Combining Imaging + RV’s
Jus n R. Crepp
Frank M. Freimann Professor Department of Physics University of Notre Dame [email protected] Connec ng Two Dis nct Techniques Doppler Trends
30 HD 19467 20
10
0
ï10 dv/dt=é1.37+/é0.09 m/s/yr
Radial Velocity [m/s] ï20 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 Mo va on • High detec on efficiency. • Break sin(i) degeneracy, construct 3d-orbits. • Dynamical masses of companions. • Calibrate evolu onary models.
Kappa Andromedae Determine fpl for wide orbits.
“b”
HR 8799 Marois et al. 2010 Carson et al. 2012 Par al Orbits
Isaac Newton What informa on can be gleaned? (1) Minimum dynamical mass (Torres 1999). (2) Physical separa on (Howard et al. 2010). Minimum Mass
2 ⎛ M ⎞ ⎛ d ρ ⎞ d(RV ) min⎜ B ⎟ =1.39E − 5 ⎜ ⎟ ⎝ MSun ⎠ ⎝ pc arcsec⎠ dt where RV accelera on is measured in m s-1 yr-1. € Physical Separa on
d(RV ) GMB ⎛ rAB ⎞ ρ = 2 cosθ, ⎜ ⎟ sinθ = dt rAB ⎝ AU⎠ π
where rAB is the instantaneous true physical separa on.
€ € 3d-Orbits and Dynamical Mass
Radial Velocities:
K, P, e, ω, tp Astrometry: * a, i, P, e, ω, tp, Ω
How much informa on do we need to calculate the orbit and mass? Murray & Dermo 1998 At What Point Can We Calculate the Orbit?
4π 2a3 p2 = G (M + m) * No ce that precise parallax is essen al (Dupuy & Kraus 2013) € When do constraints become interes ng? s “with accuracies of a few mas, it is possible to place strong constraints on KBO orbits even from very short (e.g., 24 hour) arcs … reflex mo on is easily σast detected giving a measure of distance to “Yes, of course the the target in less than an hour.” “I don’t trust anyone’s results inclina on falls out - Bernstein & Khushalani 2000 un l I see 2 phase wraps.” immediately.” - Andy Boden (Caltech) - Jessica Lu (IfA) Galac c Center Comets
PMS Binaries
100 10-1 10-2 10-3 10-4 10-5 10-6 10-7 We are here. σast / s
How Do We Fit Observa ons?
• Bayesian sta s cal framework. (“Data Analysis: A Bayesian Tutorial” by Sivia) • Assume we understand uncertain es. • Keplerian orbit model. Single companion. • Markov-Chain Monte Carlo (MCMC) analysis. (“Quan fying the Uncertainty in the Orbits of Extrasolar Planets”, Ford 2005, AJ, 129, 1706) Markov Chain Monte Carlo
• Many degrees of freedom {P, e, i, ω, tp, Ω, …}. • Metropolis-Has ngs algorithm.
P
e X (Working in N-dimensional space) i
Metropolis-Hastings Pseudo-Code:
(1) Cull P, a, e, i, ω, tp, Ω, + nuissance parameters. (2) Calculate Likelihood, L, for parameters. Use both RV, Imaging data simultaneously.
(3) if (L > L0) record new chain entries. else