Cosmic Flows
Mike Hudson
U. Waterloo / IAP
Our Local Group
has a peculiar
velocity of ~600
km/s with respect
to the Cosmic
Microwave
Background
Tully et al ‘08 have shown that the LG’s motion is
shared by the Local Sheet (<7 Mpc) - linear
What masses are the source of this motion?
On what scale is the Universe at rest?
Outline
Two approaches:
Direct:
Peculiar velocity measurements
Parametric:
Density field predictions, and comparison to peculiar
velocity measurements
Mapping the Peculiar
Velocity Field
Mapping the Peculiar Velocity Field
The peculiar velocity of a single object (e.g. LG) cannot
distinguish between a small nearby overdensity or a
massive distant overdensity.
By mapping the peculiar velocity field, can find the d
field:
• Infall into an attractor suggests local source.
• Coherent “bulk flow” with little shear suggests
distant sources.
Measuring peculiar velocities
cz = H0 r + vpec
• Measuring redshifts (cz) is easy.
• Measuring r is much more difficult, and
less accurate.
• Tully-Fisher (~20% uncertainty per gal)
• Fundamental Plane (~20% uncertainty)
• Type Ia SNe (~10% uncertainty)
Bulk flow
The bulk flow is the average peculiar
velocity over some volume.
Less noisy than individual measurements
It is sensitive to structures on scales
larger than the scale being averaged.
Some history…
• Rubin et al. (‘76) detected a large bulk motion
of Sc galaxies.
• They assmed that Sc galaxies were standard
candles. No dynamics (rotation curves or
linewidths).
Sayings of the Samurai
“A mean motion of ellipticals toward l = 312 +-11 deg, b = 6
+- 10 deg at 599 +- 104 km/s is observed …. It is inferred
that the Local Group motion with respect to the MWB is
primarily due to mass concentrations at V greater than
5000 km/s.”
Dressler et al. 1987
“A flow toward a great attractor centered on l=307, b=9 at a
distance of 4350 km/s … gives a much better fit to the
motions of the ellipticals than the bulk motion considered
earlier.”
Lynden-Bell et al. 1988
Tidal Field
in Local
Surveys
Velocity from
beyond
60 h-1 Mpc
• Mark III:
366+/-125 km/s
• SFI: 255 km/s
Hoffman, Eldar,
Zaroubi & Dekel 2001
Bulk Flows on Larger Scales
Measure bulk motion on scales larger than the
distance to the GA, (hope to) separate local
(GA) from distant sources.
However, recent (1999+) large-scale surveys
measured bulk flow statistics that are
apparently in conflict.
The L & P Smoking Gun?
LP
? No HST data
HST
snapshots
(Laine et al 03)
SMAC
discussion in MH et al ‘04
show
filamentary
dust in some
BCGs.
Bulk Flows in the CMB frame
Survey
Meth
od
V
km/s
Random
error
(km/s)
l
b
LG
CMB
627
22
276
30
LP (Lauer & Postman 95)
BCG
830
220
330
39
SC (Dale, Giovanelli et al 99)
TF
80
100
290
20
Willick (00)
TF
1100
450
270
27
SMAC (MH et al 99)
FP
650
180
260
-4
Shellflow (Courteau et al 00)
TF
70
90
~330
~30
EFAR (Colless, Wegner et al 01)
FP
650
350
50
10
SNIa (6000 < Hr < 15000 kms)
SNIa
790
210
298
3
MH Analysis of Tonry et al. (2003) SNIa compilation
Errors?
At this stage, many people attributed the
discrepancies to putative systematic errors.
BUT quoted errors are for bulk flows of the
sample, not of the volume and the large-scale
surveys are sparse.
