The California Planet Survey IV: A Planet Orbiting the Giant Star HD 145934 and Updates to Seven Systems with Long-Period Planets

1.6 MB·English

by  Y. Katherina Feng

Draft version January 6, 2015
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
THE CALIFORNIA PLANET SURVEY IV: A PLANET ORBITING THE GIANT STAR HD 145934 AND
UPDATES TO SEVEN SYSTEMS WITH LONG-PERIOD PLANETS *
Y. Katherina Feng1,4, Jason T. Wright1, Benjamin Nelson1, Sharon X. Wang1, Eric B. Ford1, Geoffrey W.
Marcy2, Howard Isaacson2, and Andrew W. Howard3
(Received 2014 August 11; Accepted 2014 November 26)
Draft version January 6, 2015
ABSTRACT
We present an update to seven stars with long-period planets or planetary candidates using new
and archival radial velocities from Keck-HIRES and literature velocities from other telescopes. Our
5
updatedanalysisbetterconstrainsorbitalparametersfortheseplanets,fourofwhichareknownmulti-
1
planet systems. HD 24040 b and HD 183263 c are super-Jupiters with circular orbits and periods
0
longer than 8 yr. We present a previously unseen linear trend in the residuals of HD 66428 indicative
2
on an additional planetary companion. We confirm that GJ 849 is a multi-planet system and find a

n goodorbitalsolutionforthec component: itisa1M planetina15yrorbit(thelongestknownfor
Jup
a a planet orbiting an M dwarf). We update the HD 74156 double-planet system. We also announce
J
the detection of HD 145934 b, a 2M planet in a 7.5 yr orbit around a giant star. Two of our stars,
Jup
4 HD 187123 and HD 217107, at present host the only known examples of systems comprising a hot
Jupiter and a planet with a well constrained period >5 yr, and with no evidence of giant planets in

] between. Ourenlargementandimprovementoflong-periodplanetparameterswillaidfutureanalysis
P
of origins, diversity, and evolution of planetary systems.
E
Subject headings: planetary systems — stars: individual (HD 145934, GJ 849) — techniques: radial
.
h velocity
p
-
o 1. INTRODUCTION ter (HIRES; Vogt et al. 1994) and complementary, pub-
r lished velocities from other telescopes, where available.
t 1.1. Background
s So far, the number of planets discovered by RVs with
a The possibility of “Earth 2.0”, and especially another periods greater than 1000 days is 103, only 16 of which
[ planet that hosts life, drives much of the search for exo- have periods longer than 3000 days. This is in contrast
1 planets. As of 2014 November, efforts over two decades tothe336suchplanetswithshorterperiods.6 Ourintent
v haveuncoveredmorethan1400planetsandalmost4000 is to enlarge this sample of long-period planets to search
3 planetarycandidates(Hanetal.2014;Burkeetal.2014, forplanetarysystemswithJupiteranalogs. Ofthese103
3 ; exoplanets.org). The variety of discoveries, from lone planets, 31 are in multi-planet systems. The study of
6 Jupiter-massplanetsinfew-dayorbitstopackedsystems multi-planet systems addresses planetary formation, mi-
0 with multiple planets that fit within Mercury’s orbit, gration, and dynamics. Having a large sample can also
0 raises a significant question as to the nature of our solar contributetotheunderstandingoftheevolutionandlife-
1. system: Are we unique? time of stable planetary systems. Studies can examine
0 To search for analogs of the solar system, we target the orbital eccentricities and perform dynamic simula-
5 multi-planetsystemsandlong-periodgiantplanets,rem- tions and probe migration.
1 iniscent of our own outer solar system. Because we seek For the purposes of this discussion, we follow Wang et
: planets with orbits of at least a few hundred days, the al. (2012) and define a Jupiter analog as a planet with
v radialvelocity(RV)methodofexoplanetdetectionisad- P > 8 yr, 4 > Msin(i) > 0.5M , and e < 0.3, but we
Xi vantageous(e.g.Wright&Gaudi2013;Butleretal.1996; also adopt an upper period limJitupof P < 16 yr. Of the
Mayor & Queloz 1995). The RV method is the longest confirmed RV planets, only 13 planets fit the above cri-
r
a running, with multiple surveys studying thousands of teria(Hanetal.2014). Anothermotivationforstudying
stars. In our study, we utilize up-to-date velocities from systemswithlong-periodJupiteranalogsistherolesuch
Keck observatory’s High Resolution Echelle Spectrome- aJovianplanetmayplayinthehabitabilityofanEarth-
like planet in the same system. Wetherill (1994) argued
∗Based in part on observations obtained at the W. M. Keck thatJupiteractsasashieldthatdeflectscometsoriginat-
Observatory, which is operated by the University of California ing from the Oort Cloud or Kuiper Belt, protecting the
andtheCaliforniaInstituteofTechnology.
inner solar system. Without Jupiter, Wetherill (1994)
1CenterforExoplanetsandHabitableWorlds,Departmentof
Astronomy & Astrophysics, 525 Davey Lab, The Pennsylvania suggested an increase in the frequency of cometary im-
StateUniversity,UniversityPark,PA16802,USA; pacts on Earth by 1000 – 10,000 times the present–day
Correspondingauthor: [email protected] value. Multi-planet systems serve not only as examples
2Department of Astronomy, University of California, Berke-
ofplanet-planetinteractionbutalsoasmodelsforplane–
ley,CA94720-3411,USA
3Institute for Astronomy, University of Hawaii, 2680 Wood- comet dynamics.
lawnDrive,Honolulu,HI96822,USA
4AlsoatDepartmentofAstronomy&Astrophysics,1156High
Street,MS:UCO/LICK,UniversityofCalifornia,SantaCruz, 6 WefollowHanetal.(2014),whoadoptaupperlimitonmin-
CA95064,USA. imummassof24timesthemassofJupiter
2 Feng et al.
1.2. Plan tween the pre- and post-upgrade time series (e.g., Kane
et al. 2014); (2) we collect literature RVs for the sys-
Section 2 gives an outline of the steps taken for char-
tem from other telescopes, if available (3) we use the
acterizing the planetary systems. In Section 3, we de-
published orbital parameters (which we collect from the
scribe the planetary systems orbiting seven stars. Each
Exoplanet Orbit Database; Han et al. 2014) as initial
of these systems already has at least one planet known
guesses for the planets’ orbits; (3) we use RVLIN to fit
and exhibits RV residuals indicative of an outer com-
the system anew (with additional planets contributing
panion. Additionally, each can have its planetary orbits
five model parameters each, if necessary); (4) we use the
significantly refined with our new velocities from Keck,
and in some cases we show that an outer, decade-long reduced χ2(χ2ν) to describe the goodness of fit.
