2. Comments
• Tried
to
be
complementary
to
other
lectures
– Please
ask
ques<ons
(some<mes
controversial
on
purpose!)
• Only
talks
about
cosmological
surveys
– Galaxy
evolu<on,
planetary
searches,
galaxy
archeology
are
not
covered
• Personal
views
– Many
surveys
will
be
missed
(sorry)
– Op<cal
astronomy
(radio/CMB
surveys
are
poorly
represented)
3. Advice
• Wise
• Old
• ForgoKen
• Sartre
"You're
free,
choose,
that
is,
invent.”
• Nichol
“I
told
you
so!
I’m
always
right”
11. Fermilab
Industrial
Astronomy
(driven
by
technology)
Random
galaxy
Apache
Point
UWashington
2009
Nobel
Prize
for
CCDs
and
op<cal
fibers
12. LSS
Observational Definition
pcs
rs
-‐
420
M
SDSS
i llion
ly
1.37
b
CfA2
2dfGRS
Structures
larger
than
clusters,
typically
>
10Mpc
(larger
than
a
galaxy
could
have
moved
in
a
Hubble
<me)
14. Measuring
ξ(r)
or
P(k)
Simple
es<mator:
Data
ξ(r)
=
DD(r)/RR(r)
-‐
1
Advanced
es<mator:
ξ(r)
=
(DD-‐RR)2/RR-‐1
r
The
la,er
does
a
be,er
job
with
edge
effects,
which
cause
a
bias
to
the
Random
mean
density
of
points
Usually
10x
as
many
random
points
over
SAME
area
/
volume
Same
techniques
for
P(k)
-‐
take
Fourier
transform
of
density
field
rela<ve
to
a
random
catalog
over
same
volume.
Several
techniques
for
this
-‐
see
Tegmark
et
al.
and
Pope
et
al.
Also
“weighted”
and
mark
correla<ons
15. Measuring
ξ(r)
II
Essential the random catalog looks like the real data!
16. Errors
on
ξ(r)
Hardest
part
of
esLmaLng
these
staLsLcs
On
small
scales,
the
errors
are
Poisson
On
large
scales,
errors
correlated
and
typically
larger
than
Poisson
• Use
mocks
catalogs
• PROS:
True
measure
of
cosmic
variance
• CONS:
Hard
to
include
all
observa<onal
effects
and
model
clustering
• Use
jack-‐knifes
(JK)
• PROS:
Uses
the
data
directly
• CONS:
Noisy
and
unstable
matrices
17. Jack-‐knife
Errors
Real
Data
•
Split
data
into
N
equal
subregions
1
3
•
Remove
each
subregion
in
turn
and
compute
ξ(r
)
4
5
6
•
Measure
variance
between
regions
as
func<on
of
scale
N=6
2
3
N
(N −1)
σ = 2
N i=1
∑ (ξi − ξ ) 2
4
5
6
Note
the
(N-‐1)
factor
because
there
are
N-‐1
es<mates
of
mean
18. SDSS
&
WMAP
• Now
the
most
successful
astronomical
facili<es
in
the
world
• 4187
papers
with
162913
cita<ons
(Jan
17th
at
4pm)
• At
least
a
paper
a
day!
21.
The
DETF
figure-‐of-‐merit
is
the
reciprocal
of
the
area
of
the
error
ellipse
enclosing
the
95%
confidence
limit
in
the
w0–wa
plane.
Stage
II
–
today
(ish)
Stage
III
–
factor
of
3
Stage
IV
–
factor
of
10
22. Dark
Energy
is
bad
for
Astronomy
(ArXiv:0704.2291)
1. Cultural differences: HEPs are
fundamentalists (“specialists”) and
astronomers are generalists. Respect each
others cultures
2. Don’t over optimize your surveys - plan for
the unexpected
3. Don’t over prioritize DE surveys to the
expense of others
4. Be inclusive and publish your data
5. Nurture young talent and give recognition
where due
The
SDSS
is
the
last
of
its
kind!
23. Don’t over-optimize
• Dark energy is now systematics limited - young
scientists should do PhD’s in dust and biasing.
• DETF proposed diversity in experiments
• All the new surveys are building this in, e.g., DES will
get less SNe but (hopefully) understand them better
• These will greatly benefit astrophysics and I would
argue would not be done without the driving force of
DE (unfocused science is also a risk and can be
expensive)
• DE experiments will deliver more numbers and area.
Excellent for cosmic variance, environment studies
and high-dimensional parameter searches.
• Larger field of views are driven by technology, so we
would do large area surveys anyway.
24. BOSS
in
a
nutshell
8,000 deg2 footprint in Spring (Eisenstein
et
al.
2011)
3,000 deg2 footprint in Fall
• Upgraded spectrographs (with better throughput)
• 1000x 2-arcsec fibers in cartridges
• Increase wavelength range to 3600-10,000A
(R=1500-2600)
• Finished ~3,000 deg2 southern imaging in Fall 2008.
• Released as part of DR8, published in ApJS (2011).