Effects of Sparse Sampling
SC
SMAC
Blue=In
Red=Out
EFAR
GA
PP
Errors Including Sampling
cf. Watkins & Feldman
Survey
Meth
od
LP (Lauer & Postman 95)
BCG
SC (Dale, Giovanelli et al 99)
V
km/s
Random
error
(km/s)
Sampling
error
(km/s)
l
b
830
220
110
330
39
TF
80
100
170
290
20
Willick (00)
TF
1100
450
220
270
27
SMAC (MH et al 99)
FP
650
180
180
260
-4
EFAR (Colless, Wegner et al 01)
FP
650
350
210
50
10
SNIa (6000 < Hr < 15000 kms)
SNIa
790
210
130
298
3
Sampling errors are not negligible
With Feldman and Watkins, we have
devised a new weighting scheme that
minimizes small-scale aliasing
… See Hume’s talk for more details …
Once we correct for sparse sampling, all of
the peculiar velocity surveys agree with each
other
(except for Lauer & Postman’s BCG survey)
Comparison
SFI++ (Giovanelli et al)
“DEEP”
Tully-Fisher
in field
Various methods, clusters
SMAC, Willick, EFAR, SC, SNe
Bulk flow:
430 +- ~90 km/s
towards
l=284, b=12
Bulk flow:
387 +- ~100 km/s
towards
l=295, b=10
Combined: 407+- 81 km/s, towards l=287, b=8
Combined
cluster/Sne
sample
(excluding
LP)
Blue=In
Red=Out
GA
Seven
Samurai
Bulk Flow:
387 +/- 80 km/s
towards l=295,
b=10
All surveys are
consistent with
this value
Expectations for the Bulk Flow
Consistency with LCDM Models
What bulk flow do we expect for this
combined sample?
Allowing for the sparse sampling and
assuming a flat LCDM power spectrum with
WMAP5 parameters n=0.96, Wm h2 = 0.13
and s8 ~0.8 then the cosmic r.m.s. is ~ 110
km/s.
… but we measure 400 km/s.
This model is then rejected at the 99% CL.
Likelihood
contours in
Wm h2 vs s8
WMAP in colour;
bulk flows in black.
WMAP5
cosmology
rejected at 2.5 s
(98%).
Watkins, Feldman and MH 2009
Kinetic SZ effect
The kinetic SZ (kSZ) effect: the CMB
temperature decrement will depend on the
velocity of the cluster w.r.t to the CMB frame.
The signal depends on the amount of hot
plasma in the cluster and the cluster’s line-ofsight velocity.
Kashlinsky et al. 2008
Kashlinsky et al. averaged the kSZ effect
from 700 clusters within z < 0.3.
Their claim is that local volume out to
~900 Mpc/h was moving at a velocity
600-1000 km/s.
The direction of the flow that they find is
within 6 deg of our result.
Other probes
There are few independent ways to measure
the fluctuations in the mass density on
very large scales (~100 Mpc/h) in the
nearby Universe.
• Integrated Sachs-Wolfe effect (decay of
potential)
• SDSS galaxy power spectrum
Integrated Sachs-Wolfe effect
LCDM
predicts:
A=1
Observed:
A=2.23+-0.60
Galaxy Power Spectra
SDSS (more red gals)
and 2dFGRS (more blue)
disagree on small-scales.
Scales probed by flows
On large scales, they are
expected to agree (and
they do)
… but both exceed LCDM
predictions
Percival et al 07
Future
Conventional distances indicator methods
running out of targets
• NOAO FP Survey
• SNe (infinite but slow)
kSZ ?
NOAO Fundamental
Plane Survey
NOAO = National Optical Astronomical Observatories
Picture: Kitt Peak,Arizona
Conclusions I
• Once sampling effects are considered no conflict
between large-scale sparse peculiar velocity
surveys (except possibly Lauer & Postman).
• Combined sample bulk flow (~100 Mpc):
• Combined: 407+- 81 km/s, towards l=287, b=8
• No convergence to CMB yet …
• Not clear what is causing the flow.
• Marginally inconsistent with LCDM (98%)
• Some other measures also suggest large
fluctuations on large scales.