planetaryorbithasfinallycompleted. Wepresentasum- We calculate most orbital parameter uncertainties us-
maryoftheradialvelocitydata,meanuncertainties,and ing BOOTTRAN (Wang et al. 2012), which uses RVLIN and
telescope offsets in Table 1. Table 2 lists the stellar pa- a bootstrapping method to compute the distribution of
rameters of the target stars. Table 3 lists the orbital parametersconsistentwiththedata. Becauseuncertain-
parameters of the planets presented in this paper. We ties can be highly non-Gaussian for planets with incom-
also present figures showing the RV curves and residuals plete orbits, we also examine the minimum χ2 surface in
for each system. Section 4 presents an analysis of a new minimum mass-period space (Section 3).
planet,HD145934b. Wediscussourfindingsandfuture For our fits, we choose values for the jitter (Wright
prospects in Section 5. 2005, and references therein) that yield χ2 values close
ν
to 1; usually we pick a value similar to the rms of the
2. METHODOLOGY initial fit which does not incorporate jitter. If a star has
data from several (more than three) instruments taken
2.1. Radial Velocity Sources and Analysis
by multiple teams, we apply jitter on an instrument-
Wecombinepreviouslypublisheddatafromothertele- by-instrument basis. To do so, we ran the fit with no
scopes to complement the time span and quantity of assumed jitter, calculated for each instrument the stan-
Keck-HIRES observations obtained by the California dard deviation of the residuals, and added that value in
Planet Survey (Howard et al. 2010; Johnson et al. 2010; quadrature to the velocities. After that, we rerun the
Wright et al. 2011) for many purposes, including as part fit and that yielded the best-fit parameters. We utilized
oftheη⊕ survey(Howardetal.2009,2010,2011a,b). At an instrument-by-instrument jitter for HD 24040, HD
the time of the first confirmed RV planet, 51 Pegasi b 74156, and HD 217107. In general, we are confident in
(Mayor & Queloz 1995), several surveys were underway the relative instrumental uncertainties in the pre- and
andactivelymonitoringstarsforthesignsofplanets(e.g. post-upgrade HIRES data, and we use a common jitter
Cochran&Hatzes1994;Fischeretal.2014). Thediscov- value for both.7 Table 2 lists the stellar parameters of
ery team for 51 Pegasi used the ELODIE spectrograph hoststars,andTable3liststheorbitalparametersofthe
(Baranne et al. 1996), which was part of the Northern planetswediscussedbelow. CorrespondingRVplotsand
ExtrasolarPlanetSearchuntiltheSOPHIEspectrograph additional figures follow the text.
(Bouchy & Sophie Team 2006) replaced it in 2006. The Sun-like stars are known to have magnetic cycles with
CORALIE spectrograph (e.g. Queloz et al. 2000) was periodscomparabletotheperiodofJupiter(Baliunaset
situated in Chile as part of the Southern Sky extrasolar al.1995). Apersistentconcerninthehuntforlong-term
PlanetsearchProgramme. IthasbeenjoinedbyHARPS RVsignalsfromJupiteranalogshasbeenthattheymight
(Mayor et al. 2003), also located in Chile. We make bemimickedbytheeffectsofsuchmagneticcycles,which
use of literature data from all four spectrographs in this could alter convective patterns such that the magnitude
work. Other data come from High Resolution Spectro- ofthedisk-integratedconvectiveblueshiftofastarmight
graph(HRS)oftheHobby-EberlyTelescope(Tull1998), vary with the stellar cycle (Dravins 1985; Walker et al.
the Tull Spectrograph at the 2.7-m telescope of McDon- 1995; Deming et al. 1987; Santos et al. 2010).
ald Observatory (Tull et al. 1995), and the Hamilton A common way to check that magnetic effects are not
spectrograph at Lick Observatory (Vogt 1987). responsible for RV variations is to measure correlations
To analyze and fit the data, we use the Wright & between the RVs and activity indices such as Ca II H
Howard (2009) RVLIN package written in IDL that natu- & K. Previous work by Wright et al. (2008), Santos et
rally handles multiplanet systems using data from mul- al. (2010), and Lovis et al. (2011) find that the observed
tiple telescopes in systems where planet–planet interac- activity-cycle-induced RV amplitudes are typically quite
tionsarenegligiblegiventheprecisionandthespanofthe small (a few m s−1 or less), although there are sugges-
observations. Inthispackage,RVcurvesaredescribedby tions that a few stars may show abnormally high levels
both non-linear and linear parameters, and the package of correlation (at the level of 10–20 m s−1).
performs least-squares fitting on them separately. The We have checked for activity cycles in these stars to
package uses a simple linear least-squares solution for see if they have similar periods and phases to the RV
thelinearparameters,andtheLevenberg–Marquardtal- measurements. To do this, we have used the Ca II H
gorithm for the nonlinear parameters. RVLIN supplies
a sum-of-Keplerians model (plus optional secular trend 7 Theeffectofjitteronthebestfitvaluesofanorbitalsolution
and offsets between instruments) to MPFIT , the IDL im- is to give more even weight to points with different measurement
plementationoftheLMmethoddevelopedbyMarkwardt uncertainties; in the cases of the well-detected planets we discuss
(2009). in this work, the exact value of the jitter has very little effect on
thesebest-fitvalues. Becausewedeterminemostofourparameter
We fit for these new planetary system as follows: (1)
uncertaintiesviabootstrapping,ouruncertaintiesarenotstrongly
we collect Keck RVs before and after the 2004 HIRES affectedbyourchoiceofjitter,andsothereisnoneedtofindthe
upgrade separately to account for any (small) offsets be- precisejittervaluethatyieldsχ2ν =1.0.
Updates to Long-Period Planets 3
& K chromospheric activity measurements from Wright
et al. (2004, hereafter W04), Isaacson & Fischer (2010, 60
hereafterIF10),andmorerecentmeasurementsmadeus-
40
ingthesamedatastreamandpipelineasthelatterwork. s)
m/
Fboetrwseoemnetshtearms,etahsuerreemapepntesarptuoblbisehceadlibbyraWtio0n4dainffderethnoceses city ( 20
made using the IF10 pipeline, necessitating a rescaling elo 0
or application of an offset to one of the streams. This al V −20
is most apparent in “flat activity” stars which show no di
a
variationbutoccasionallyexhibitalargejumpinactivity R
−40
level between the two data streams.
