HATS-36b and 24 other transiting/eclipsing systems from the HATSouth - K2 Campaign 7 program

We report on the result of a campaign to monitor 25 HATSouth candidates using the K2 space telescope during Campaign 7 of the K2 mission. We discover HATS-36b (EPIC 215969174b), a hot Jupiter with a mass of 2.79$\pm$0.40 M$_J$ and a radius of 1.263$\pm$0.045 R$_J$ which transits a solar-type G0V star (V=14.386) in a 4.1752d period. We also refine the properties of three previously discovered HATSouth transiting planets (HATS-9b, HATS-11b, and HATS-12b) and search the K2 data for TTVs and additional transiting planets in these systems. In addition we also report on a further three systems that remain as Jupiter-radius transiting exoplanet candidates. These candidates do not have determined masses, however pass all of our other vetting observations. Finally we report on the 18 candidates which we are now able to classify as eclipsing binary or blended eclipsing binary systems based on a combination of the HATSouth data, the K2 data, and follow-up ground-based photometry and spectroscopy. These range in periods from 0.7 days to 16.7 days, and down to 1.5 mmag in eclipse depths. Our results show the power of combining ground-based imaging and spectroscopy with higher precision space-based photometry, and serve as an illustration as to what will be possible when combining ground-based observations with TESS data.

Transiting planet systems are the most valuable systems for exoplanet studies due to the wide range of characterisation observations that can been made for them, both during transit and also in secondary eclipse. However since the geometric probability of transit is low, large numbers of stars must be monitored in order to discover such systems. This task began over a decade ago with lenses-based wide-field ground-based surveys, most successfully HATNet (Bakos et al. 2004), WASP (Pollacco et al. 2006), and KELT (Pepper et al. 2007). These were followed by space-based surveys CoRoT (Barge et al. 2008) and Kepler (Borucki et al. 2010). In this paper we combine data from two ongoing transit surveys: HATSouth (Bakos et al. 2013, hereafter HS) and K2 (Howell et al. 2014). While the HATSouth survey has been monitoring selected fields in the southern Hemisphere over the last six years, the K2 survey has been monitoring selected ecliptic plane fields since 2014. K2 Campaign 7 (C7) monitored a field centered at α = 19 h 11 m 19s, δ = −23 • 21 ′ 36 ′′ between 2015 October 4 to 2015 December 26. By chance, approximately 25% of this field had already been monitored by the HATSouth survey five years earlier between 2010 March to 2011 August. It was therefore possible for us to propose for K2 targets based on the HATSouth light curves that showed transit-like signals, and this was done as part of the K2 Guest Observer program (GO7067; PI: Bakos). In addition we were also able to select as K2 targets the HAT-South transiting planets that had already been confirmed as bona fide planets, namely HATS-9b , HATS-11b and HATS-12b (Rabus et al. 2016); this formed the K2 Guest Observer program GO7066 (PI:Bakos). In total 25 stars were monitored by K2 as part of these programs, which we collectively name "HS-K2C7" targets. Details for each of the HS-K2C7 candi-dates are set out in Table 1.
This paper is organised as follows. In Section 2 we describe the observations made for each HS-K2C7 target, including the initial HATSouth discovery photometry, reconnaissance spectroscopy and photometry, radial velocity measurements, and the K2 photometry. In Section 3 we analyse each HS-K2C7 candidate, including the discovery of a new transiting exoplanet (HATS-36b), refinement of the parameters of three known HATSouth transiting planets (HATS-9b, HATS-11b, HATS-12b), identification of three Jupiter-radii candidates, and classification of 18 HS-K2C7 candidates as eclipsing binaries or blended eclipsing binaries. Finally in Section 4 we discuss the results and implications of this joint ground/spacebased photometry project.

HATSouth Photometry
The HATSouth global telescope network consists of six HS4 units, two each at Siding Spring Observatory (SSO), Las Campanas Observatory (LCO), and the High Energy Spectroscopic Survey (H. E.S.S.) site. Each HS4 unit holds four Takahashi astrographs (f /2.8, 18 cm apertures) which are each coupled to an Apogee U16M 4K×4K CCD camera. Imaging is performed using Sloan r-band filters and with exposure times of 240 s. A full description of the HATSouth telescope network hardware and operations can be found in Bakos et al. (2013).
