ARTEMiS (Automated Robotic Terrestrial Exoplanet Microlensing Search) - A possible expert-system based cooperative effort to hunt for planets of Earth mass and below
M. Dominik, K. Horne, A. Allan, N.J. Rattenbury, Y. Tsapras, C. Snodgrass, M.F. Bode, M.J. Burgdorf, S.N. Fraser, E. Kerins, C.J. Mottram, I.A. Steele, R.A. Street, P.J. Wheatley, L. Wyrzykowski
aa r X i v : . [ a s t r o - ph ] J a n Astron. Nachr. / AN , No. , 1 – 4 (2007) /
DOI
ARTEMiS (Automated Robotic Terrestrial Exoplanet MicrolensingSearch) – A possible expert-system based cooperative effort to hunt forplanets of Earth mass and below
M. Dominik ,⋆,⋆⋆ K. Horne , A. Allan , N.J. Rattenbury , Y. Tsapras , C. Snodgrass , M.F. Bode ,M.J. Burgdorf , S.N. Fraser , E. Kerins , C.J. Mottram , I.A. Steele , R.A. Street , P.J. Wheatley ,Ł. Wyrzykowski , SUPA, University of St Andrews, School of Physics & Astronomy, North Haugh, St Andrews, KY16 9SS, UnitedKingdom School of Physics, University of Exeter, Stocker Road, Exeter EX4 4QL, United Kingdom Jodrell Bank Observatory, Macclesfield, Cheshire, SK11 9DL, United Kingdom Astrophysics Research Institute, Liverpool John Moores University, Twelve Quays House, Egerton Wharf, Birkenhead,CH41 1LD, United Kingdom European Southern Observatory (ESO), Casilla 19001, Santiago de Chile, Chile Las Cumbres Observatory Global Telescopes Network, 6740B Cortona Dr, Goleta, CA 93117, United States of America Department of Physics, University of Warwick, Coventry, CV4 7AL, United Kingdom Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA, United Kingdom Warsaw University Astronomical Observatory, Al. Ujazdowskie 4, 00-478 Warszawa, PolandThe dates of receipt and acceptance should be inserted laterThe technique of gravitational microlensing is currently unique in its ability to provide a sample of terrestrial exoplan-ets around both Galactic disk and bulge stars, allowing to measure their abundance and determine their distribution withrespect to mass and orbital separation. Thus, valuable information for testing models of planet formation and orbital mi-gration is gathered, constituting an important piece in the puzzle for the existence of life forms throughout the Universe.In order to achieve these goals in reasonable time, a well-coordinated effort involving a network of either 2m or 4 × c (cid:13) If an observed star in the Galactic bulge at distance D S hap-pens to be aligned with a foreground star with mass M atdistance D L within an angle θ < ∼ θ E , where θ E = r GMc (cid:0) D − − D − (cid:1) (1)is the angular Einstein radius , it exhibits a transient bright-ening (Einstein 1936) over about a month due to the bending ⋆ Royal Society University Research Fellow ⋆⋆ Corresponding author: e-mail: [email protected] of its light by the gravitational field of the foreground star,constituting a gravitational microlensing event .If moreover a planet orbiting the foreground ’lens’ starhappens to be separated by an angle δ ∼ θ E , it can re-veal its existence by means of a short distortion to the ob-served light curve (Mao & Paczy´nski 1991), depending onits mass lasting between several hours and several days. Be-sides three other current claims (Bond et al. 2004; Udalskiet al. 2005; Gould et al. 2006), microlensing led to the dis-covery of OGLE-2005-BLG-390Lb, the first cool rocky/icyexoplanet ever found (Beaulieu et al. 2006; Dominik, Horne& Bode 2006).Microlensing observations can provide samples of plan-ets orbiting stars in two distinct populations, namely the c (cid:13) M. Dominik et al.: ARTEMiS – An expert system to hunt for planets of Earth mass and below
Galactic disk and the Galactic bulge, rather than facing arestriction to the Solar neighbourhood. This will allow anestimate of the abundance of planets in the Universe andprovide a powerful test of theoretical models of planet for-mation and orbital migration.Given the properties of the two stellar populations, atypical microlensing event on an observed bulge star at D S ∼ . kpc involves a lens star with M ∼ . M ⊙ at D L ∼ . kpc. This implies that θ E ∼ µ as, so that microlens-ing is most sensitive to the detection of planets at a sepa-ration of 1 to 10 AU. Since the respective orbital period byfar exceeds the duration of the planetary signal, the latterreflects a snapshot of the planet at its current position. With only about one in a million monitored stars being sig-nificantly brightened by the gravitational field of a fore-ground star (Kiraga & Paczy´nski 1994), the OGLE (OpticalGravitational Lensing Experiment) and MOA (Microlens-ing Observations in Astrophysics) surveys monitor more than100 million stars on a daily basis, which results in 700-1000microlensing events per year being alerted on-line whilethey are in progress (Udalski et al. 1992; Muraki et al. 1999;Bond et al. 2001; Udalski 2003). Their sampling is howeverinsufficient for detecting planets with masses significantlybelow that of Jupiter (Snodgrass, Tsapras, & Horne 2004).The first microlensing follow-up network combining hourlysampling with a round-the-clock coverage was establishedby PLANET (Probing Lensing Anomalies Network) in 1995(Albrow et al. 1998; Dominik et al. 2002). While this net-work of 1m-class telescopes relies on human observers anddedicated observing time, the demand of not only an imme-diate response, but also a flexible scheduling makes robotictelescopes ideally suited to carry out such an observing pro-gramme. Since 2004 – and since 2005 in cooperation withPLANET –, microlensing observations have been carriedout with the RoboNet-1.0 network of UK-built 2m robotictelescopes (Burgdorf et al. 2007), using a priority algorithmthat selects those targets for which a detection of a planetis most likely to occur. In contrast to PLANET/RoboNet,the MicroFUN team concentrate on a few quite promisingevents, with a network only being activated on target-of-opportunity basis.With the deployment of the SIGNALMEN anomalydetector (Dominik et al. 2007), a combined effort of mi-crolensing campaigns can realize a three-step strategy ofsurvey, follow-up, and anomaly monitoring that allows for asubstantial detection efficiency even for Earth-mass planets.