Insixofourstars,thereisnoappreciableactivityvari- −60
ation(i.e.,theyare“flatactivity”stars, Saaretal.1998), 0 5 10
Time since first observation (yr)
making it very unlikely that the large RV variations we
see are due to solar-type activity cycles. The seventh
star, HD 183263 does show a significant cycle, however. 60
TheW04activitylevelsdecreasefrom2002to2004, and
40
theIF10showacontinueddecreasestartinginlate2004, s)
wvehloicchitibeoststhomows aoumtiinnimaumminiinm2u0m05inan2d01a2.mTaxhiemaucmtuianl city (m/ 20
2012, thus exhibiting a shorter period than the actual o 0
el
a2c0t1i2viatyndcytchlee.veTryhehingehgaatmivpelitcuodrreeloaftitohnebReVtwseigenna2ls00a5r–e dial V −20 11028000000
a 600
inconsistent with typical stars with RV-activity correla- R 400
−40 200
tion seen in Wright et al. (2008) and described by Lovis 0
−200
et al. (2011). It is thus very unlikely that any of the −60 0 5 10
long-period signals we describe in this work are due to 0 5 10
Time since first observation (yr)
stellar magnetic activity cycles.
60
2.2. Minimum Masses from Linear Trends Alone
In some cases, we find that a secular increase or de- s) 40
crease in the observed radial velocities is present (a “lin- m/
ear trend”), which is presumably a small portion of a y ( 20
Kepleriansignal from amassive companion, typicallyan ocit 0
outer planet, or a secondary star or brown dwarf (Crepp Vel 100
et al. 2012, 2013a,b, 2014; Montet et al. 2014; Knutson al −20 50
et al. 2014). adi 0
R
−40 −50
When the trend shows no curvature and we have no
−100
AO imagery to put limits on the mass and angular sep- −60 0 50 100 150 200
aration of companions, we usually say very little about 0 5 10
the companion beyond a minimum mass (and a maxi- Time since first observation (yr)
mum luminosity from the fact that its spectrum did not
Fig. 1.— Fifty synthetic RV measurements made over 8 yr by
complicatetheRVanalysis). Thescenariothatgivesthe
anunluckyobserverofahypotheticalsystemwithGaussianerrors
minimummasstoacompaniongeneratingalineartrend of 3 m s−1. Top: RV curve of a planet with Msini=1.97MJup,
of a given magnitude is one that has e ∼ 0.5, ω = 90◦, P = 10 yr, K = 30 m s−1, e = 0.5 and ω = 90◦. The observer
which produces a sawtooth-like RV curve with a long, might conclude, incorrectly, that they were seeing the effects of a
distant exoplanet with P (cid:29) 8 yr and K (cid:29) 30 m s−1. Middle:
nearly linear component for ∼ 80% of the orbit with a
brief, high-acceleration component during periastron for R34V5cmursv−e1o,fea=pl0a.n9e7tawnidthωM=si2n0i◦.=A6l.t6hMouJugph,tPhe=pe1r1iodyr,PKan=d
the other ∼ 20% (Wright (2006), and see top panel of
magnitude of the observed trend are about the same as that in
Figure 1 for an example; other panels show other patho- thetoppanel,thetruesemi-amplitudeoftheorbitismuchlarger.
logical cases with radically different periods and semi- The inset illustrates the complete RV curve, with the same units
amplitudes that mimic the same trend). asthemainfigure. Bottom: RVcurveforhypotheticalplanetwith
Theminimummassofaplanetarycompaniondetected Msini = 11.7MJup, P = 200 yr, K = 65 m s−1, e = 0.5 and
◦
ω = 270 . Although the magnitude of the observed trend is the
only by its strongly detected constant acceleration γ˙,
sameasthatinthetopandmiddlepanels,theperiodinthiscaseis
is thus derived by solving the mass function for the
muchlonger,whileK isonlymodestlylarger. Theinsetillustrates
minimum mass (e.g., Wright & Howard 2009) assuming the RV curve, with the same units as the main figure, over more
P ∼ 1.25τ (where τ is the span of the observations), than a complete orbit. The box in the inset illustrates the span
of the main panel. All three panels are reproduced from Wright
e∼0.5, and K ∼τγ˙:
(2006).
Mminimum ≈(0.0164MJup)(cid:18)yτr(cid:19)4/3(cid:12)(cid:12)(cid:12)(cid:12)ms−1γ˙yr−1(cid:12)(cid:12)(cid:12)(cid:12)(cid:18)MM∗(cid:19)2/3.
(cid:12)
(1)
4 Feng et al.
3. REFINEDORBITALPARAMETERSFORSEVEN
PLANETARYSYSTEMS 150 HD 24040 b
Our sample includes many known planetary systems
100
of interest because of the presence of a linear trend in s)
m/
theresidualsindicativeofanadditionalcompanion;some y ( 50
with known trends with significant curvature; and some cit
o
knowntohaveoutercompanionswithpoorly-constrained el 0
V
parameters. The first six, HD 24040, HD 66428, HD al
74156, HD 183263, HD 187123, and HD 217107 are G di −50 SOPHIE
a
stars; the seventh GJ 849, is an M dwarf. R Keck (pre−2004)
Keck (2004−)
Table 1 presents the time span of sets of observations, −100 ELODIE
the number of points from each set, the number of new
2000 2005 2010 2015
points, the mean uncertainty in velocities from each set,
Date (year)
and the offsets between instruments.
(a) HD24040b
3.1. HD 24040
Wright et al. (2007) reported a substellar compan-
80
ion to the star HD 24040 with a wide range of possi-
ble periods (10 yr < P < 100 yr) and minimum masses 60
SOPHIE
(5<Msini<20M ). Boisse et al. (2012), combining
Jup Keck (pre−2004)
velocities from HIRES, SOPHIE, and ELODIE, deter- m/s) 40 Keck (2004−)
minedanorbitof3668+−116791days(correspondingto 10yr) al ( 20 ELODIE
andaminimummassof4.01±0.49M forHD24040b. u
Jup d
Boisse et al. (2012) also found a linear trend of 3.85+1.43 esi 0
−1.29 R
m s−1 yr−1, indicative of a third body in the system.
−20
Boisse et al. (2012) also investigated potential long-term
correlation between SOPHIE measurements and stellar −40
activity indices but did not find such behavior. 2000 2005 2010
We present an updated fit with more recent Keck- Date (year)
HIRES velocities, seen in Figure 2 and Table 4. We use
(b) Residuals
HIRES data and published SOPHIE and ELODIE data,
so in our fit we applied jitter instrument-by-instrument. Fig. 2.—RadialvelocityandKeplerianfitforHD24040b. Solid
With107velocitiesintotal,47ofwhicharefromHIRES, lines represent the best-fit Keplerian orbits. The fit includes a
lineartrendof1.8±0.4ms−1 yr−1.