As part of the HATSouth survey (http://hatsouth.org/), we monitored a 64 square degree field (centered at 19 h 30 m 00s,−22 • 30 ′ 00 ′′ ) between 2010 March to 2011 August. In total we obtained 10137 images, with a cadence of approximately 300 s. In Table 2 we set out a summary of the HATSouth observations of this field, including the number of images from each site, the observation dates, and the cadence.
HATSouth raw images are reduced to light curves using an automated aperture photometric pipeline detailed in Penev et al. (2013). The light curves are detrended using External Parameter Decorrelation (EPD; Bakos et al. 2010) and the Trend Filtering Algorithm (TFA; Kovács et al. 2005). These light curves are then combined for transit-like features using the box-fitting least squares algorithm Kovács et al. (2002). We found 25 candidates with transit-like signals which were also were on-silicon for the K2 Campaign 7 (see Section 2.5). We designate these as "HS-K2C7 candidates", and summarise them in Table 1. Three of these have already been published as confirmed transiting planets HATS-9b , HATS-11b (Rabus et al. 2016), and HATS-12b (Rabus et al. 2016). We therefore do not discuss these further in this Section.
HATSouth light curves for all stars overlapping with the K2 Campaign 7, including the HS-K2C7 candidates, are publicly available at http://data.hatsurveys.org/.
By way of example, we present the HATSouth light curve for HATS-36b in Figure 1, which shows the 18 mmag transit-like dip when phase-folded at P = 4.1752379 days. All HATSouth photometric data used in this paper is set out in Table 3. We note that as a result of the EPD and TFA detrending, and also due to blending from neighbors, the apparent transit depth in the HATSouth   light curves is somewhat shallower than that of the true depth in the Sloan r filter (the apparent depth is typically 85% that of the true depth).

Reconnaissance Spectroscopy
As part of the usual HATSouth program to follow-up transiting planet candidates, follow-up spectroscopy was obtained for the HS-K2C7 candidates, primarily using WiFeS (Dopita et al. 2007) on the ANU 2.3 m telescope at SSO. Full details for this observing program are given in Bayliss et al. (2013). In summary, we obtain a flux calibrated, high S/N, R = λ/∆λ = 3000 spectrum for each candidate in order to determine the spectral type and class of the host star. These spectra are compared with a grid of synthetic templates from the MARCS model atmospheres (Gustafsson et al. 2008) in order to estimate T eff . The results for the HS-K2C7 candidates are set out in Table 4. For scheduling reasons, we obtained reconnaissance spectra for two candidates (HATS579-014 and HATS579-036) using FEROS (Kaufer & Pasquini 1998) on the MPG 2.2 m telescope at the ESO observatory in La Silla, Chile (LSO). Spectral parameters were derived from these spectra using the CERES code (Brahm et al. 2017a) and these are also tabulated in Table 4. We note that the spectrum of HATS579-014 showed that the candidate was a spectroscopic binary. We were not able to obtain a spectrum for two candidates, HATS579-050 and HATS624-003, as they were too optically faint (V=16.221 and V=15.537 respectively).
We also use WiFeS on the ANU 2.3 m telescope to obtain multi-epoch medium resolution (R = λ/∆λ = 7000) spectra to check for large amplitude (K>2 km s −1 ) radial velocity variations in phase with the photometric signal. This allows us to identify candidates which are eclipsing binaries without the need of more resourceintensive high-precision radial velocity monitoring. For those targets with multi-epoch spectra we list the measured semi-amplitudes in Table 4. Again for the candidate HATS579-036, FEROS was used instead of WiFeS to measure the radial velocity semi-amplitude.

Reconnaissance Photometry
In order to further rule out eclipsing binaries, and refine our ephemerides, we obtained ground-based photo-  metric follow-up for four HS-K2C7 candidates. In this section we detail all of these observations.