• Currently doing only spectroscopy
• 1.5 million galaxies, i<19.9, z<0.8, over 10,000 deg2
• 150,000 QSOs, g<22, 2.3<z<3, over 8,000 deg2
25. Data
so
far
et
al.
2011,
Ho
et
al.
2012,
Seo
et
al.
2012
Ross
Current
status
Done
by
2014
27. BOSS
• BOSS
is
designed
as
a
“stage
III”
project
to
constrain
DE
using
the
baryon
acous<c
oscilla<on
(BAO)
method
– Galaxies
z~0.1-‐0.7
1%
dA,
2%
H(z),
z~0.35
&
0.6
– QSOs
(LyAF)
z~2-‐3
1.5%
dA,H
at
z~2.5
28. AS3:
e-‐BOSS
• gri selection conducted on a single plate based on DR8 photometry
(targeting the CFHT-LS W3 field)
• 78% redshift success efficiency - ~68% in 0.6<z<1
DES
overlap
BOSS
e-BOSS
MaNGA
Start
2014?
J.P. Kneib
28
29. The Dark Energy Survey
Blanco
4-‐meter
at
CTIO
• Survey project using 4
complementary techniques:
I. Cluster Counts
II. Weak Lensing
III. Large-scale Structure
IV. Supernovae
• Two multiband surveys:
5000 deg2 grizY to 24th mag
30 deg2 repeat (SNe)
• Build new 3 deg2 FOV camera
and Data management system
Survey 2012-2017 (525 nights)
Facility instrument for Blanco
29
30. DECam
C4
in
its
cell
(UCL)
New
flat-‐field
Screen
(CTIO)
Completed
Imager
(FNAL)
31. DES Science Summary
Forecast
Constraints
on
DE
Equa<on
of
State
Four Probes of Dark Energy
• Galaxy Clusters DES
• ~100,000 clusters to z>1
• Synergy with SPT, VHS
• Sensitive to growth of structure and geometry
• Weak Lensing
• Shape measurements of 300 million galaxies
• Sensitive to growth of structure and geometry
• Large-scale Structure
• 300 million galaxies to z = 1 and beyond
Planck
prior
assumed
• Sensitive to geometry
• Supernovae
• 30 sq deg time-domain survey
• ~4000 well-sampled SNe Ia to z ~1
Factor
3-‐5
improvement
over
• Sensitive to geometry
Stage
II
DETF
Figure
of
Merit
31
37. The
Euclid
machine
Space-based Vis and NIR observations of galaxies
VIS
Imaging
NIR
Spectroscopy
NIR
Photometry
NIR
Imaging
Tomographic
shear
Redshib
machine
machine
Dark
MaRer
and
Galaxy
PowerSpectra-‐meters
Astronomical
data
base
for
Explorer
of
gravity
and
expansion
Legacy
science
38.
39. Area
requirements
• FoM
increases
with
increasing
area/volume
and
galaxy
number
density.
• This
ignores
that
any
survey
is
limited
by
cost:
<me
is
finite
• weak-‐lensing,
intrinsic
alignments
become
increasingly
important
for
shallower
surveys
• This
changes
the
trade-‐off
between
area
and
depth
• 6-‐year
dura<on,
WL+GC
gives
op<mal
survey
area
of
15,000
deg2
40. Euclid
clustering
measurements
20%
of
the
Euclid
data,
assuming
the
slitless
baseline
at
z~1
Distance-‐redshib
rela<on
moves
P(k)
42. Measuring
Modified
Gravity
• The
growth
factor
[or
its
deriva<ve,
the
growth
rate
f(z)]
quan<fies
the
efficiency
with
which
cosmological
structure
is
built.
• The
growth
rate
well
described
by
f(z)=Ωm(z)γ.
• A
detec<on
of
γ≠0.55
would
indicate
a
devia<on
from
General
Rela<vity,
and
thus
a
completely
different
origin
of
cosmic
accelera<on,
rather
than
dark
energy.
• Euclid
can
constrain
this
parameter
to
0.01
(where
ΛCDM
corresponds
to
γ=0.55).
• the
γ-‐parameterisa<on
is
merely
an
example.
In
general,
Euclid
will
provide
<ght
constraints
on
the
cosmological
growth
rate.
48. Most
people
look
at
about
20
galaxies.
All
galaxies
looked
at
by
at
least
20
people
(median
38).
49. Karen
Masters:
The
Enigma
of
Red
Spirals.
Wednesday
9th
December
2009
49
50. Summary
• Era
of
surveys
is
here
– More
to
come
(DR9,
DES,
Euclid).
– By
end
of
decade,
billions
of
galaxies
in
public
domain
– Only
held
back
by
your
imagina<on!
– Wonderful
technologies
to
share
and
collaborate
with
such
data
• Era
of
maximal
ignorance
– We
know
“nothing”,
but
not
what
caused
it
or
what
it
could
be
– Progress
will
only
be
made
through
observa<on!
– Don’t
let
anyone
tell
you
it’s
a
“crazy
idea”