13 from SOPHIE, and from 47 ELODIE (Boisse et al.
2(a)Keck,SOPHIE,andELODIERVsoverplottedbybest-fitone-
2012),wefindforthebest-fitone-planetKeplerianmodel planetKeplerianmodel. 2(b)ResidualsoftheRVswiththebest-fit
anrmsof13.62ms−1 andχ2 of0.93. HD24040b orbits one-planetKeplerianmodelsubtracted.
ν
at a semimajor axis of 4.637 ± 0.067 AU, correspond-
m s−1 yr−1 (corresponding to a minimum mass for the
ing to a period of 9.5 yr, making it a good Jupiter ana-
outercompanionofatleast1.77M ,byEquation(1)).
log in terms of its orbit (however, its minimum mass is Jup
We run the fit with no jitter and no trend in order
4.10±0.12M ). Thelineartrendis1.8±0.4ms−1 yr−1
Jup to see the significance of the detected trend. For that
(lower than reported in Boisse et al. 2012), a minimum
case, χ2 is 52.56, and the rms of the residuals is 7.46
mass of at least 1.44 M according to Equation (1). ν
Jup m s−1. To compare, we found an rms of 3.14 m s−1
Our fit for HD 24040 b, with a period of 3490±25 days
and χ2 of 8.23 for seven free parameters (including the
andminimummassof4.10±0.12MJup, isingoodagree- ν
trend) and no jitter. Given the improvement in the fit
ment with the solution from Boisse et al. (2012).
with a trend included, the trend is significant. We also
3.2. HD 66428 note that the eccentricity of the orbit is large: 0.442±
0.016. Thetrendmayindicatethattheoutercompanion
Butler et al. (2006) announced HD 66428 b, a planet
has influenced the orbit of the b component. Further
with P = 1973±31 d (5.4 yr), e = 0.465±0.030, and
monitoringwilldeterminethenatureofthesourceofthe
Msin(i) = 2.82±0.27M . We update the orbital pa-
Jup trend (i.e., whether it is due to a stellar or planetary
rameters with a total of 55 velocities from HIRES (see
companion).
Figure 3). The original fit used 29 velocities taken with
HIRESfrom2000to2006. Ournewfitadds26newdata
3.3. HD 74156
pointsthroughlate2013. Capturingtwocompleteorbits
ofHD66428b,thefithasanrmsof3.14ms−1 wherewe Naef et al. (2004) described the HD 74156 two-planet
assumedajitterof3ms−1 andχ2 of0.96. Wedetermine systemasa1.86±0.03M planetina51.64±0.011day
ν Jup
a period of 2293.9±6.4 days, or 6.3 yr. We determine period with a 6.17 ±0.23M outer companion in a 5.5
Jup
a minimum mass of 3.195±0.066M , which is more yr orbit. Multiple authors have suspected a third planet
Jup
massive than reported in Butler et al. (2006). in the system. Barnes & Raymond (2004) predicted one
Given our larger set of radial velocities, it is under- based on the Packed Planetary System hypothesis, and
standable that our solution does not match with the so- Bean et al. (2008) claimed the discovery of a compan-
lution announced in Butler et al. (2006). The final fit ion with P = 336 days as the third planet. Based on
finds a previously unreported linear trend of −3.4 ± 0.2 analysis of RV jitter, Baluev (2009) questioned the va-
Updates to Long-Period Planets 5
50
HD 66428 b 200 Keck (pre−2004)
Keck (pre−2004) HD 74156 Keck (2004−)
m/s) Keck (2004−) m/s) 100 HELROSDIE
ocity ( 0 ocity ( 0 CORALIE
el el
V V −100
al −50 al
Radi Radi −200
−100 −300
2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014
Date (year) Date (year)
(a) HD66428b (a) HD74156system
15
60 Keck (pre−2004)
Keck (2004−)
s) 10 KKeecckk ((p2r0e0−42−0)04) m/s) 40 HELROSDIE
sidual (m/ 05 Velocity ( 20 CORALIE
Re al 0
di
−5 a
R −20
−10
−40
2000 2002 2004 2006 2008 2010 2012 2014
2000 2002 2004 2006 2008 2010 2012 2014
Date (year)
Date (year)
(b) Residuals
(b) Residuals
Fig. 3.—RadialvelocityandKeplerianfitforHD66428b,with
a trend of 3.4 m s−1 yr−1 incorporated. Solid lines represent the
best-fit Keplerian orbits. 3(a) Keck RVs overplotted by best-fit 50 HD 74156 b
one-planet Keplerian model. 3(b) Residuals of the RVs with the
best-fitone-planetKeplerianmodelsubtracted.
s) 0
m/
lsiydsitteymoaftHicDe7rr4o1r5s6f“rodm”aHsRaSfa.lsWeidtetteencmtiyoenrdeuteatlo. a(n20n0u9a)l city ( −50 Keck (pre−2004)
concluded that the third planet was unlikely to be real, Velo −100 KHeRcSk (2004−)
andMeschiarietal.(2011)updatedthesystemwithfur- al ELODIE
ther observations and reached the same conclusion. adi −150 CORALIE
R
Here, we combine 226 velocities from CORALIE and
−200
ELODIE(44and51observationsNaefetal.2004), HRS
(82 Bean et al. 2008), and HIRES (52) (see Figure 4).
0.0 0.2 0.4 0.6 0.8 1.0
We apply a two-planet Keplerian model. We added jit- Phase
ter instrument-by-instrument, and our fit has an rms of
11.03 m s−1 and χ2 of 0.97. We have captured at least (c) HD74156b
ν
two orbits of HD 74156 c, making our orbital solution
more robust than previously reported solutions. Table 3
lists the orbital parameters. HD 74156 c is one of the 200
HD 74156 c
more massive planets we have examined, with minimum
mass7.997±0.095M . Bothplanetshavelargeorbital s) Keck (pre−2004)
eccentricities (e = 0.6Ju4pand e = 0.38 for b and c respec- y (m/ 100 KHeRcSk (2004−)
tively). cit ELODIE
In Figure 5 we plot the Lomb–Scargle periodogram elo CORALIE
V 0
(Scargle 1982; Horne & Baliunas 1986) of the residuals al
to our best two-planet fit. There is no indication of any di
a
power at the period of the purported d component, a R
−100
result which is consistent with prior refutations of this
signal (indeed, our analysis here uses much of the same
0.0 0.2 0.4 0.6 0.8 1.0
data as previous work on the topic). Indeed, there is no
Phase
hint of significant power at any period, indicating that
there is no detectable third planetary companion in this (d) HD74156c
Fig. 4.— Radial velocity and Keplerian fits for the HD 74156
system. Solidlinesrepresentthebest-fitKeplerianorbits.