• HATS-36 (EPIC 215969174): we obtained initial photometric follow-up on the night of 2013 Jul 1 using the 0.3 m Perth Exoplanet Survey Telescope (PEST) in Perth, Australia. For a full description of the PEST facility see Zhou et al. (2014) and the PEST website (http://pestobservatory.com). Imaging was carried out in the R C band with ex-posure times of 120 s. In total 137 exposures of HATS-36 were taken. Data was reduced via aperture photometry as described in Zhou et al. (2014). The resulting light curve is plotted in Figure 2 and the data is provided in Table 3. A full transit is clearly detected with a depth and duration consistent with the HATSouth discovery data. These data allowed us to refine the transit ephemeris.
The following year we observed two consecutive partial transits of HATS-36b with the multiband   The light curves are plotted in Figure 2 and the data is set out in Table 3. These high precision light curves confirmed the transit depth was colour independent and were consistent with a transiting planet. Both the PEST and GROND light curves for HATS-36 are used in the global fitting described in Section 3.
• HATS579-037(EPIC 215234145): we obtained reconnaissance photometric follow-up from the ANU2.3 m imaging camera at SSO on 2012 September 8 in I band which showed a "V"-shaped transit with a depth of 20 mmag (compared with the 10 mmag transit observed in the HATSouth discovery data). This colour dependent depth difference was confirmed with observations from the PEST 0.3 m telescope which observed a 10 mmag transit in RC band on 2013 July 7, and a 20 mmag transiting I band on 2015 August 11. The photometric data for this candidate is set out in Table 3, while the follow-up light-curves are plotted in Figure 3. The color dependent depths are a strong indication that this candidate is an eclipsing binary. 27. None of these observations showed a transit feature. It is likely that uncertainties in the HAT-South ephemeris for this candidate were responsible for us missing the transit event for this candidate during these photometric follow-up observations.
• HATS579-039 (EPIC 215353525): we obtained reconnaissance photometric follow-up in I band from the SWOPE 1 m at LCO on 2013 August 20. A deep (25 mmag) V-shaped transit was observed, which was consistent with the r band discovery data. This observation confirmed the transit feature and refined the ephemeris for this candidate.
The data for this observation are set out in Table 3.
A summary of all photometric observations are set out in Table 2. All of the follow-up photometric data are set out in Table 3.

High Resolution Spectroscopy -Radial Velocities
We obtain radial velocity measurements for HS-K2C7 candidates that remain after the reconnaissance spectroscopy and photometry set out in Sections 2.2 & 2.3 respectively. An additional magnitude constraint of V<14.5 is placed on candidates at this stage, as the radial velocity monitoring of candidates fainter than V=14.5 is beyond the reach of most telescopes/instruments. Only in exceptional cases such as an M-dwarf host (e.g. HATS-6b; Hartman et al. 2015)) do we attempt high precision radial velocity measurements for such faint candidates. By this criteria just seven candidates remained: three of which have already been published (HATS-9b, HATS-11b, and HATS-12b) and four which are presented below: • HATS-36 (EPIC 215969174): We measured the radial velocity of HATS-36 using FEROS on the MPG 2.2 m at LSO between 2013 July 16 and 2014 July 24. In total 16 measurements were made spread over the phase of the photometric period (4.1752 day). These data were reduced using the CERES FEROS echelle spectrograph pipeline described in Brahm et al. (2017a). We find a radial velocity variation in phase with the photometric period and with an amplitude of K=324±45 m s −1 , indicating the transiting companion was of a planetary mass. We present these radial velocity data in Table 5 and plot the data along with a best fit circular orbit in Figure 4. These data are used in Section 3 to model the global parameters of the system -primarily determining the mass of the transiting planet.
• HATS579-007 (EPIC 215626177): We obtained multiple high resolution spectra of this target from several different instruments, however the data showed no radial velocity variation above 10 m s −1 .
Coupled with the lack of a transit in the reconnaissance photometry (see Section 2.3), we put the monitoring of this candidate on hold until K2 data became available (see Section 2.5).
• HATS579-015 (EPIC 217149884): We obtained four FEROS observations for this candidate between 2016 May 18 and 2017 May 31. The observations show an in-phase radial velocity variation with K = 13.49715.68 ± 0.011 km s −1 , indicating the companion is of stellar mass and the candidate is therefore an eclipsing binary.
• HATS579-037 (EPIC 215234145): We obtained multiple high resolution spectra of this target from several different instruments, however the data showed no radial velocity variation above 10 m s −1 .
