Post on 03-Feb-2022
Andr eas PawlikLeiden Univer sit y
PhD super visor : J oop Schaye
Claudio Dalla VecchiaHuub Röttgering
Andreas Pawlik, Leiden University
• I nt r oduct ion
• Ther mal f eedback (posit ive / negat ive)
• TRAPHI C - r adiat ive t r ansf er f or SPH
Andreas Pawlik, Leiden University
r edshif t z 6 0
1 13.7 Gyr
Fir st st ar s
Recombinat ion
Reionizat ion
Andreas Pawlik, Leiden University
• Gadget -2(Springel ‘05)
• Def ault r uns: L = 6.25 Mpc/ h, N = 2 x 2563
• St ar f or mat ion (Schaye & Dalla Vecchia ‘07)
• UV backgr ound (z < 9) (Haardt & Madau ’01; cooling tables: Wiersma et al. ’08)
Andreas Pawlik, Leiden University10log
-1.0 0.0 1.0 2.0
z = 9 z = 9r ef er ence r eheat ing f or z < 9
3.125 Mpc / h 3.125 Mpc / h
Andreas Pawlik, Leiden University10log
-1.0 0.0 1.0 2.0
3.125 Mpc / h
z = 6
3.125 Mpc / h
z = 6r ef er ence r eheat ing f or z < 9
Andreas Pawlik, Leiden University
• Smoot hing of f luct uat ions -> lower ing of r ecombinat ion r at e-> r educes r equir ed st ar f or mat ion r at e-> posit ive f eedback
• Phot o-evapor at ion of low-mass halos-> r educes st ar f or mat ion r at e->negat ive f eedback
AP, J. Schaye & E. van Scherpenzeel (arXiv:0807.3963)
Andreas Pawlik, Leiden University
• Dalla Vecchia & Schaye (2008) (Springel & Hernquist ‘03)
• v = 600 km s-1
• Mass loading = 2
I solat ed galaxy, 1012 Msun/ hDalla Vecchia & Schaye (2008)
Andreas Pawlik, Leiden University
• Bot h r eheat ing and super nova f eedback r educe t he SFR -> negat ive f eedback
• Reheat ing and super nova f eedback mut ually st r engt hen each ot her
• Ef f ect incr eases wit h r esolut ion
AP & J. Schaye (in preparation)
Andreas Pawlik, Leiden University
• Lar ge r epr esent at ive volumes(cosmic var iance, long wavelengt hs)
• High r esolut ion(f ir st galaxies, at omic cooler s ~108 Msun)
• Accur at e gas dist r ibut ion (r ecombinat ion r at e)
• Many sour ces(st ellar and r ecombinat ion r adiat ion)
Andreas Pawlik, Leiden University
• Radiat ive t r ansf er on hydr odynamics (vs. N-body/ semi-analyt ics)
• Spat ially adapt ive(vs. unif or m mesh)
• Par allel on dist r ibut ed memor y(vs. ser ial; par allel on shar ed memor y)
• Avoid scaling wit h # sour ces (vs. linear scaling)
TRAPHI C – r adiat ive t r ansf er f or SPH
Andreas Pawlik, Leiden University
•Adapt ive: Dir ect ly on SPH par t icles
•Par allel: Tr anspor t employs t he SPH par t icle-neighbor scheme
•Ef f icient : Comput at ion t ime independentof # sour ces
Andreas Pawlik, Leiden University
• Spat ial r esolut ion: # SPH neighbor s• Angular r esolut ion: # Cones• Tempor al r esolut ion: Clocks• Comput at ion t ime: Phot on packet mer ging
AP & J. Schaye (2008), MNRAS 389, 651
Andreas Pawlik, Leiden University
• Reionizat ion simulat ions r equir e t ailor ed appr oaches f or solving t he r adiat ivet r ansf er pr oblem
• TRAPHI C – r adiat ive t r ansf er f or SPHadapt ive, par allel, ef f icient
AP & J. Schaye (2008), MNRAS 389, 651
Andreas Pawlik, Leiden University
• Ther mal coupling r equir es mult i-f r equency t r eat ment
• Example: St ar wit h blackbody spect r um T = 105 K in homogeneous hydr ogen-only medium init ially neut r al and at 100 K
neut r al & cold
ionized & hot
Andreas Pawlik, Leiden University
Nor malized dist ance t o st ar
neut r al /ionizedf r act ion
Temp.
blackbodyTbb = 105 K
t = t rec