4(a)CORALIE,ELODIE,HRS,andKeckRVsoverplottedbybest-
fittwo-planetKeplerianmodel. 4(b)ResidualsoftheRVswiththe
best-fittwo-planetKeplerianmodelsubtracted. 4(c)and4(d): the
RVcurvesforHD74156b andc,respectively.
6 Feng et al.
15
99% FAP HD 183263
s) 100 Keck (pre−2004)
m/ Keck (2004−)
10 y (
wer ocit 0
o el
P V
5 al
adi −100
R
0 −200
1 10 100 336 1000 10000 2002 2004 2006 2008 2010 2012 2014
Period (days) Date (year)
Fig. 5.— Periodogram of the residuals to our best two-planet, (a) HD183263system
five-instrumentfittotheRVdataforHD74156. Thereisnoindi-
cationofsignificantpoweratanyperiod,orofanypoweratallat
336days,theperiodofthepurportedbutdisprovendcomponent. 15
Wehavecomputedthe99%falsealarmprobabilityinthisfigureby Keck (pre−2004)
calculating the highest peak in each of 10,000 such periodograms 10 Keck (2004−)
calculatedforsyntheticdatasetsofRVresiduals(e.g.,Howardet
al. 2009). We calculated each of the 10,000 synthetic sets by ran- s) 5
domlyassigningtheactualresiduals(drawnwithreplacement)to m/
e9a9c%hooffctahseesttimheestaollfesotbspeeravkathioandspoofwtehrebealcotwua1l3o.5b.servations. In ual ( 0
d
system. si
e −5
R
3.4. HD 183263
−10
First reported by Marcy et al. (2005), the HD 183263
−15
system showed a residual linear trend in addition to a
2002 2004 2006 2008 2010 2012 2014
3.7MJup planetina634-dayperiod. Wrightetal.(2007) Date (year)
attributedthenewandsignificantcurvatureintheresid-
(b) Residuals
uals to an outer companion. Wright et al. (2009) fol-
lowed up and constrained the minimum mass (3.57 ±
0.55 M ) and period (8.4 ± 0.3 yr) for the outer com-
Jup 100
panion,HD183263c,towhichwereportanupdatedset
of parameters.
With 66 velocities from HIRES, we implemented a fit s) 50 HD 183263 b
m/
wmitsh−a1.n rFmigsuorfe36.68prmesesn−t1satnhde aRnVascsuurmveesdfjoitrtetrheofs3y.s2- city ( 0 Keck (pre−2004)
tem as well as the residuals. The orbit for HD 183263 elo Keck (2004−)
c appears to have finally closed, and it is significantly al V −50
closertocircular(e=0.051±0.010)andhasalongerpe- di
a
riod than the solution from Wright et al. (2009), which R −100
found e=0.239±0.64 and P ∼8.5 yr. We find for HD
183263 c, that P = 4684±71 days, or 9.1 yr; Msini is −150
0.0 0.2 0.4 0.6 0.8 1.0
6.90±0.12M . Whileourbestfitorbitalsolutiondoes
Jup Phase
not match well with the previous orbital solution, our
solution resides comfortably within the stable portion in (c) HD183263b
the P –M sini space found by Wright et al. (2009, see
c c c
their Figure 3).
3.5. HD 187123
Butler et al. (1998) discovered HD 187123 b, a s) 50
m/
0.52MJup planet in a 3-day orbit. After many years of y (
continuedmonitoringofthissystem,Wrightetal.(2007) cit
announced a long-period outer companion with P > 10 elo 0
yrandaminimummassbetween1.5MJup and10MJup. al V HD 183263 c
Wrightetal.(2009)presentedasolutionthatconstrained di
a
the mass and period of an outer companion to within R −50 Keck (pre−2004)
Keck (2004−)
20%,withP =10.4±1.2yrandMsini=2.0±0.3M .
Jup
Figure 7 shows an updated fit with HIRES data. Naef
0.0 0.2 0.4 0.6 0.8 1.0
et al. (2004) provide ELODIE velocities; however, since
Phase
they have significantly worse precision and do not add
temporal coverage, we do not use them here. The 108 (d) HD183263c
Fig. 6.— Radial velocity and Keplerian fits for the HD 183263
system. Solidlinesrepresentthebest-fitKeplerianorbits.
6(a)KeckRVsoverplottedbybest-fittwo-planetKeplerianmodel.
6(b) Residuals of the RVs with the best-fit two-planet Keplerian
model subtracted. 6(c) and 6(d): the RV curves for HD 183263 b
andc,respectively.
Updates to Long-Period Planets 7
Keck observations still cover multiple orbits of the plan-
ets; assuming a jitter of 2.23 m s−1, we find an rms of 100
2.66 m s−1. From our fit, the period of HD 187123 c is
9H.D1±108.71132y3rcanadptpheearmsintoimbuemamJauspsitiser1.a8n1a8l±og0,.0a3l5thMoJuugph. m/s) 50
its orbit is somewhat eccentric at e=0.280±0.022. city ( 0
o
el
3.6. HD 217107 V −50
al
Fischer et al. (1999) presented HD 217107 b as a 1.27 di
a
MJup planetina7.12-dayperiod. Afewyearslater,Fis- R −100 HD 187123 Keck (pre−2004)
cheretal.(2001)identifiedalineartrendintheresiduals,
Keck (2004−)
which was likely caused by an outer companion. Vogt et −150
al. (2005) reported the first orbit for HD 217107 c, mod- 2000 2005 2010
Date (year)
estly constrained at P = 8.6 yr and Msini = 2.1M .
Jup
Wright et al. (2009) constrained the orbit and mass of (a) HD187123system
HD 217107 c to almost within 10%, with P ∼ 11.7 yr
and the minimum mass ∼2.6M .