Due to this lack of variation, along with the colordependent transit depths (see Section 2.3), we put this candidate on hold until K2 data became available (see Section 2.5).

K2 Photometry
The K2 mission (Howell et al. 2014) uses the Kepler space telescope (Borucki et al. 2010) to monitor selected fields in the ecliptic plane for campaigns of approximately 80 d each. Between 2015 Oct 4 and 2015 Dec 2, K2 Campaign 7 monitored a field centered on 19 h 11 m 19s, −23 • 21 ′ 36 ′′ . Since this K2 field overlapped with the previously monitored HATSouth field described in 2.1, we were able to obtain K2 data for the 25 HS-K2C7 candidates listed in Table 2 via the K2 Guest Observer programs GO7066 and GO7067. A summary of the K2 imaging is set out in Table 1.
The K2 data was made available on 2016 April 20. We used three different versions of the K2 light curves downloaded from the Mikulski Archive for Space Telescopes (http://archive.stsci.edu). We used the K2 PDC light curves, the light curves produced by the "self-flatfielding" technique described in Vanderburg & Johnson (2014), and the K2 light curves from the everest opensource pipeline fully described in Luger et al. (2016Luger et al. ( , 2017. In addition we used the K2SC light curves (Aigrain et al. 2016) which were provided to us upon request (B. Pope 2016, private communication).
All four of these data products are derived from the same raw pixel data. However the different apertures and detrending techniques result in light curves with sometimes quite marked differences. We therefore analyzed all four in order to help categorize our HS-K2C7 candidates. Each HS-K2C7 candidate was detrended individually to take into account the fact that each light curve potentially contained a combiniation of K2 systematics, variability due to stellar rotation, and elipsoidal variability. We detrended the light curves by first masking the transit/eclipse signal, and then flattening the curve using a fifth order SavitzkyGolay filter (Savitzky & Golay 1964), with the window length selected to detrend variations due to K2 systematics and variability due to stellar rotation, but to avoid flattening any potential ellipsoidal variation or secondary eclipse. The resulting 25 light curves are presented in the Appendix. In most instances we utilized the Everest light curves, however for some light curves the Everest algorithm removed real astrophysical variability, so in those cases we used the PDC light curves. Features seen in these light curves, such as secondary eclipses and out-of-transit variability, are noted in Table 4.
Due to the space-based environment, the large aperture (1 m), and near continuous 80 day coverage, all of the K2 light curves we used are of very high precisionfor K2 Campaign 7 the median 6.5-hr combined differential photometric precision (CDPP) for a Kep=12 mag dwarf star was 120 ppm. With this very high precision we are able to see features not visible or ambiguous in the HATSouth discovery light curves. Most importantly, we can search for secondary eclipses, out-of-transit ellipsoidal variation, and odd/even transit depth differences. These features, in an optical light curve and at significant amplitudes, are characteristic of eclipsing binary systems rather than transiting exoplanets.
The timing was such that the K2 observations and data followed after we had already completed the photometric and spectroscopic follow-up of the HS-K2C7 candidates detailed in Sections 2.3, 2.2, & 2.4. In this respect some candidates had already been robustly identified as either transiting exoplanets or eclipsing binaries before the K2 data was analyzed. However in other cases the K2 data were critical to our classification of the candidate. Here we detail the findings for each HS-K2C7 candidate.
• HATS-36 (EPIC 215969174): The phase folded K2 light curve for HATS-36 is presented in Figure 5, and the data is tabulated in Table 3. It shows a 15mmag U-shaped transit consistent with the discovery and follow-up photometry presented in Sections 2.1 & 2.3 respectively. There is no secondary eclipse, odd/even depth difference, or out-of-transit variation present to the limits of the photometry. This light curve is used in our global analysis of this newly discovered transiting exoplanet system in Section 3.
• Known HATSouth Planets: The K2 data for HATS-9, HATS-11, and HATS-12 are consistent with the previous published exoplanet discoveriesi.e. there was no evidence of any secondary eclipses, odd/even depth differences, or out-of-transit variations. We analyze these light curves for additional planets, TTVs and phase modulations in Section 3.
