApJ, 725, 1995
ApJ, 627, L69 Mowlavi, N., Eggenberger, P., Meynet, G., et al. 2012, A&A, 541, A41
Fortney, J. J., Saumon, D., Marley, M. S., Lodders, K., & Freedman, R. S. 2006, Naoz, S., Farr, W. M., Lithwick, Y., et al. 2011, Nature, 473, 187
ApJ, 642, 495 Nymeyer, S., Harrington, J., Hardy, R., et al. 2011, ApJ, 742, 35
Fortney, J. J., Marley, M. S., & Barnes, J. W. 2007, ApJ, 659, 1661 Pirard, J.-F., Kissler-Patig, M., Moorwood, A., et al. 2004, SPIE, 5492, 510
Fortney, J. J., Lodders, K., Marley, M. S., & Freedman, R. S. 2008, ApJ, 678, Schneider, J. 2011, http://exoplanet.eu/catalog-all.php
1419 Schwarz, G. E. 1978, Annals of Statistics, 6, 461
Gelman, A., & Rubin D. 1992, Statist. Sci., 7, 457 Seager, S., & Malln-Ornelas, G. 2003, ApJ, 585, 1038
Gibson, N. P., Aigrain, S., Pollaco, D. L., et al. 2010, MNRAS, 404, L14 Stetson, P. B. 1987, PASP, 99, 111
Gillon, M., Lanotte, A. A., Barman, T., et al. 2010, A&A, 511, A3 Torres, G., Andersen, J., & Gimnez, A. 2010, A&ARv, 18, 67
Gillon, M., Jehin, E., Magain, P., et al. 2011a, Detection and Dynamics of Watson, C. A., Littlefair, S. P., Collier Cameron, A., et al. 2010, MNRAS, 408,
Transiting Exoplanets, Proceedings of OHP Colloquium, 2327 August 2010, 1606
ed. F. Bouchy, R. F. Diaz, & C. Moutou (Platypus Press) Triaud, A. H. M. J. 2011, Ph.D. Thesis, Geneva University, No. Sc. 4342,
Gillon, M., Doyle, A., Maxted, P., et al. 2011b, A&A, 533, A88 http://archive-ouverte.unige.ch/unige:18065
Gimnez, A. 2006, ApJ, 650, 408 Triaud, A. H. M. J., Collier Cameron, A., Queloz, D., et al. 2010, A&A, 524,
Gregory, P. C. 2005, Bayesian Logical Data Analysis for the Physical Sciences A25
(Cambridge University Press) Winn, J. N. 2010, Exoplanets, ed. S. Seager (University of Arizona Press), 55
Gu, P.-G., Lin, D. N. C., & Bodenheimer, P. H. 2003, ApJ, 588, 509 Wu, Y., & Lithwick, Y. 2011, ApJ, 735, 109

Pages 14 to 15 are available in the electronic edition of the journal at http://www.aanda.org

A4, page 13 of 15
 A&A 542, A4 (2012)

Table 1. WASP-43 b photometric eclipse time-series used in this work.

