the Potential of the
Milky Way
adrian price-whelan
kathryn johnston
TIDAL STREAMS
&
david hogg
Slide 2
Slide 2 text
Tidal Streams
Distances 10–100 kpc
Slide 3
Slide 3 text
Tidal Streams
Distances 10–100 kpc
∼20 known around Milky Way
(100 total?)
Slide 4
Slide 4 text
Tidal Streams
Distances 10–100 kpc
∼20 known around Milky Way
(100 total?)
Sgr, Orphan, GD1, Pal 5 most prominent
(Many other short streams, many w/ no progenitors)
Slide 5
Slide 5 text
Tidal Streams
Stream morphology set by progenitor orbit
Dynamically cold
(small spreads in orbital properties)
(and internal dynamics…)
Slide 6
Slide 6 text
Tidal Streams
Stream morphology set by progenitor orbit
Dynamically cold
(small spreads in orbital properties)
(and internal dynamics…)
Can measure host galaxy potential
Slide 7
Slide 7 text
With near-future surveys we
can measure 6D kinematics
(l, b, D, µl, µb, v
los
)
Slide 8
Slide 8 text
RR Lyrae
as dynamical tracers
Slide 9
Slide 9 text
RR Lyrae
as dynamical tracers
Bright, MV ~ 0.5
Slide 10
Slide 10 text
RR Lyrae
as dynamical tracers
Found in many streams
(some may only have a few…)
Bright, MV ~ 0.5
Slide 11
Slide 11 text
RR Lyrae
as dynamical tracers
Found in many streams
(some may only have a few…)
Bright, MV ~ 0.5
Easy to identify from light curve
Slide 12
Slide 12 text
RR Lyrae
as dynamical tracers
Found in many streams
(some may only have a few…)
Bright, MV ~ 0.5
Easy to identify from light curve
Period-Luminosity relation for precise distances
(2% uncertainty in relative distances)
Slide 13
Slide 13 text
“Orphan stream”
(no uncertainty)
Slide 14
Slide 14 text
M giants
(15-20%)
Slide 15
Slide 15 text
BHB stars
(10%)
Slide 16
Slide 16 text
RR Lyrae
with PL (2%)
Slide 17
Slide 17 text
RR Lyrae in Gaia
post-launch estimates
Transverse
velocity error
Slide 18
Slide 18 text
post-launch estimates
15 km/s at 30 kpc
RR Lyrae in Gaia
Slide 19
Slide 19 text
“Orphan stream”
(no uncertainty)
Slide 20
Slide 20 text
HSTPROMO
(~0.1 mas/yr)
Slide 21
Slide 21 text
Gaia
(~0.03 mas/yr)
Slide 22
Slide 22 text
How do we use these data?
Slide 23
Slide 23 text
How do we use these data?
We need fast generative models
Slide 24
Slide 24 text
How do we use these data?
We need fast generative models
N-body
Slide 25
Slide 25 text
Rewinder
Slide 26
Slide 26 text
potential center
L1
L2
Stream formation
Slide 27
Slide 27 text
potential center
L1
L2
Stream formation
Slide 28
Slide 28 text
t = 0
Price-Whelan et al. (2014)
Rewinder
L1
L2
stream stars
progenitor
Slide 29
Slide 29 text
t = 0
Price-Whelan et al. (2014)
Rewinder
L1
L2
Slide 30
Slide 30 text
Rewinder
t = -1
Price-Whelan et al. (2014)
Slide 31
Slide 31 text
Rewinder
t = -1
evaluate
likelihood
Price-Whelan et al. (2014)
Slide 32
Slide 32 text
Rewinder
t = -2
Price-Whelan et al. (2014)
Slide 33
Slide 33 text
Rewinder
t = -2
evaluate
likelihood
Price-Whelan et al. (2014)
Slide 34
Slide 34 text
Rewinder
t = -3
Price-Whelan et al. (2014)
Slide 35
Slide 35 text
⌧ub
K
unbinding time
leading/trailing tail
mass vs. time
any parametrization
per star
progenitor
potential
(l, b, d, µl, µb, vr)
(l, b, d, µl, µb, vr)
Price-Whelan et al. (2014)
Rewinder
M(t)
Slide 36
Slide 36 text
⌧ub
K
unbinding time
leading/trailing tail
mass vs. time
any parametrization
per star
progenitor
potential
(l, b, d, µl, µb, vr)
(l, b, d, µl, µb, vr)
marginalize out
Price-Whelan et al. (2014)
Rewinder
M(t)
Slide 37
Slide 37 text
⌧ub
K
unbinding time
leading/trailing tail
mass vs. time
any parametrization
per star
progenitor
potential
(l, b, d, µl, µb, vr)
(l, b, d, µl, µb, vr)
marginalize out
FFFUUUU
Price-Whelan et al. (2014)
Rewinder
M(t)
Slide 38
Slide 38 text
−100 −50 0 50
X [kpc]
−100 −50 0
X [kpc]
x
“DATA”
8 “RR Lyrae” stars
Gaia velocity errors
2% distance errors
+ progenitor
Price-Whelan et al. (2014)
We must move away from
static, analytic potentials!
With restrictive models, we will make
uninterpretable, biased measurements
Slide 42
Slide 42 text
For example,
Basis Function Expansions
⇢nlm = ˜
⇢nl(r) Ylm(✓, )
nlm = ˜
nl(r) Ylm(✓, )
Slide 43
Slide 43 text
Current stream models can recover
true parameters of true potential
(from “observed” simulations)
Slide 44
Slide 44 text
Current stream models can recover
true parameters of true potential
(from “observed” simulations)
Moving to more general potential models &
real obs. uncertainties is going to hurt
(but we have to do it)
Slide 45
Slide 45 text
Current stream models can recover
true parameters of true potential
(from “observed” simulations)
Rewinder can do this, but
more work to be done!
Moving to more general potential models &
real obs. uncertainties is going to hurt
(but we have to do it)