• HATSouth Candidates: HATS579-040, HATS579-044, and HATS579-048 all have K2 light curves that are consistent with the HATSouth discovery data with no sign of secondary eclipses, odd/even depth differences, or out-of-transit variations. The light curves are set out in the Appendix.
•  The over-plotted model is based on a global fit of all available data, and accounts for differences in limb darkening between the R-and i-bands, but assumes all other light curve parameters are the same. The poor fit of this model shows clear differences in the observed transit depths between the R-band and i-band observations, which allow us to classify this candidate as a eclipsing binary.
ble 4, and the light curves are set out in the Appendix. For HATS579-009 we do not detect a secondary eclipse, however we detected a high amplitude (K=20 km s −1 ) in-phase radial velocity variation (see Section 2.4) indicating it is an eclipsing binary. Likewise HATS579-037 does not have a detectable secondary eclipse, however it had color dependent transit depths (see Figure 3) and the K2 light curve also shows out-of-transit variation when phase-folded at ×2 the discovery period. Therefore we classify this candidate as a eclipsing binary.

ANALYSIS
In this section we analyze the newly discovered transiting exoplanet HATS-36b, the three known HATSouth with K2 data (HATS-9b, HATS-11b, and HATS-12b), and the three HS-K2C7 targets that remain as transiting exoplanet candidates.

HATS-36b -A high-mass transiting hot Jupiter
To derive the physical properties of HATS-36 we obtain initial stellar parameters from the high-resolution spectra of HATS-36 from FEROS, together with ZASPE (Brahm et al. 2017b). This provides a first estimate of the temperature (T eff⋆ ), surface gravity (log g ⋆ ), metallicity ([Fe/H]), and projected equatorial rotation velocity (v sin i) of HATS-36. The T eff⋆ and [Fe/H] values are then used with the stellar density ρ ⋆ , determined from the combined light-curve and radial velocity analysis, to determine a first estimate of the stellar physical parameters following the method described in Sozzetti et al. (2007). We use the Yonsei-Yale isochrones (Y2; Yi et al. 2001) to derive the stellar mass, radius and age that best fit our estimated T eff⋆ , [Fe/H] and ρ ⋆ values. We then determine a revised value of log g ⋆ and perform a second iteration of ZASPE holding log g ⋆ fixed to this value while fitting for T eff⋆ , [Fe/H] and v sin i. We again compare this new value of ρ ⋆ to the Y2 isochrones to produce our final adopted values for the physical stellar parameters. Figure 6 shows the final log g ⋆ value plotted against T eff⋆ , with 1-σ and 2-σ confidence ellipsoids and the appropriate Y2 isochrones for various stellar ages. The final parameters indicate that HATS-36 is a G0V dwarf star (T eff⋆ =5970 ± 160 K, log g ⋆ =4.344 ± 0.020) with a mass and radius of M ⋆ =1.186 ± 0.036 M ⊙ and   Table 6.
To exclude blended eclipsing binary scenarios for HATS-36b, we carried out an analysis of all the data fol- lowing the methodology set out in Hartman et al. (2012). We model the photometric data, including the K2 observations, as an eclipsing binary system blended with a third star. The stars in the model are constrained using the Padova isochrones (Girardi et al. 2000), and also must have a blended spectrum consistent with the determined atmospheric parameters. We also simulate composite cross-correlation functions (CCFs) and use them to predict radial velocities and bisector spans for each blend scenario. All blend models tested can be rejected with > 6σ confidence based on the photometry alone. Moreover, none of the blend models tested would produce RV variations or non-variable bisectors consistent with the observations.
We use a Differential Evolution Markov Chain Monte Carlo procedure to determine the posterior distribution of the parameters. We compare the Bayesian evidence for an e=0 fixed eccentricity model to a free-e model, and find that the fixed circular model is preferred. We therefore adopt the circular orbit model. The 95% confidence upper limit on the eccentricity is e < 0.294. The planetary parameters resulting from this global fit are set out in Table 7. HATS-36b is a high mass (M p =2.79 ± 0.40 M J ), Jupiter-sized (R p =1.263 ± 0.045 R J ) transiting planet with an orbital period of 4.1752379 ± 0.0000021d. It has a bulk density of 1.72 ± 0.29 g cm −3 .
