Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
AGN feeding and feedback:beyond sub-grid physics
Andrea Negri
In collaboration with Marta Volonteri
Hidden Monsters: Obscured AGN and Connections to Galaxy EvolutionFriday 12 August 2016
~last decade: hydro cosmological simulations
dark matter + stars + gas
Evolution of the large scale structure
Andrea Negri – AGN feeding and feedbackIllustris simulation
~last decade: hydro cosmological simulations
dark matter + stars + gas
Evolution of the large scale structure
Andrea Negri – AGN feeding and feedbackIllustris simulation
Huge dynamical range!currenly impossible to simulate
~last decade: hydro cosmological simulations
dark matter + stars + gas
Evolution of the large scale structure
Andrea Negri – AGN feeding and feedbackIllustris simulation
Huge dynamical range!currenly impossible to simulate
Simulations are forced to adopt lower resolution
~ 6 kpc for MassiveBlack at z=0 (Di Matteo+2012)
~ 1 kpc for Horizon-AGN (Dubois+2012)
> 750 pc Illustris (10 pc hydro, Vogelsberger+14)350 pc EAGLE simulation, z=0 (Schaye+14)
need to resort to subgrid models for BH accretion
Subgrid models for BH accretion
Andrea Negri – AGN feeding and feedback
Subgrid models for BH accretion
Various models are present in literature, most of them based on Bondi Bondi accretion
M BH=α4π G2 M BH
2 ρ
c s3
formal solution of: ● spherically symmetric accretion problem● adiabatic, no feedback, no rotation
Andrea Negri – AGN feeding and feedback
Subgrid models for BH accretion
Various models are present in literature, most of them based on Bondi Bondi accretion
M BH=α4π G2 M BH
2 ρ
c s3
In simulations we have:● non spherical accretion● multiphase gas● rotation
formal solution of: ● spherically symmetric accretion problem● adiabatic, no feedback, no rotation
Andrea Negri – AGN feeding and feedback
How accretion is implemented
and cs are calculated using information on the
surrounding gas, plenty of ways (every code/author)
M BH=α4π G2 M BH
2 ρ
c s3
Andrea Negri – AGN feeding and feedback
● Volume weighted (GADGET, Booth+09, Vogelsberger+13)● Mass weighted (Dubois+12)● ISM subgrid model (hot/cold phase, Pelupessy+07)●Global mass w. averages from all the cells (Choi+2012)●Direct accretion of hot and cold gas when hydro is more resolved than gravity (Steinborn+15)
Sometimes it is not even mentioned!
How accretion is implemented
and cs are calculated using information on the
surrounding gas, plenty of ways (every code/author)
M BH=α4π G2 M BH
2 ρ
c s3
Andrea Negri – AGN feeding and feedback
● Volume weighted (GADGET, Booth+09, Vogelsberger+13)● Mass weighted (Dubois+12)● ISM subgrid model (hot/cold phase, Pelupessy+07)●Global mass w. averages from all the cells (Choi+2012)●Direct accretion of hot and cold gas when hydro is more resolved than gravity (Steinborn+15)
Sometimes it is not even mentioned!
Roughly classified as● MassMass weighted
●Volume Volume weighted (SPH)
How accretion is implemented
(~100-300) is a boost factor, depends on resolution and sub-grid models of ISM
Andrea Negri – AGN feeding and feedback
M BH=α4π G2 M BH
2 ρ
c s3
● Constant (Springle+05, Dubois+12, Curtis+15)● Depends on ISM density (Booth+09, Steinborn+15)● Depends on feedback (Vogelsberger+13)●Sometimes not used (Pelupessy+07, Choi+12)●Sometimes is not even mentioned!
How accretion is implemented
(~100-300) is a boost factor, depends on resolution and sub-grid models of ISM
Andrea Negri – AGN feeding and feedback
M BH=α4π G2 M BH
2 ρ
c s3
● Constant (Springle+05, Dubois+12, Curtis+15)● Depends on ISM density (Booth+09, Steinborn+15)● Depends on feedback (Vogelsberger+13)●Sometimes not used (Pelupessy+07, Choi+12)●Sometimes is not even mentioned!
