Khee-Gan Lee

Mapping the z > 2 Cosmic Web with
3D Lyα Forest Tomography
“From Wall To Web” Meeting, Berlin, Germany
Khee-Gan (K.G.) Lee
Hubble Fellow, Lawrence Berkeley National Lab
July 27, 2016
K.G. Lee
Lyα Forest Tomography
The Cosmic Web and Cosmology
z = 10
z=2
z=1
z = 0.5
z=0
Credit: Anatoly Klypin (NMSU) & Andrei Kravtsov (Chicago)
I Pattern of voids, filaments and nodes in the large-scale distribution of
DM + baryons
I Caused by gravitational evolution of Gaussian random-phase initial
conditions from inflation
I Detection of cosmic web in the 1980s was key evidence supporting
inflationary cold dark matter paradigm
I Galaxy formation and evolution is influenced by cosmic web environment
I The evolution over time probes gravity models and the cosmological
constant
K.G. Lee
Lyα Forest Tomography
Why study cosmic web at ‘Cosmic Noon’ ?
Peak of cosmic star-formation + AGN activity occured at z ∼ 2 − 3
Bouwens+2011
K.G. Lee
Lyα Forest Tomography
Lyman-α Forest as Probe of z > 2 Large-Scale Structure
I
Seen in quasar spectra in their restframe
λ < 121.6nm wavelengths
I
Caused by neutral hydrogen in the IGM
I
Absorption is non-linear tracer of underlying
LSS density in mildly overdense regime
(ρ/hρi ∼ few):
T0 −0.7
τ(x) ∝
Γ
I
ρ(x)
hρi
Ellison+2000
2−0.7(γ−1)
In this talk, I will ignore astrophysics!
Absorption ↔ Cosmic Web Density
Credit: AmSci/R. Simcoe
K.G. Lee
Lyα Forest Tomography
Lyα Forest Tomography at 2.15 < z < 2.55
(Wiener-filtered)
Visualization Credit: Thomas Mueller (Haus der Astronomie)
Transverse footprint: 24h−1 Mpc × 18h−1 Mpc, LOS Pathlength: 340 h−1 Mpc
K.G. Lee
Lyα Forest Tomography
Come check out the VR visualization!
https://youtu.be/xV2Ng8n61Xc
K.G. Lee
Lyα Forest Tomography
CLAMATO Survey
(COSMOS Lyman-Alpha Mapping And Tomography Observations)
I Co-PIs: Schlegel & White
I Ongoing program targeting ∼ 1 sq deg of COSMOS field
I Require ∼ 240hrs on-sky with Keck-LRIS → ∼ 30 nights over 3 years
I Target ∼ 1000 LBGs at 2.3 . z . 3 for R ∼ 1000 spectroscopy
→ hzi ∼ 2.3 LSS map over 106 h−3 Mpc3 ∼ (100 h−1 Mpc)3
I Similar spatial resolution (∼ 3 h−1 Mpc) and volume (∼ 106 h−3 Mpc3 ) to
GAMA survey at z < 0.3!
K.G. Lee
Lyα Forest Tomography
Latest Observations
I 10.5 nights through UC and NAOJ,
∼ 36hrs on-sky (inc time swaps)
I 14 masks within roughly 15 0 × 21 0
I Targeted 2.3 < z < 3.0 LBGs and
QSOs, spec-z’s + Ilbert 2009 photo-z’s
I 2-4hr LRIS integrations with B600/4000
and R400/7500
I 274 spectra extracted, 151 good
redshifts (63 new + 88 retargeted)
I 124 spectra with 2.15 < z < 2.55 Lyα
absorption
K.G. Lee
Lyα Forest Tomography
Tomographic Reconstruction
Measure Lyα forest transmission δF = F/hFi − 1 (’data’), pixel noise
estimates σF , and [x, y, z] positions. Perform Wiener filtering on these
inputs to estimate the map:
M = CMD · (CDD + N)−1 · D
The noise term provides some noise-weighting to the data. We assume
Gaussian correlation function in the map, where
CDD = CMD = C(r1 , r2 ), and
"
#
(∆rk )2
(∆r⊥ )2
2
C(r1 , r2 ) = σF exp −
exp −
,
(1)
2L2k
2L2⊥
with L⊥ = 2.5h−1 Mpc and Lk = 2.0 h−1 Mpc, and σF = 0.8 (Note
average sightline separation hd⊥ i ≈ 2.5 h−1 Mpc).
K.G. Lee
Lyα Forest Tomography
A Galaxy Protocluster at z = 2.44 (Lee et al 2016)
I See one large (∼ 20 h−1 Mpc) overdensity
in our absorption map (3σ significance)
I Correlated with z = 2.45 galaxy
protocluster from LBGs and LAEs
(Diener+2015, Chiang+2015)
I Comparison to sims gives descendant
mass estimates:
M(z = 0) = (3 ± 1.5) × 1014 h−1 Mpc
(∼ Virgo cluster)
I Elongated morphology suggests possible
fragmentation into two z ∼ 0 clusters
HETDEX Pilot LAEs (stars, Chiang+2015); LBGs
(squares, Diener+2015); Open circles: sightline positions
K.G. Lee
Lyα Forest Tomography
A Forming Supercluster at z ∼ 2.48?