From our radial velocity monitoring we measure a high jitter (105 ± 26 m s −1 ), hinting at strong stellar activity.
Jitter of this amplitude is seen in other other hot Jupiter hosts, e.g. HATS-27b with RV jitter =72 m s −1 (Espinoza et al. 2016). However in this case we have a precise K2 light curve with which to examine the underlying nature of this radial velocity jitter. In order to do this we take the HATS-36 K2 light curve (Vanderburg & Johnson 2014), remove the transit events, and examine the longer time-scale variability using a Lomb Scargle analysis. From this analysis we detect a strong modulation at 14.4 d with a peak-to-peak amplitude of approximately 0.5%. Such a rotational period is typical for a star of the spectral type and class of HATS-36 (McQuillan et al. 2014). We confirm this rotational signal is also present in the HATSouth discovery light curve, again showing modulation at P=14 d. The Lomb-Scargle power spectrum is presented in Figure 7 for both the K2 data and the HATSouth data. Assuming R ⋆ = 1.186 ± 0.036, this rotation period would result in an equatorial rotational of v e = 4.17 km s −1 , in contrast the spectroscopic derived v sin i = 5.47 ± 0.59 km s −1 . The fact that the derived rotational period is 2 − σ below the spectroscopic v sin i may be due to non-equatorial spots and solar-like differential rotation. 11b,12b High precision K2 data allow us to revisit the three transiting exoplanets already published by the HAT-South team, namely HATS-9b  and HATS-11b & HATS-12b (Rabus et al. 2016). We augment our originally published photometric and spectroscopic data with the new K2 photometry and re-run the global modelling with the methodology described in those papers and in Section 3. 1. Here we make use of the (Espinoza et al. 2016) light curves rather than the everest light curves, and we apply filter low frequency variations due to stellar activity and instrumental errors prior to fitting the transit model. For HATS-9b and HATS-11b short cadence data was available, and we made use of these observations, rather than the long cadence, in our analysis.
The results allow us to improve the precision for the planetary parameters for these systems, and we list these in Table 8. Most of the revisions to the planetary parameters are relatively minor. The largest change is for the radius of HATS-9b, which is revised upwards by almost 10%. This is primarily due to the limited photometric follow-up that was available when the parameters of HATS-9b were calculated in the original analysis of Brahm et al. (2015).
For the case of HATS-11b our global modelling, assuming a circular orbit, also finds a secondary eclipse with a planet-to-star flux ratio of 0.0032±0.0012. In addition to improving the parameters for these planets, we utilize the K2 light curves to search for additional transiting planets. The discoveries of WASP-47d and WASP-47e (Becker et al. 2015) show that hot Jupiters can have nearby planetary neighbours. After removing the transit events from the hot Jupiters, we search the K2 light curves of HATS-9, HATS-11, and HATS-12, using the BLS search algorithm as discussed in Section 2. 1. We do not find any significant evidence for additional transiting planets in any of these systems.
Using the original light curves and the K2 data, we check for any changes in the timing of the transits for these exoplanets. We do this by fitting the best fit transit model (using the parameters set out in Table 8) to each individual transit event, and measure the difference between the best fit central transit time and the expected transit time from a purely Keplerian orbit. For all three systems we find no evidence of any transit timing variations.
Finally we analyze the K2 light curves for evidence for of additional variability. HATS-11 and HATS-12 only show long term drift in the K2 data that are likely caused by systematics from the spacecraft. HATS-9 shows an additional modulation with a period of approximately 8.8 d and a peak-to-peak amplitude of approximately 0.1%. If this modulation is indeed due to stellar rotation, it would imply an equatorial rotation of v e = 8.6 km s −1 . Given the spectroscopic v sin i of HATS-9 is 4.58 ± 0.90 km s −1 , this would mean the spin axis of the star is inclined with respect to our line of sight, and that the planet is on a mis-aligned orbit.
• HATS579-040 (EPIC 216231580): The transiting candidate has a period of P=3.905 d and shows a 17 mmag "U"-shaped transit. The host star is a faint (V=14.963) G9 dwarf (T eff =5304±300 K). Assuming a host star mass of 0.9 M ⋆ , and with a radial velocity semi-amplitude of K< 2.0 km s −1 , we can constrain the companion mass to be less than 15M J for an unblended single-host/single companion system. Gaia DR1 (Lindegren et al. 2016) shows a single source (G Gaia =14.702) having no neighbours within 15 ′′ down to the Gaia DR1 magnitude limit (G Gaia ∼ 20) and separation limit (∼ 1 ′′ ).