Philosophy behind Low resolution: ISM cold phase not resolved
Bondi radius not resolved●underestimated● c
s overestimated
BH accretion rate underestimatedunderestimated
Past attempts to calibrate the BH accretion by simulating the Bondi solution
has been performedwithout feedback
Andrea Negri – AGN feeding and feedback
see Wurster+13, Elahi+16
But they used Bondi accretion to simulate Bondi solution!
Past attempts to calibrate the BH accretion by simulating the Bondi solution
has been performedwithout feedback
Andrea Negri – AGN feeding and feedback
see Wurster+13, Elahi+16
But they used Bondi accretion to simulate Bondi solution!
What happens with feedbackfeedback?
A simple ideaA simple idea
Sims with same setup but with Bondi accretion:● Different schemes of weighting
●Different resolution
Fiducial simulations comparison with
Andrea Negri – AGN feeding and feedback
Simulations of an isolated galaxy:
● High resolution ~ 0.1 pc0.1 pc (cold and hot phase)● Well resolvedWell resolved Bondi radius for all the T
● NoNo parametrized accretion
A simple ideaA simple idea
Sims with same setup but with Bondi accretion:● Different schemes of weighting
●Different resolution
Fiducial simulations comparison with
Andrea Negri – AGN feeding and feedback
Simulations in a nutshell
Code: ZEUSMP (modified in Novak et al. 2011)
●2D axisymmetric
●MBH
= 3 × 107 M ☉
● Rbondi
= 0.1 pc at
T=108 K
●r from 0.1 pc to 250 kpc 0.1 pc to 250 kpc
●Mechanical Feedback Mechanical Feedback from broad absorption line (BAL) winds
● Radiative Radiative feedback
●Compton heating/cooling
●Radiative cooling
Andrea Negri – AGN feeding and feedback
Simulations in a nutshell
Code: ZEUSMP (modified in Novak et al. 2011)
●2D axisymmetric
●MBH
= 3 × 107 M ☉
● Rbondi
= 0.1 pc at
T=108 K
●r from 0.1 pc to 250 kpc 0.1 pc to 250 kpc
●Mechanical Feedback Mechanical Feedback from broad absorption line (BAL) winds
● Radiative Radiative feedback
●Compton heating/cooling
●Radiative cooling
Andrea Negri – AGN feeding and feedback
fiducial model:fiducial model:BH accretion is fluxflux on the innermost
radial grid
Andrea Negri – AGN feeding and feedback
cooling – accretion – feedback – hot bubble cycle
Andrea Negri – AGN feeding and feedback
cooling – accretion – feedback – hot bubble cycle
CHAOTIC ACCRETIONCHAOTIC ACCRETION
Andrea Negri – AGN feeding and feedback
cooling – accretion – feedback – hot bubble cycle
CHAOTIC ACCRETIONCHAOTIC ACCRETION
Feedback self-limits accretion to sub-Eddington values
Andrea Negri – AGN feeding and feedback
What happens with Bondi?
1)Define accretion radius racc
2) Calculate and cs inside r
acc
3) no boost factor
Computational grid
Andrea Negri – AGN feeding and feedback
What happens with Bondi?
1)Define accretion radius racc
2) Calculate and cs inside r
acc
3) no boost factor
Computational grid
● High vs low resolution runs
Andrea Negri – AGN feeding and feedback
What happens with Bondi?
1)Define accretion radius racc
2) Calculate and cs inside r
acc
3) no boost factor
Computational grid
● High vs low resolution runs● Mass vs Volume weighted
Andrea Negri – AGN feeding and feedback
Bondi low resolution mass weightedNO AGN
The central resolution is 3, 30 and 300 pc
The expected trend is recoveredin absenceabsence of AGN feedback
Andrea Negri – AGN feeding and feedback
Bondi low resolutionmass weighted
Explored racc
= 3, 30, 300 pc
Andrea Negri – AGN feeding and feedback
Trend: the larger rthe larger raccacc the larger the larger MMBHBH!!
Bondi low resolutionmass weighted
Explored racc
= 3, 30, 300 pc
Andrea Negri – AGN feeding and feedback
Trend: the larger rthe larger raccacc the larger the larger MMBHBH!!