On 2005 August 10, PLANET/RoboNet, OGLE, and MOAobserved a 15 % deviation to the light curve of microlensing event OGLE 2005-BLG-390 over about a day, which wasshown to be due to a planet of about 5 Earth masses, orbitinga star with . M ⊙ at 3 AU with a period of 10 years, whereall these values are uncertain to a factor of two (Beaulieuet al. 2006). An Earth-mass planet in the same spot wouldstill have caused a signal amplitude of 3 % and duration of ∼ h. If one assumes photometric uncertainties of ∼ %,the discovery of such a planet would only have been possi-ble if the standard follow-up sampling of 2 h had been re-placed by high-cadence (10–15 min) anomaly monitoringtriggered upon the first suspicion of a deviation. Real-timephotometry and a prompt response from the telescopes al-low the SIGNALMEN anomaly detector to identify ongo-ing anomalies by successively requesting further observa-tions until an anomaly can be confirmed or rejected withthe required significance. By triggering on residuals whoseabsolute value is among the largest 5 % of all data for therespective site (such trigger points being marked by arrowsin Fig. 1), and eliminating the effect of outliers by meansof robust-fitting techniques, SIGNALMEN carefully ad-dresses the fact that reported photometric error bars fre-quently do not properly represent the true uncertainties andin general do not follow a Gaussian distribution.The giant source star ( R ⋆ ∼ . R ⊙ ) that was observedin OGLE 2005-BLG-390 yielded a larger probability to de-tect a planetary signal and increased its duration, but re-duced its amplitude as compared to a main-sequence star.However, the SIGNALMEN anomaly detector would alsoallow to reveal an Earth-mass planet from a 5 % deviation ifthe observed source is a main-sequence star ( R ⋆ ∼ . R ⊙ ),provided that exposure times are chosen long enough forachieving a photometric accuracy of 1–2 %. While the mi-crolensing searches face a reliable chance of first detectingan extra-solar planet of Earth mass, the detection of plan-ets with masses as small as 0.1 M ⊕ is challenging both bymeans of the short signal duration and the tiny probabilityfor signals of appropriate amplitudes to occur, but possiblein principle (Dominik et al. 2007). The detection of a significant number of terrestrial extra-solar planets in reasonable time requires a well-coordinatedeffort involving a network of either 2m or 4 ×
1m tele-scopes at its sites and further smaller telescopes, in orderto ensure that sufficient data can be acquired on an ongoingplanetary anomaly, which could require that observationsare being scheduled at a suitable site within 10 min of analert by the SIGNALMEN anomaly detector. Comprised ofthe Liverpool Telescope (LT), the Faulkes Telescope North(FTN), and the Faulkes Telescope South (FTS), RoboNet-1.0 constituted the prototype of such a network, with thetwo Faulkes telescopes now having been acquired by LasCumbres Observatory, and further such instruments can beexpected to be deployed over the next three years. c (cid:13) stron. Nachr. / AN (2007) 3 Fig. 1
The possible detection of planets of Earth mass or below (simulations) in three different configurations (where d = δ/θ E ) with the robotic 2m-telescopes that constitute the RoboNet-1.0 network, namely the Liverpool Telescope(LT), the Faulkes Telescope North (FTN), and the Faulkes Telescope South (FTS), augmented by two similar hypotheticaltelescopes located in Chile and South Africa. Arrows indicate the epochs where the SIGNALMEN anomaly detectorrequested further observations. (top) 1- M ⊕ planet in the same spot as OGLE-2005-BLG-390Lb, with the original modellight curve for the respective event also plotted; (middle) 1- M ⊕ planet in the same spot, but with a main-sequence sourcestar; (bottom) 0.1- M ⊕ planet at a closer distance ( d = 1 . instead of d = 1 . ) to the lens star, and a main-sequencesource star. For the two latter cases, the orientation angle of the source trajectory with regard to the planet-star axis hasbeen modified from the OGLE-2005-BLG-390 model in order to produce a deviation of the desired amplitude. c (cid:13) M. Dominik et al.: ARTEMiS – An expert system to hunt for planets of Earth mass and below