Jup
AswiththecaseofHD74156,wealsohavedatataken 15
Keck (pre−2004)
by different teams from several instruments, we added
Keck (2004−)
jitterinstrument-by-instrument. Inourfit,weuseveloc- s) 10
ities from Keck (128 observations), Lick (Wright et al. m/
2009, 121), and CORALIE (63 Naef et al. 2001) to find city ( 5
a fit an rms of 10.29 m s−1 (see Figure 8). elo
Because the outer planet has only barely (apparently) V 0
al
completed an orbit, its orbital parameters may be es- di
a
pecially uncertain (and are particularly sensitive to the R −5
assumptionthatthereisnotathird,longer-periodplanet
contributing significantly to the velocities). To explore −10
therobustnessofourderivedorbitalperiodoftheccom- 2000 2005 2010
Date (year)
ponent as a function of its minimum mass, we have con-
structed a χ2 map in P–Msini space (a variety of what (b) Residuals
Knutson et al. (2014) call “Wright diagrams”; see Pa-
tel et al. (2007), Wright et al. (2009) and similar ap-
proaches taken in, e.g., Dumusque et al. (2011); Boisse 100
et al. (2012)). In this map all orbital parameters have HD 187123 b
bhoeeondoinptiamχiz2edm(ini.iem.,utmheysenasree)aftortheeairchmpaaxiirmoufmPlikaenlid- m/s) 50 Keck (pre−2004)
Mcsinic in the map (except for the offsets amoncg the city ( Keck (2004−)
four instruments, which are held constant at their over- o 0
el
all best-fit values). V
Figure9showstheχ2 contourmap, revealingthatthe dial
a −50
orbital period and minimum mass for HD 217107 c are R
well constrained with P =14.215+0.045 yr and Msini=
−0.04
4.51+0.07M . These uncertainties are roughly consis- −100
−0.02 Jup 0.0 0.2 0.4 0.6 0.8 1.0
tent with the uncertainties determined via bootstrap-
Phase
ping, which yields P = 14.215±0.06 yr and Msini =
4.51±0.07M . This validates our choice of stellar jit- (c) HD187123b
Jup
ter for this star, since the contours in the χ2 maps are
sensitive to the choice of jitter, while the bootstrapping
uncertainties are almost completely independent of it. 30
We report the bootstrapping uncertainties in Table 3.
To test the importance of our assumption that there s) 20
m/
are only two planets contributing detectable accelera- y ( 10
tionstothestar,werepeatedourbootstrappinganalysis cit
withamodelthatincludesanadditional, lineartrendto elo 0
V
tshuechdaatat.reTndhoiunghoutrhemreodiseln,ogisvtiantgistoicuarlmneoeddeltothinecflruedee- dial −10 HD 187123 c
a
dom to include one could, in principle, affect the best-fit R −20 Keck (pre−2004)
parameters for the outer planet. Indeed, though the pa- −30 Keck (2004−)
rameters of the b component do not change significantly
0.0 0.2 0.4 0.6 0.8 1.0
in this model (as expected given its high frequency) we
Phase
find a slightly different best-fit with such a model, with
P ,T ,andK allchangingby2–4σ,resultinginamin- (d) HD187123c
c p,c c
Fig. 7.— Radial velocity and Keplerian fits for the HD 187123
system. Solidlinesrepresentthebest-fitKeplerianorbits.
7(a)KeckRVsoverplottedbybest-fittwo-planetKeplerianmodel.
7(b) Residuals of the RVs with the best-fit two-planet Keplerian
model subtracted. 7(c) and 7(d): the RV curves for HD 187123 b
andc,respectively.
8 Feng et al.
imum mass for the outer companion of 4.37M . The
Jup
uncertainties on the parameters of the c component in 300 HD 217107 Keck (pre−2004)
the with-trend model are larger by a factor of 2–4, com- Keck (2004−)
fortably including most of the parameter estimates from s) 200 CORALIE
the no-trend model. We conclude that our choice not to m/ Hamilton
include a linear trend does not have a large effect on our city ( 100
conclusions or parameter estimations. o
Vel 0
3.7. GJ 849 al
adi −100
3.7.1. Orbital Fit R
Unlike the other stars in this work, GJ 849 is an M3.5 −200
dwarf. Various studies of this star’s composition have
2000 2005 2010
all found similar, super-solar abundances: Rojas-Ayala
Date (year)
et al. (2012) find [Fe/H] = 0.31±0.17 (from K-band fea-
tures); O¨nehag et al. (2012) find 0.35±0.10 (using J- (a) HD217107system
band); and Terrien et al. (2012) found 0.31±0.12 (using
K-band).
40
GJ 849 hosts the first planet discovered orbiting an
M-dwarf with a semi-major axis greater than 0.21 AU.
Butler et al. (2006) announced GJ 849 b, with P =5.16 m/s) 20
eyvriadnendcmeoinfiamluinmeamratrsesn0d.8o2fM−J4u.p7.5Amtst−h1eytirm−1e,,itnhdeirceatwivaes city ( 0
o
of a second companion. Bonfils et al. (2013) also fitted el
V −20
the system with one planet and a linear trend of −4.0 al
m s−1 yr−1, adding their HARPS data to the published adi KKeecckk ((p2r0e0−42−0)04)
R −40
HIRES velocities. CORALIE
Stellar magnetic activity had to be ruled out as the Hamilton
−60
source of the trend. Gomes da Silva et al. (2012) mon-
2000 2005 2010
itored several M-dwarfs from the HARPS program for Date (year)
long-term magnetic activity. For GJ 849, they saw mild
correlation in our velocities with the Na I index data. (b) Residuals
However, the amplitude was not large enough. Mon-
tet et al. (2014) provided the first orbital parameters for
200
on GJ 849 c, finding Msini = 0.70 ± 0.31M , and
Jup HD 217107 b
P =19.3+17.1 yr, and found no correlation between stel-
lar magn−et5i.c9 activity and the long-period signal of this m/s) 100 KKeecckk ((p2r0e0−42−0)04)
outer companion. y ( CORALIE
Ourfit,using35velocitiesfromHARPS(Bonfilsetal. cit Hamilton
o
2013) and 82 velocities from HIRES spanning from 1997 el 0
V
through early 2014, has further constrained the orbital al
parametersoftheGJ849system. Weincorporateajitter di
a
of 3 m s−1, and our fit has an rms of 3.72 m s−1. R −100
GJ849b isa0.911M planetina5.27yrperiodwith
Jup
an orbital eccentricity of 0.038. GJ 849 c is a 0.944±
0.0 0.2 0.4 0.6 0.8 1.0
0.07M planetina15.1±0.66yrperiodwithanorbital
Jup Phase
eccentricity of 0.087±0.06.
