From Darkness, Light:From Darkness, Light:Computing Cosmological ReionizationComputing Cosmological Reionization

Romeel DavéWith: Kristian Finlator, Feryal Özel, Ben Oppenheimer

Movie by T. di Matteo

HI

James Webb Space Telescope“First Light Machine”

Planck: CMBat high precision

Atacama Large Millimeter Array

Square Kilometer Array

Reionization: A Multiwavelength Approach

When did reionization occur (start and end)?

What are the sources that reionized the Universe?

How did reionization transpire in space and time?

Fundamental Questions of Reionization

Observational Constraints: HI Optical Depth

Bolton & Haehnelt (2007)

Quasar spectra show sudden rise in HI at z~6.

SDSS

Observational Constraints: CMB Polarization

CMB T-E cross-correlation @ low-l shows enhanced signal from free electrons.

zreion~10±1

Observational constraints: z>6 galaxies

Bouwens et al. (2008)

z=4z=5z=6z=7

z=6.96: Iye et al (2008)

Hundreds of (putative) z>6 galaxies now seen.No unambiguous signature of reionization.

JWST

- Reionization began at z>~10 (0.5 Gyr)

- Reionization ended at z~6 (1 Gyr)

- There are plenty of galaxies (and very few quasars) seen at z>~6.

Do galaxies alone emit enough photons to reionize?

Observations tell us…

QI = Volume-averaged filling factor of ionized gasnph = # of ionizing photons per unit volumetrec = recombination time

Clumping factor CHII = <nHII ne>/<nHII><ne>

Can (in principle) measure dnph/dt.

But to solve reionization, need CHII, i.e. topology.

Analytic Reionization

z=9, 1 Mpc/h, dark matter

Outside-in: Voids reionize first, then dense regions

Inside-out: Regions around galaxies ionize first, then voids

Competition between:- Sources forming in overdense regions- Galaxies are highly clustered at early epochsvs.- High recombination rates in dense regions- Dense regions more self-shielded (shadowing)

Analytic results highly assumption-dependent. Simulate!

Topology of Reionization

Simulating Reionization: Physics

(1) structure formation with gas (density, temperature evolution)

(2) sources of photons (normal stars, Pop III stars, AGN, exotica)

(3) non-equilibrium thermal state

(4) non-equilibrium ionization state

(5) radiation transport

1 2

3

Code Comparison

Code method c=? shadows scaling comments

C2-ray ray-tracing ∞ yes N*Nx3 n-body

ART ray-tracing ∞ ? N*Nx3

FLASH-HC ray-tracing ∞ yes N*Nx3

TRAPHIC ray-tracing c yes NcN

SPHmass resolution limit?

CRASH Monte carlo ∞ yes N*N time dependence?

OTVET* moments << c ? Nx3 optically-thin fEdd

MARCH* moments c yes Nx~4

accurate fEdd; highly flexible; use SPH sim's

Zel'dovich approximation +irradiated boundary

* Have been implemented into a cosmological radiation-hydro code

The Moments of the Transfer Equation

- Derive fEdd from accurate long-characteristics → minimize artifacts; enhance shadowing

See also Auer & Mihalas (1970); Stone, Mihalas, & Norman (1992)

Eddington tensor

The Future: HI 21cm maps

Redshifted 21cm (~100 MHz) traces HI directly.dTbright~xHI(Tspin-TCMB)

In principle, map HI distributionand get clumping factor.

Kinetically coupledto gasTs~(1+z)2

Returnsto CMBTs~(1+z)Santos+10

Summary

Reionization is a frontier for both observations and theory- Planck, JWST, LOFAR, SKA, … all will play major roles- Simulations will require next-gen techniques + technology

Our new fast & accurate cosmological rad-hydro code…- Makes no physical approximations (outside of galaxies)- Moment-based; doesn’t scale with # of sources- Still limited by dynamic range

Interesting early results:- Galaxies with normal star formation can reionize- But… hard to match all current constraints with normal SF- Topology of reionization: inside-outside-middle- Clumping factor decreases rapidly with time; ~2-3 by z~6