Bondi low resolutionmass weighted
I was expecting the opposite! Why?
Explored racc
= 3, 30, 300 pc
Sims with large racc
there is a stronger stronger feedback feedback but a
larger gaslarger gas mass to heat/sweep away tostop the accretion
Andrea Negri – AGN feeding and feedback
Bondi high resolution
Sims with large racc
there is a stronger stronger feedback feedback but a
larger gaslarger gas mass to heat/sweep away tostop the accretion
Andrea Negri – AGN feeding and feedback
Bondi high resolution
Sims with large racc
there is a stronger stronger feedback feedback but a
larger gaslarger gas mass to heat/sweep away tostop the accretion
Andrea Negri – AGN feeding and feedback
Bondi high resolution
At small racc
the AGN feedback is more effective in stopping the accretion
Again, at low resolution the AGN feedback isless efficient
Compensates the fact that the accretion is dominated by hot modehot mode
Andrea Negri – AGN feeding and feedback
Bondi low resolutionvolume weighted
Again, at low resolution the AGN feedback isless efficient
Compensates the fact that the accretion is dominated by hot modehot mode
Andrea Negri – AGN feeding and feedback
Bondi low resolutionvolume weighted
How you calculate density and T mattersmatters!
Andrea Negri – AGN feeding and feedback
Mass weighted sims
Andrea Negri – AGN feeding and feedback
Volume weighted sims
Andrea Negri – AGN feeding and feedback
Take home points● At high resolution high oscillating BH accretion: self-limitedself-limited evolution, kept sub-Eddington without artificial limiters
● The common assumption of low resolution = low accretion is not verified not verified in presence of feedback
● EfficencyEfficency of (mechanical) feedback in stopping accretion is low at low resolutionlow at low resolution
●Accretion boosting Accretion boosting is justified only with volume weighted algorithms (SPH and Illustris simulation) but you need to be very careful!
●alpha is resolution and problem dependant
Additional slides
Simulations in a nutshell
Code: ZEUSMP (modified in Novak et al. 2011)
●2D axisymmetric
●Spherical coordinates
●r from 0.1 pc to 250 kpc 0.1 pc to 250 kpc
●NO gas self-gravity (important in 3D)
●SN Ia & SNII
●Mechanical Feedback Mechanical Feedback from broad absorption line (BAL) winds
●Star formation (KS prescription)
●Compton heating/cooling, radiation pressure
●Radiative cooling
Andrea Negri – AGN feeding and feedback
●Spherical stellar population (Jaffe model) Reff = 5,3 kpc M
star = 3 × 1010 M☉
●BH potential well MBH
=3 × 107 M☉
+ Singular Isothermal Sphere (vc=150 km/s)
●Initial ISM density profile: flat at low r, at large radii r -2
●Two different central ISM density: 1 and 0.01 cm -3
● (Almost) no rotation
Initial conditions
Andrea Negri – AGN feeding and feedback
●Spherical stellar population (Jaffe model) Reff = 5,3 kpc M
star = 3 × 1010 M☉
●BH potential well MBH
=3 × 107 M☉
+ Singular Isothermal Sphere (vc=150 km/s)
●Initial ISM density profile: flat at low r, at large radii r -2
●Two different central ISM density: 1 and 0.01 cm -3
● (Almost) no rotation
Initial conditions
Andrea Negri – AGN feeding and feedback
MBH
= 3 × 107 M ☉ with 0.1 pc resolution
we are able to resolve the Bondi radius up to T = 108 K
●Spherical stellar population (Jaffe model) Reff = 5,3 kpc M
star = 3 × 1010 M☉
●BH potential well MBH
=3 × 107 M☉
+ Singular Isothermal Sphere (vc=150 km/s)
●Initial ISM density profile: flat at low r, at large radii r -2
●Two different central ISM density: 1 and 0.01 cm -3
● (Almost) no rotation
Initial conditions
Andrea Negri – AGN feeding and feedback
BH accretion is computed as the fluxflux on the innermost radial grid
MBH
= 3 × 107 M ☉ with 0.1 pc resolution
we are able to resolve the Bondi radius up to T = 108 K