GJ 849 c has the longest robustly measured orbital (c) HD217107b
semimajor axis of any planet orbiting an M dwarf dis-
covered to date. Indeed, it has one of the longest well-
measuredperiodsofexoplanetsorbitinganykindofstar. 150 HD 217107 c
Keck (pre−2004)
Exoplanets with similar period and period uncertainties Keck (2004−)
intheliteratureinclude55Cncd(Marcyetal.2002;Endl s) 100 CORALIE
m/ Hamilton
eatl.a2l.01230)1;2)a;nHdDH1D6617329431bba(nHdoHwDard21e9t07a7l.b2(0M10a)rm—ierbuett city ( 50
these all orbit stars with M > 0.8M(cid:12) and the two from elo
Marmier et al. show significant eccentricity. The exo- al V 0
planetwiththelongestrobustlymeasuredorbitalperiod di
a
is β Pictoris (P =20.5+2.9 yr Macintosh et al. 2014). R −50
−1.4
We estimated the model parameters for GJ 849 in two
additional ways to check for consistency and robustness. −100
0.0 0.2 0.4 0.6 0.8 1.0
It is unclear whether the bootstrap resampling proce-
Phase
dure provides an accurate estimate of GJ 849 c’s orbital
parameters. In particular, the poor phase coverage be- (d) HD217107c
Fig. 8.— Radial velocity and Keplerian fits for the HD 217107
system. Solidlinesrepresentthebest-fitKeplerianorbits.
8(a)KeckRVsoverplottedbybest-fittwo-planetKeplerianmodel.
8(b) Residuals of the RVs with the best-fit two-planet Keplerian
model subtracted. 8(c) and 8(d): the RV curves for HD 217107 b
andc,respectively.
Updates to Long-Period Planets 9
The similarity of the parameter uncertainties from all
three methods verifies that the orbit of GJ 849 c is well
4.7 constrained and validates the BOOTTRAN and χ2 map
approaches(inthiscase)andourchoiceofjitter. Weuse
M)Jup 4.6 tehteermunorceerctoaninsteirevsaitnivTeaχb2leco3n.tours do determine param-
n i ( 4.5
si 3.7.2. Stability
m
Because this system is not “highly hierarchical”
4.4
(Wright 2010) in mass or orbital period, we have per-
formed n-body simulations to establish the dynamical
4.3
stability of our orbital solutions. The 1000 BOOTTRAN re-
14.0 14.1 14.2 14.3 14.4
Period (yr) alizations of the GJ 849 RV data are used to determine
parameteruncertaintiesisassociatedwithacompleteset
of Keplerian orbital parameters for the two planets (P,
Fig. 9.— Best-fit 100×100 χ2 map for fixed values of Pc and e,ω,K,andTp foreachplanet,plusanoverallRVoffset
Mcsinic for HD 217107 c. This confirms that the period and γ and two offsets among the three RV data sets). All
mass are well-constrained. We have illustrated the contours of
the 1σ, 2σ, and 3σ (defined by χ2 = χ2 +{2.30,6.17,11.8}) of these realizations returned reasonable fits, indicating
confidencelevels,basedonforthenumbermoifndegreesoffreedomin that the fitting procedure did not fail in any case.
theproblem(Pressetal.2002). Thecenterand1σ limitsinboth We performed long-term dynamical integrations for
parametersareconsistentwiththebootstrappinguncertaintiesfor all 1000 fits to these realizations of the data using the
theseparameters.
MERCURY symplectic integrator (Chambers 1999). Each
simulation runs for 107 orbits of the inner-most planet
fore 2001 (see Figure 10) results in several clear outlier
(∼ 2×1010 days). This integration timescale is short
models in the joint parameter distributions.
relative to the lifetime of the star but sufficiently long
In our first check for accuracy in the parameters and
enough to show a significant fraction of our models un-
uncertainties,aswithHD217107c,weconstructedaP–
dergo an instability, described below.
Msini χ2 map to confirm that the orbital period of GJ
An instability occurs if at any point during the inte-
849 c is well constrained (assuming no additional plan-
gration either planet crosses the other’s Hill sphere or
ets and a stellar jitter of 3 m s−1), despite having just
either of the planets’ semi-major axes change by more
completedanorbit,andfindthatthe68%confidencein-
than 50% of their initial value.
terval contours corresponds to uncertainties in P of less
None of our models resulted in a collision over the
than 5%. As Figure 11 shows, the χ2 map uncertain-
course of the integration. However, we find 67 models
ties in minimum mass are 0.07 M , exactly consistent
Jup undergo the second listed mode of instability (| [a −
with our bootstrapping errors; the uncertainties in pe- final
a ]/a |>0.5)whentheperiastronpassageofGJ
riodare∼1.1yr, whichislargerthanthebootstrapping initial initial
849 c is less than 3.5 AU (Figure 12). The instability
errors of 0.66 yr, probably because the χ2 contours are
times are logarithmically uniform from ∼10 to ∼107 yr.
asymmetric.
We removed the unstable BOOTTRAN realizations from
As a second check, we turn to a Bayesian approach
our calculations of the uncertainties in the orbital pa-
for performing parameter estimation via Markov chain
rameters we report in Table 3.
Monte Carlo. We adopt the usual broad priors for Ke-
plerian orbital parameters and likelihood assuming un-
correlated,Gaussianmeasurementerrorswithdispersion 4. A2MJUP PLANETAROUNDHD145934
based on the quadrature sum of the reported measure- We here announce a new long-period planet orbit-
ment uncertainties and an unknown jitter term (Ford ing the giant star HD 145934, a 1.748±0.105 M(cid:12) star
2006). Given the potential for mutual planetary inter- (Takeda et al. 2007). This star was not known to be
actions, we apply RUN DMC8, a well tested code that a giant when the California Planet Survey began mon-
combines n-body integration with differential evolution itoring it in 1997 at Keck Observatory. Since then, its
Markov chain Monte Carlo (Nelson et al. 2014). Al- log(g)valueandmassfromTakedaetal.(2007)indicate
though the GJ 849 planets are well approximated by that it is a giant. Visual inspection of the gravity sensi-
Keplerian orbits, the differential evolution proposal in tive sodium and magnesium lines confirm this diagnosis.