Date Instrument Filter Np Epoch Baseline 120s w r CF

function [%] [%]
06 Dec. 2010 TRAPPIST I+z 393 11 p(t2 ) 0.35, 0.36 0.14, 0.17 1.26, 1.30 1.00, 1.90 1.26, 2.47
09 Dec. 2010 VLT/HAWK-I NB2090 183 13.5 p(t ) + p(l ) + p(xy ) 0.055, 0.055 0.036, 0.036
2 1 1
1.04, 1.04 1.09, 1.14 1.14, 1.18
15 Dec. 2010 TRAPPIST I+z 442 22 p(t2 ) 0.36, 0.36 0.12, 0.14 0.97, 0.97 1.03, 1.13 1.00, 1.10
19 Dec. 2010 TRAPPIST I+z 480 27 p(t2 ) 0.27, 0.28 0.11, 0.12 1.00, 1.01 1.00, 1.18 1.00, 1.19
28 Dec. 2010 TRAPPIST I+z 574 38 p(t2 ) 0.32, 0.32 0.12, 0.12 1.01, 1.02 1.29, 1.50 1.30, 1.53
28 Dec. 2010 Euler Gunn-r 111 38 p(t ) + p(xy )
2 2
0.10, 0.10 0.10, 0.10 1.54, 1.64 1.04, 1.15 1.61, 1.90
30 Dec. 2010 TRAPPIST z 332 40.5 p(t2 ) 0.30, 0.30 0.14, 0.14 1.04, 1.04 1.00, 1.00 1.04, 1.04
01 Jan. 2011 TRAPPIST I+z 407 43 p(t2 ) + p(xy2 ) 0.27, 0.27 0.11, 0.12 1.00, 1.01 1.00, 1.00 1.00, 1.01
06 Jan. 2011 TRAPPIST I+z 273 49 p(t2 ) 0.24, 0.24 0.11, 0.11 1.07, 1.09 1.00, 1.38 1.07, 1.50
09 Jan. 2011 VLT/HAWK-I NB1190 115 51.5 p(t ) + p(b1 ) + p(xy1 ) 0.087, 0.087 0.047, 0.048
2
2.22, 2.22 1.00, 1.00 2.22, 2.22
14 Jan. 2011 TRAPPIST I+z 237 59 p(t2 ) 0.17, 0.18 0.09, 0.10 0.91, 0.93 1.04, 1.34 0.95, 1.24
2
19 Jan. 2011 TRAPPIST I+z 279 65 p(t ) 0.18, 0.19 0.09, 0.10 0.91, 0.95 1.00, 2.16 0.91, 2.05
23 Jan. 2011 TRAPPIST I+z 244 70 p(t2 ) + p(xy2 ) 0.20, 0.20 0.08, 0.09 1.02, 1.02 1.00, 1.00 1.02, 1.02
28 Jan. 2011 Euler Gunn-r 114 76 p(t2 ) + p(xy1 ) 0.13, 0.13 0.13, 0.13 1.36, 1.42 1.17, 1.24 1.59, 1.76
06 Feb. 2011 Euler Gunn-r 107 87 2
p(t ) 0.13, 0.15 0.13, 0.15 0.92, 1.07 1.10, 1.86 1.02, 1.99
14 Feb. 2011 TRAPPIST I+z 311 97 p(t2 ) 0.21, 0.22 0.11, 0.12 1.05, 1.08 1.00, 1.42 1.05, 1.53
08 Mar. 2011 TRAPPIST I+z 285 124 p(t2 ) 0.21, 0.22 0.11, 0.13 1.11, 1.15 1.00, 2.00 1.11, 2.30
10 Mar. 2011 TRAPPIST z 216 126.5 p(t2 ) 0.22, 0.22 0.17, 0.17 1.09, 1.09 1.00, 1.00 1.09, 1.09
2
21 Mar. 2011 TRAPPIST I+z 322 140 p(t ) 0.20, 0.21 0.11, 0.13 0.83, 0.88 1.20, 2.04 1.00, 1.80
22 Mar. 2011 TRAPPIST I+z 237 141 p(t2 ) 0.21, 0.22 0.10, 0.11 1.10, 1.14 1.00, 1.67 1.11, 1.91
28 Mar. 2011 TRAPPIST z 195 148.5 p(t2 ) 0.28, 0.28 0.18, 0.18 0.81, 0.81 1.03, 1.04 0.84, 0.84
2
31 Mar. 2011 TRAPPIST I+z 238 152 p(t ) 0.24, 0.26 0.12, 0.17 1.04, 1.15 1.00, 2.28 1.04, 2.63
02 Apr. 2011 TRAPPIST z 160 154.5 p(t2 ) + p(b1 ) 0.18, 0.18 0.11, 0.11 0.93, 0.93 1.27, 1.30 1.19, 1.21
13 Apr. 2011 TRAPPIST I+z 293 168 p(t2 ) + p(xy2 ) 0.23, 0.23 0.12, 0.13 1.00, 1.01 1.00, 1.00 1.00, 1.01
2
17 Apr. 2011 TRAPPIST I+z 289 173 p(t ) 0.26, 0.27 0.13, 0.15 1.05, 1.09 1.00, 1.26 1.05, 1.36
30 Apr. 2011 TRAPPIST I+z 327 189 p(t2 ) 0.32, 0.33 0.16, 0.17 1.31, 1.33 1.07, 1.30 1.40, 1.73
09 May 2011 TRAPPIST I+z 280 200 p(t2 ) 0.19, 0.20 0.10, 0.12 0.98, 1.05 1.00, 1.77 0.98, 1.85
11 May 2011 TRAPPIST z 214 202.5 2
p(t ) 0.21, 0.21 0.14, 0.14 1.03, 1.03 1.23, 1.29 1.27, 1.33
18 May 2011 TRAPPIST I+z 250 211 p(t2 ) 0.18, 0.18 0.10, 0.11 0.93, 0.96 1.10, 1.48 1.02, 1.42
13 Jun. 2011 TRAPPIST I+z 605 243 p(t2 ) 0.33, 0.34 0.13, 0.14 0.97, 0.99 1.25, 1.87 1.22, 1.84

Notes. For each light curve, this table shows the date of acquisition, the used instrument and filter, the number of data points, the epoch based
on the transit ephemeris presented in H11, the selected baseline function (see Sect. 3.1), the standard deviation of the best-fit residuals (unbinned
and binned per intervals of 2 min), and the deduced values for w , r , and CF = r w (see Sect. 3.1). For the baseline function, p(
N ) denotes,
respectively, a N-order polynomial function of time (
= t), the logarithm of time (
= l), x and y positions (
= xy), and background (
= b). For
the last five columns, the first and second value correspond, respectively, to the individual analysis of the light curve and to the global analysis of
all data.