Multiple known
overdensities in our map,
within ∼ 100Mpc
comoving
I LAE/LBG overdensity at
z = 2.44
I DSFG/LBG overdensity
at z = 2.48
I Sub-mm overdensity with
X-ray extended source at
z = 2.51
Also a candidate
overdensity at z = 2.20
K.G. Lee
Lyα Forest Tomography
Voids and Protoclusters
M = 1.1 × 1015 h−1 M cluster progenitor
R = 11.7 h−1 Mpc cosmic void
Casey Stark (UC Berkeley) studied detectability of z ∼ 2.5 protoclusters and
voids with Lyα forest tomography in sims (Stark+2015a, 2015b)
I L = 256 h−1 Mpc TreePM sim with IGM absorption from FGPA
I Generated DM density field and mock tomographic maps with sightline
sampling + noise consistent with real data
I Protoclusters: Look for 3σ peaks in smoothed map, which gives > 90%
completeness and ∼ 75% purity for M > 3 × 1014 h−1 M progenitors
I Voids: Search for spherical low-density regions in DM field and
tomographic map → ∼ 65% volume overlap for ∼ 15% filling factor
K.G. Lee
Lyα Forest Tomography
Voids and Protoclusters in CLAMATO
Right: Central part of COSMOS Field
65 h−1 Mpc
6
Magenta CLAMATO 0.8 sq deg
65 h−1 Mpc
Blue Pilot field (2014-2015)
Dots Photo-z and spectro-z
z = 2.4 − 3.0 LBG targets
Below: Simulated protoclusters and voids
(approx to scale)
?
K.G. Lee
Lyα Forest Tomography
Cross-Correlation between Lyα Forest and Galaxies
I Cross-correlate 2.0 < z < 2.6 CLAMATO spectra with 662 spec-z’s
available in COSMOS field, → 20 σ detection significance
I Will assume forest bF (1 + βf ) is known (2% constraints from
Blomqvist+2015) to estimate galaxy bias + RSD, redshift errors
Preliminary!
K.G. Lee
Lyα Forest Tomography
Improving Photometric Redshifts with Lyα Forest
I Schmittfull & White, arXiv:1606.03399
I Spec-z’s extremely challenging for quiescent z > 1.5 galaxies, and
photo-z’s are limited to ∆(z) ∼ 0.1 precisions
→ ∆χ ∼ 100 h−1 Mpc LOS uncertainty
I Massive galaxies live in high-density environments, so can constrain LOS
positions to ∼ 3× better precision (∆(z) ∼ 0.03)
K.G. Lee
Lyα Forest Tomography
Classifying the Cosmic Web at z ∼ 2 (arXiv:1603.04441)
After Zel’dovich 1970 (... Hahn+2007, Forero-Romero+2009, Eardley+2015).
Analyze deformation tensor of density field:
Tij =
∂2 Φ
∂xi ∂xj
Easy to do in Fourier space, since ∇2 Φ̃ = k2 Φ̃ = 4πGδk :
T̃ij =
ki kj δk
k2
(2)
Compute 3 eigenvalues at each point, and count # above eigenvalue threshold
λth :
I No eigenvalue above threshold: void
I 1 eigenvalue above threshold: sheet
I 2 eigenvalues above threshold: filament
I 3 eigenvalues above threshold: node/knot
For Ly-a forest flux field δf = F/hFi − 1, just flip the signs!
K.G. Lee
Lyα Forest Tomography
Classifying the z = 2.5 Cosmic Web with CLAMATO
Analyzed L = 256 h−1 Mpc FGPA sim to find voids (white), sheets, filament,
nodes
CLAMATO-like: hd⊥ i = 2.5 h−1 Mpc
70% of volume is correctly classified w.r.t. underlying DM (c.f 75% in
GAMA at z ∼ 0.2!)
K.G. Lee
Lyα Forest Tomography
Constraining Galaxy Spin-Filament Alignments at z ∼ 2
Project with Alex Krolewski (Berkeley grad student) and Zarija Lukic
K.G. Lee
Lyα Forest Tomography
Preliminary example: Angle between e1 eigenvector of tidal tensor
vs projection of 104 galaxies within CLAMATO-like reconstructions
within L = 100 h−1 Mpc Nyx volume
I
Will probably require future surveys: > 103 galaxies combined
with hd⊥ i ∼ 1h−1 Mpc maps (c.f. hd⊥ i ∼ 2.5 h−1 Mpc in
CLAMATO)
K.G. Lee
Lyα Forest Tomography
Inferring Primordial Density Fields
Project with Grigor Aslanyan (LBL) and Martin White
I Solve for the quasi-linear density field using L-BFGS minimization, with
log-normal assumption
I Preliminary tests using log-normal flux field, 163 grid with 4Mpc/h bins
I Will try to infer primordial initial conditions based on z = 2.5 tomographic
maps, use to run N-body sim!
K.G. Lee
Lyα Forest Tomography
Conclusion
I
Observations of z ∼ 2 − 3 QSOs + LBGs at high area densities allow
3D reconstructions of foreground Lyα forest absorption
I
Ideal for detecting extended z ∼ 2 structures: voids + protoclusters
I
Good recovery of z ∼ 2 cosmic web sheets and filaments —
comparable to z ∼ 0.1 galaxy surveys!
I
IGM Tomography already works! CLAMATO Survey on
Keck-I/LRIS (2016-2019) to map ∼ 106 h−3 Mpc3 probing scales of
∼ 3 h−1 Mpc
I
A lot of theory work needed to interpret and exploit observations!
I
Next generation spectroscopic facilities (Subaru PFS, Maunakea
Spectroscopic Explorer, Keck-FOBOS) will exploit this technique.
K.G. Lee
Lyα Forest Tomography
Example Spectra
First systematic use of LBGs for Lyα forest analysis!
K.G. Lee
Lyα Forest Tomography
Cross-Correlation Galaxy Samples
K.G. Lee
Lyα Forest Tomography