• HATS579-044 (EPIC 217393088): The transiting candidate has a period of P=1.320 d and shows a 10 mmag "U"-shaped transit. The host star is faint (V=15.575) G0 dwarf (T eff =5945 ± 300 K). Assuming a host star with 1.1 M ⋆ , and with a radial velocity semi-amplitude of K< 2.0 km s −1 , we can constrain the companion mass to be less than 11.5M J for an unblended single-host/single companion system. Gaia DR1 shows the host is a single source (G Gaia =15.288), having no neighbours within 15 ′′ .
• HATS579-048 (EPIC 215816368): The transiting candidate has a relatively long period of P=10.148 d and shows a 20 mmag "U"-shaped transit. The host star is a faint (V=15.835) G9 dwarf (T eff =5320 ± 300 K). Assuming a host star mass of 0.9 M ⋆ , with a radial velocity semiamplitude of K7.24 ± 2.21 km s −1 , we would determine a companion mass of 70M J -approximately at the lower limit for H-burning. If confirmed, this would join the small population of known transiting brown dwarfs, and would follow the trend of having a longer orbital period than typical hot Jupiters (Bayliss et al. 2017). From the Gaia DR1 we see a primary source (G Gaia =15.610), with a neighbour at 9.5 ′′ (G Gaia =18.182). However by analyzing multiple pixel-aperture sizes from the SFF K2 light curves (Vanderburg & Johnson 2014), we determine the detected transit signal originates from the primary candidate star rather than the fainter neighbor.

DISCUSSION
This is the first time we have vetted HATSouth candidates using the high precision photometry afforded by  0.029 the Kepler telescope, although it has been done for HAT-Net under K2 program GO0116 (PI Bakos) resulting in the discovery of HAT-P-56b (Huang et al. 2015). In the cases of both HATS-36b and HAT-P-56b the radial velocity semi-amplitudes are high (∼ 300 m s −1 ), but the stellar jitter is also high (∼ 100 m s −1 ) and thus the K2 data is especially helpful in robustly confirming the nature of the systems. 4. 1. HATS-36b HATS-36b is a hot Jupiter with a typical orbital period (P=4.1752379 ± 0.0000021 d). The star is active, which we see manifest in both the variability in the LC and the high jitter in the radial velocity measurements. Due to its high mass compared with the known population of hot Jupiters, HATS-36b lies in a relatively sparsely populated region of the mass-density relationship for gas giant exoplanets (see Figure 8). However its bulk density fits well on the mass-density sequence of gas giants.

Candidates
Secondary eclipses in the K2 allow us to robustly rule out 17 of the 25 HS-K2C7 candidates. However we have three candidates that remain active. All three candidates require future radial velocity monitoring in order to determine if they are transiting exoplanets/brown dwarfs or (blended) eclipsing binaries. However such a task is extremely difficult to due to faintness of the host stars. HATS579-048 (EPIC 215816368) is the most promising as we have an indication of a velocity variation of K=7.24 ± 2.21 km s −1 . However the host star is also the faintest of the three candidates at (V=15.835).

Outlook
This program shows the benefit of using high precision K2 space-based photometry to vet candidates identified from ground-based surveys. This concept will naturally extend to the TESS mission (Ricker et al. 2014). The primary differences will be twofold. Firstly, TESS will monitor most stars for only 27 days, about one-third the duration of the K2 campaigns. For many of our HS-K2C7 candidates such duration would result in only one or two transits to be observed. In these cases the value of combining the TESS data with ground-based monitoring is greatly enhanced. Secondly, the spatial resolution of TESS is just 21 ′′ .1 pixel −1 , meaning many blended systems that can be resolved in K2, such as the blended eclipsing binaries presented in this work, will not be readily identifiable from TESS data alone. In these cases ground based data such as that from HATSouth (3 ′′ .7 pixel −1 ) will be highly beneficial in resolving the nature of the systems.