A4, page 14 of 15
 M. Gillon et al.: Thirty eclipses of the ultra-short period planet WASP-43 b

Table 4. Median and 1- errors of the posterior PDFs deduced for the jump parameters from the individual analysis of the eclipse light curves.

Epoch Filter dFocc b dF W T0 OC

[%] [%] [h] [BJDTDB -2 450 000] [min]
+0.028 +0.14 +0.038 +0.00043
11 I+z 0.6820.035 2.920.13 1.2500.034 5537.816480.00048 0.09 0.69
+0.015
13.5 NB2090 0.1560.016
+0.030 +0.10 +0.024
22 I+z 0.6730.036 2.500.11 1.2340.029 5546.76490 0.00020 0.16 0.29
+0.019 +0.00014
27 I+z 0.7160.021 2.721 0.077 1.250 0.018 5550.832280.00013 0.15 0.20
+0.043 +0.11 +0.029
38 I+z 0.6450.052 2.430.09 1.2240.026 5559.78038 0.00021 0.07 0.30
+0.019
38 Gunn-r +0.035
0.6400.041 +0.088
2.4230.083 1.1810.020 5559.78092 0.00016 0.71 0.22
40.5 z +0.061
0.0900.053
+0.043 +0.022
43 I+z 0.5730.047 2.60 0.17 1.1760.024 5563.84773 0.00023 0.12 0.33
+0.042 +0.090 +0.020 +0.00018
49 I+z 0.6110.048 2.4250.079 1.2100.021 5568.728300.00017 0.54 0.26
+0.035
51.5 NB1190 0.0820.037
+0.022 +0.067
59 I+z 0.6430.027 2.5340.070 1.219 0.029 5576.86380 0.00015 0.51 0.22
+0.025 +0.064 +0.00012
65 I+z 0.6510.030 2.3780.060 1.212 0.014 5581.744120.00013 0.28 0.19
+0.026 +0.24 +0.041 +0.00039
70 I+z 0.6670.032 2.600.22 1.2220.038 5585.812110.00038 0.59 0.56
76 Gunn-r +0.058
0.5460.073 +0.10
2.370.09 +0.023
1.1580.020 +0.00017
5590.692600.00016 0.05 0.25
87 Gunn-r +0.030
0.6560.037 +0.089
2.6490.087 +0.019
1.2570.020 5599.64042 0.00017 0.57 0.25
+0.032 +0.068 +0.017 +0.00014
97 I+z 0.6430.041 2.4560.075 1.1880.018 5607.775150.00015 0.63 0.22
+0.028 +0.064 +0.00014
124 I+z 0.6190.029 2.4560.059 1.213 0.014 5629.740020.00013 0.79 0.20
126.5 z +0.033
0.0340.023
+0.020 +0.070 +0.00016
140 I+z 0.6980.024 2.5700.072 1.223 0.018 5642.754510.00017 0.86 0.25
+0.029 +0.093 +0.00019
141 I+z 0.6330.037 2.7120.094 1.240 0.020 5643.568710.00018 0.18 0.27
148.5 z +0.042
0.0390.027
+0.023 +0.024 +0.00021
152 I+z 0.7250.027 2.82 0.11 1.2680.023 5652.515810.00023 1.48 0.33
154.5 z +0.050
0.0420.029
+0.033 +0.018 +0.00022
168 I+z 0.6060.036 2.42 0.12 1.1910.016 5665.532220.00021 0.36 0.32
+0.029 +0.098 +0.029
173 I+z 0.6700.030 2.7410.084 1.2740.026 5669.59946 0.00025 0.58 0.36
+0.051 +0.10 +0.024 +0.00020
189 I+z 0.5960.072 2.410.11 1.2140.026 5682.615920.00021 0.61 0.30
+0.017 +0.060 +0.016 +0.00013
200 I+z 0.6770.021 2.7380.070 1.2180.015 5691.563780.00012 0.05 0.19
202.5 z +0.054
0.0650.044
+0.022 +0.067 +0.014 +0.00015
211 I+z 0.6420.031 2.4000.073 1.2070.017 5700.512320.00014 0.47 0.22
+0.047 +0.19 +0.050 +0.00056
243 I+z 0.6870.052 2.560.16 1.2360.043 5726.544360.00044 1.57 0.81

Notes. For each light curve, this table shows the epoch based on the transit ephemeris presented in H11, the filter, and the derived values for the
occultation depth, impact parameter, transit depth, transit duration, and transit time of minimum light. The last column shows for the transits the
dierence (and its error) between the measured timing and the one deduced from the best-fitting transit ephemeris computed by linear regression,
T (N) = 2 455 528.868289(0.000072) + N 0.81347728(0.00000060) BJDTDB .

A4, page 15 of 15