Predicting Peculiar
Velocities using the
Galaxy Density Field
Mass vs Light
SC
SC
GA
GA
PP
PP
Mass
(Far-infrared) Light
(Zaroubi inversion)
(Branchini)
“Biasing”
Averaged over large scales, one might assume
that
The simplest case is
d g  bd
where b is the linear biasing parameter.
but this need not be true: biasing might be
non-linear and depend on other parameters
• Use galaxy dg=b d
• Model external flows as bulk flow
Comparison with 2MASS
SNe
Surface
Brightness
Fluctuations
Pike &
MH 05
Tully-Fisher
For nearby peculiar velocity samples, 2MASS works well
Cosmological Parameters
Measure from
clustering:
s8,K =0.90 +- 0.10
s8,IRAS=0.85 +- 0.05
Measure from peculiar velocity surveys
Peculiar velocities yield
(W/0.30)^0.55 = 0.80+- 0.05
Compare with WMAP5 + BAO + SN : 0.78 +- 0.03
…. No problem on small scales
Residual Bulk Flow
• With =0.5, Pike & Hudson (2005)
found that
271+- 104 km/s
in the direction l=300 b=15
must arise from beyond 65 Mpc/h
Predicted bulk flows from IRAS PSCz
If =W0.6/b ~ 0.5,
additional sources
(not included in
PSCz) are required.
•Beyond 200/h Mpc?
•In the Zone of
Avoidance?
•Breakdown of linear
biasing - extra mass in
superclusters?
Bulk flow in spheres
Amplitude along l=300, b=10 “concordance direction”
MH et al ‘04
Cosmography
What is the source of the
large-scale ~300 km/s motion?
Must be beyond ~100 Mpc/h –
not the GA / Norma
The Usual Suspects
• The “Shapley Concentration” ?
• fit indicates that 100 +/- 60 km/s of the
LG’s motion is due to Shapley.
• (implied mass ~ 5x1016 Msun for Wm=0.3).
• Other structures such as the HorologiumReticulum supercluster complex and largescale voids may also play a role….
Predictions for =1
Growth of IRAS PSCz + Behind-the-Plane
(BTP) Gravity Dipole
H0
v(0)  
4
Not much at Shapley
Much stronger growth
than in PSCz alone

R
0
r 3
d g (r ' ) 3 d r 
| r |
Saunders et
al. 2000
conference
proceedings:
astro-ph/
0006005
… all that
has been
published so
far.
If ~0.5, then >=300 km/s
comes from beyond 100 Mpc
X-ray-selected Cluster Dipole
(Kocevski & Ebeling 2006)
VLG = Dcl
New all-sky X-ray selected
catalogue shows a large
jump (factor ~2 or ~300
km/s) at 150 h−1 Mpc.
Much of this is due to the
Shapley Concentration.
Cluster dipole is
NOT related to
IRAS dipole by
simple linear
biasing
Breakdown of Linear Biasing
In Shapley
IRAS PSCz: ~25 km/s at LG
Clusters: ~200-300 km/s at LG
Which is correct?
Shapley Overdensity
Preliminary analysis of 6dF DR2 plus “Biasing”
from mass-LK scaling relations.
Yields d ~ 1.75 on 30 Mpc scale. Compared
with LCDM s30 ~ 0.25, this is a 7s fluctuation.
(Q for theorists: extreme tail of 1pt fcn?)
Infall “only” ~50-75 km/s at LG, but exceeding
1000 km/s in foreground.
Density tracers suggest
existence of large-scale
structures, but do not agree
on what the important
structures are …
Future
• Deeper all-sky redshift surveys … 6dF + ?
• Better treatments of “biasing” (halo model)
• Better treatment of predicted peculiar
velocities (e.g. MAK)
Assume that
there is a
universal
monotonic
function
which relates
mass and
light
M/L U(M)
Clusters
Dwarfs
Groups
Galaxies
Marinoni & Hudson 02
Steepest part of this diagram is around groups
Conclusions II
• Studies of the density field are consistent with
the idea that much of the local velocity field
arises due to sources at large distance
• But what / where these sources are has not yet
been determined.