RUNDMCismuchmoreefficientthanatraditionalran- Radial velocities of HD 145934 show a clear sinusoidal
dom walk MCMC for dealing with correlated parame- modulation of planetary amplitude upon a large linear
ters, which are often present in the parameters for long- trend, indicative of a stellar binary companion.
period companions, and so by using RUN DMC we do In our analysis of the 75 HIRES velocities for HD
not have to fine tune a proposal distribution. We find 145934, we note the slight overall curvature present (see
that the marginal posterior probability distribution for Figure13(d)). ToaccountforthecurvatureusingRVLIN,
P has 68% of its mass within 0.74 yr of the median pe- which(atthemoment)onlyaccommodatespurelylinear
c
riod of 15.1 yr, only slightly larger than the uncertainty trends, we treated HD 145934 as a two-companion sys-
estimated from the bootstrap. tem, with the outer companion having a very long (60
yr)orbitalperiodandcircularorbit. Thereisnotenough
8 We used the Keplerian parameter priors given in Nel- informationinourtimeseriesfortheresultingorbitalpa-
son et al. (2014), and the algorithmic parameters n =300, rameters of the outer companion to be meaningful, but
chains
ngen=100,000,σγ=0.01,andMassScaleFactor=1.0. this approach provides us sufficient flexibility to fit out
10 Feng et al.
the low-frequency power contributed by the binary com- We confirm GJ 849 c, and find that it is the planet
panion. Equation (1) constrains the minimum mass of with the longest known period around an M dwarf so
the companion to be at least 21M . far. GJ 849 is a rare system in that it is a multi-giant-
Jup
Todeterminetheeffectsofmodelingtheostensiblestel- planet system around an M-dwarf. In all of our multi-
lar companion with our choice of orbital parameters on planet systems, the inner planet is less massive, though
the planet’s parameters, we checked first the impact of this fact is certainly influenced by the soft decrease in
letting eccentricity be a free parameter. The best-fit ec- semiamplitude with orbital distance (K ∝ a−1/2). HD
centricity is close to circular ( 0.05), so our choice of 66428maybeacasewheretheplanet’shigheccentricity
fixed e = 0 is not strongly affecting our analysis. We andthepresenceofalineartrendinthesystemaresigns
also changed the (fixed) period of the stellar companion theoutercompanionhasaffectedtheinnerplanet’sorbit,
to take values between 50 and 80 yr (guesses outside the as Kane et al. (2014) found in the case of HD 4203.
range returned poor fits, but given the nonlinear nature Allofthesesystems,butperhapsespeciallythe“highly
of the problem this does not necessarily reflect an actual hierarchical” systems (Wright 2010) HD 187123 and HD
upper limit to the companion’s period). We found that 217107, will be valuable for reconciling observations and
the choice of period did not have significant impact on the theory of planetary migration. These two systems
the parameters of the planet. For example, the best-fit are at present the only known examples of systems con-
valuesfortheperiodofHD145934b variedontheorder taining a hot Jupiter (gas giant with P < 10 days and
of 10 days for different outer companion periods. The Msini>0.1M )andavery-long-periodplanet(P >5
Jup
minimum mass varied on the order of 0.1 MJup. These yr)withawelldeterminedorbit. Inbothcases,theouter
differences are all well within 1σ of our presented set of planetis∼3timesthemassoftheinnerplanet,andthere
parameters. Weconcludethatourmodelingoftheouter is no evidence of other planets in the system.
companion is sufficiently flexible to have no important ThereareonlytwoothersystemswithhotJupitersand
effects on our estimates of the planet’s orbital parame- well-constrained long-period (P > 1 yr) outer planets:
ters. HIP 14810 (Butler et al. 2006; Wright et al. 2009) and
Given that the rms of the residuals to the fit without HAT-P-13 (Bakos et al. 2009; Winn et al. 2010). The
stellar jitter is 7.83 m s−1, we assume a stellar jitter of former case remains anomalous in that the innermost
7.5 m s−1 in our fit. Hekker et al. (2006) performed a planet is the most massive, with Msini=3.9M (the
Jup
survey of stable K giants with jitters lower than 20 m outermost planet has Msini=0.6M and P =2.6 yr;
Jup
s−1. The most stable of that sample range between 6 there is also a third, intermediate planet in the system).
and 15 m s−1, so our choice of jitter is reasonable and Thelattercasehasanespeciallyhighmassratio, having
also consistent with the residuals. The residuals to the a highly eccentric Msini > 14M outer planet and
Jup
resultingbest-fitKeplerianmodelhavermsof7.80ms−1 an inner, transiting planet with M = 0.86M . We
Jup
and χ2 of 1.05. We find that HD 145934 b has a period know from both RV studies (Wright et al. 2009) and the
ν
of 7.48 ± 0.27 yr, an orbital eccentricity of 0.053+0.053 , Kepler results (Latham et al. 2011) that “hot Jupiters
−0.063
andasemi-amplitudeof22.9±2.6ms−1. Theminimum arelonely”,atleastwhenitcomestocompanionswithin
mass of the planetary companion is 2.28 ± 0.26 M . ∼ 1 AU. Continued long-term monitoring of other hot
Jup
The presence of curvature in the binary companion’s Jupiterswillestablishwhethertheyhavefrequentlyhave
orbit implies that either it is highly eccentric and near “cold friends” at larger orbital distances (e.g., Knutson
periapse, or that we have observed a nonnegligible por- et al. 2014).
tion of its orbit. The latter is more likely, and implies
that its orbital period is a several or dozens of decades,
We thank the many observers who contributed to the
not millennia.
Lick and Keck-HIRES measurements reported here, es-
pecially John Johnson, Debra Fischer, Steven Vogt and
5. DISCUSSION R. Paul Butler. We gratefully acknowledge the efforts
Ouranalysisof13exoplanetsusesrecentKeck-HIRES and dedication of the Keck Observatory staff, especially
radial velocities and other published data. We see that Scott Dahm, Hien Tran, Grant Hill, and Gregg Dopp-
there is need for follow-up work, as in the cases of GJ mannforsupportofHIRESandGregWirthforsupport
849 and HD 145934 for better constraints and further of remote observing.
analysis. In the instance of HD 66428, whose residuals WethankNASA,theUniversityofCalifornia, andthe
show a previously unseen linear trend, we will monitor University of Hawaii for their allocations of time on the
forthecompletionoforbitsortorulethecompanionout Keck I telescope. Data presented herein were obtained
as a planet. at the W. M. Keck Observatory from telescope time al-
We have reduced the uncertainties in the parameters located to the National Aeronautics and Space Admin-
formanyplanets. Theup-to-dateHIRESdataallowedus istration through the agencys scientific partnership with
to place upper limits or constrain several orbits. From the California Institute of Technology and the Univer-
our sample, we identify two planets as Jupiter analogs sity of California. The Observatory was made possible
aroundSun-likeduetosimilaritiesinsemimajoraxis(5.2 by the generous financial support of the W. M. Keck
AU): HD 24040 b and HD 187123 c, although both are Foundation. We wish to recognize and acknowledge the
much more massive than Jupiter, and the latter’s orbit verysignificantculturalroleandreverencethatthesum-
issomewhateccentric. Wehavediscoveredanewplanet, mit of Mauna Kea has always had within the indigenous
HD145934b,anditshoststar’sresidualsshowcurvature Hawaiiancommunity. Wearemostfortunatetohavethe
whose velocity semiamplitude is indicative of a probable opportunitytoconductobservationsfromthismountain.
stellar or brown dwarf companion. We thank the many astronomers that contributed