Fig. 2
ARTEMiS (Automated Robotic Terrestrial Exoplanet Microlensing Search) and its interactions with the outsideworld.Acting as an expert system, ARTEMiS will determinethe optimal target to be observed by any site at any giventime by means of real-time modelling of real-time data re-leased by the microlensing observing campaigns and subse-quent assessment by a priority algorithm and the SIGNAL-MEN anomaly detector (see Fig. 2). With a flexible strategycatering for parameters that allow observing sites to definetheir preferences, tailored target recommendations match-ing specific strategic goals are provided.For scheduling its observations, RoboNet-1.0 alreadyused the novel software architecture developed by the eSTAR(e-Science Telescopes for Astronomical Research) project(Steele et al. 2002), which builds a virtual meta-network be-tween existing proprietary robotic-telescope networks pro-viding a uniform interface built upon a multi-agent contractmodel (Allan, Naylor & Saunders 2006). The embeddingof ARTEMiS into eSTAR will open a direct way of com-munication with robotic telescopes in the HTN (Heteroge-neous Telescope Networks) consortium. Moreover, usingthe standard adopted by the IVOA (International Virtual Ob-servatory Alliance) for representing, transmitting, publish-ing, and archiving the discovery of a transient celestial event(White et al. 2006), different levels of SIGNALMEN alertson potential or actual anomalies, defining the switch-overfrom follow-up to anomaly monitoring mode, will be dis-tributed as Virtual Observatory Events (VOEvents).Nevertheless, ARTEMiS also accounts for the fact thatmany of the current microlensing campaigns still rely onhuman observers. For these, additional means of communi-cation will be established, such as interactive webpages andSIGNALMEN alerts circulated as e-mail or SMS. In partic-ular, ARTEMiS will offer up-to-the minute visualization ofincoming data and model light curves, using the system thatwas previously operated for PLANET, and provide current information about ongoing anomalies. The real-time visu-alization and interpretation not only allows to link up withprofessional and amateur astronomers around the world, buteven offers an opportunity to communicate forefront researchin progress to the general public as “Science live to yourhome”.
References
Allan, A., Naylor, T., Saunders, E.S.: 2006, AN 327, 767Albrow, M.D., Beaulieu, J.-P., Birch, P., et al.: 1998, ApJ 509, 687Beaulieu, J.-P., Bennett, D.P., Fouqu´e, P., et al.: 2006, Nature 439,437Bond, I.A., Abe, F., Dodd, R.J., et al.: 2001, MNRAS 327, 868Bond, I.A., Udalski, A., Jaroszy´nski, M.: 2004, ApJ 606, L155Burgdorf, M.J., Bramich, D.M., Dominik, M., Bode, M. F.,Horne, K. D., Steele, I. A., Rattenbury, N., Tsapras, Y.: 2007,P&SS 55, 582Dominik, M., Albrow, M.D., Beaulieu, J.-P., et al.: 2002, P&SS 50,299Dominik, M., Horne, K., Bode M.F.: 2006, A&G 47, 3.25Dominik, M., Rattenbury, N.J., Allan, A., et al.: 2007, MN-RAS 380, 792Einstein, A.: 1936, Sci 84, 506Gould, A., Udalski, A., An, D.: 2006, ApJ 644, L37Kiraga, M., Paczy´nski, B.: 1994, ApJ 430, L110Mao, S., Paczy´nski, B.: 1991, ApJ 374, L37Muraki, Y., Sumi, T., Abe, F., et al.: 1999, PThPS 133, 233Snodgrass, C., Tsapras, Y., Horne, K.: 2004, MNRAS 351, 967Steele, I.A., Naylor, T., Allan, A., Etherton, J., Mottram, C.J.:2002, in: R.I. Kilbrick (ed.),
Advanced Global Communica-tions Technologies for Astronomy II , Proc. SPIE, Vol. 4845,13Udalski, A.: 2003, AcA 53, 291Udalski, A., Jaroszy´nski, M., Paczy´nski, B., et al.: 2005, ApJ 628,L109 c (cid:13) stron. Nachr. / AN (2007) 5Udalski, A., Szyma´nski, M., Kałuzny, J., Kubiak, M., Mateo, M.:1992, AcA 42, 253White, R.R., Allan, A., Barthelmy, S., et al.: 2006, AN 327, 775 c (cid:13)(cid:13)