How Does Feedback Affect Milky Way Satellite Formation?

How Does Feedback Affect Milky Way Satellite Formation?


We use sub-parsec resolution hydrodynamic resimulations of a Milky Way (MW) like galaxy at high redshift to investigate the formation of the MW satellite galaxies. More specifically, we assess the impact of supernova feedback on the dwarf progenitors of these satellite, and the efficiency of a simple instantaneous reionisation scenario in suppressing star formation at the low-mass end of this dwarf distribution. Identifying galaxies in our high redshift simulation and tracking them to z=0 using a dark matter halo merger tree, we compare our results to present-day observations and determine the epoch at which we deem satellite galaxy formation must be completed. We find that only the low-mass end of the population of luminous subhalos of the Milky-Way like galaxy is not complete before redshift 8, and that although supernovae feedback reduces the stellar mass of the low-mass subhalos (MM ), the number of surviving satellites around the Milky-Way like galaxy at z = 0 is the same in the run with or without supernova feedback. If a luminous halo is able to avoid accretion by the Milky-Way progenitor before redshift 3, then it is likely to survive as a MW satellite to redshift 0.


A Universe of Dwarf Galaxies, Lyon 2010

1 Introduction

The “Missing Satellite Problem” arose from the discovery that pure N-body simulations of MW-sized halos contain at least an order of magnitude more dark substructures than observed luminous satellites ([17, 13]). Although this discrepancy has recently been alleviated by the detection of a host of of new MW satellites ([14, 29]), a non negliglible (factor of a few) overabundance of dark substructures remains. Processes such as reionisation ([6, 11, 4, 25]) and supernova feedback ([7, 3]) have been invoked to reduce the number of satellite galaxies by heating and expelling potentially star-forming gas from low-mass halos.

To test this claim, we are running extremely high resolution hydrodynamic cosmological resimulations of a MW-like galaxy and its surroundings, improving on the work of other authors ([19, 31]) by at least an order of magnitude in physical resolution.

In this paper, we address the issue of whether MW satellite formation was complete by the epoch of reionisation or not. In addition, since we run identical simulations, but with and without supernova feedback, we assess the role of supernovae during the formation of satellite galaxies at high redshift. Since our resimulations containing star formation do not reach z=0, we compare our results to observations of MW satellites by twinning the halos containing galaxies in these high-redshift simulations with halos in an adiabatic gas simulation run to z=0. Along the lines of [16, 26], we track dark matter halos to z=0 to determine the properties of surviving halos. We then proceed to quantify the effect of various physical processes on dwarf galaxy formation at high redshift, and determine the importance of such processes in shaping the main properties of the MW satellite galaxies we see today.

2 Numerical Simulations

Figure 1: Comparison of stellar masses of galaxies at z=8.5 in the Cooling and Feedback run whose particles end up inside the MW virial radius at z=0, showing the fractional difference between the values in the two runs vs total virial mass in Adiabatic Run. A positive value indicates that the inclusion of supernova feedback enhances star formation, while a negative value means that it suppresses star formation. The circle size indicates stellar mass (see legend). Red circles indicate halos that survive as MW satellites to z=0; blue circles are completely disrupted by merger events by z=0.

In this work, we analyse three resimulations of a Milky Way-like galaxy from the (Nut) suite of simulations (Powell:2010inprep, in prep) named “Adiabatic”, “Cooling” and “Feedback”. We carry out these simulations using the Adaptive Mesh Refinement code RAMSES ([28]). All runs contain dark matter and baryonic gas. For the Cooling and Feedback runs, we include metal dependent radiative energy losses by the gas and its heating by a spatially-uniform UV background ([12]) after z=8.5, along with star formation. The threshold density for star formation is atoms/cm with an efficiency of 1%. For the Feedback run, we include supernova feedback, which also enriches the gas with metals. These simulations have a maximum physical resolution of 0.5pc, with a minimum mass of 5.6M for dark matter particles and 76M for star particles ([10]). With this mass resolution we are able to capture individual supernovae, although we still treat star particles as populations of stars rather than individual stars.

3 Tracking Halos

We identify dark matter (sub)halos and clusters of stars using HaloMaker with the Most Massive Subhalo Method ([30]). We then associate each cluster of stars to its nearest dark matter (sub)halo. For the adiabatic run, we use TreeMaker ([30]) to build a halo merger tree of the MW-like halo from our initial conditions to z=0. We then attach all dark matter (sub)halos in the Cooling and Feedback runs to their nearest “twins” in the Adiabatic run (based on the number of dark matter particles shared between the two halos) for a selection of redshifts before and after reionisation between z=11 and z=8 ([10]).

Using the Adiabatic run merger tree, we track each “twinned” halo to z=0 and find whether it has survived as an independent halo, as a subhalo of the MW, or has been completely disrupted in a merger with the MW or another halo. We posit that since all our galaxies are embedded within a dark matter (sub)halo, if this host halo survives to z=0 then so does the embedded galaxy.

From the 87 galaxies formed in the Feedback run at z=8.5 (just prior to reionisation) we find 13 galaxies survive as MW satellites to z=0, with 4 being destroyed in mergers with other satellite progenitor galaxies (6 in the Cooling run) and 71 destroyed in mergers with the MW (69 in the Cooling run). At z=8 (51 Myr after reionisation), one further satellite galaxy progenitor is formed that survives to z=0.

4 Feedback and Reionisation

Figure 2: Star formation histories in the Cooling run (left) and Feedback run (right). The vertical line at z=8.5 (look-back time 13.065Gyr) shows the point at which reionisation is turned on in the simulations. The grey region shows the detectable star formation rates as determined by [32].

Comparing the Cooling and Feedback runs, we find that galaxies with total mass under M generally exhibit suppressed star formation when our model of supernova feedback is included. By contrast, galaxies with total mass over M have similar masses in both runs, with some galaxies showing signs of enhanced star formation when supernova feedback is included (figure 1).

The latter is a consequence of enhanced cooling and hence higher star formation rate due to the introduction of metals from supernovae into the interstellar medium. However, supernova feedback does not appear to affect the number of galaxies formed, with roughly similar numbers of galaxies formed in both the Cooling and Feedback run. Further, the same MW satellites are found in both the Cooling and Feedback run when tracked to z=0 as described in section 3.

We find that reionisation does not immediately halt star formation, even in the smallest halos, in agreement with [31] (see figure 2). There appears to be a slight downward trend in the star formation rate after reionisation, however. As the simulations continue beyond 200Myr after reionisation, we will monitor whether this trend continues. Since we implement a uniform ionising background, our galaxies cannot self-shield. We plan to address this issue in future studies by including radiative transfer of hydrogen-ionising radiation.

5 MW Satellites Today

Figure 3: Cumulative velocity dispersion distributions of halos with galaxies and/or star clusters embedded in the Cooling (Blue) and Feedback (Red) runs, tracked to z=0 as explained in section 3. Observations (black line) use data from ([15, 1, 2, 33]). The simulated data is given as the maximum circular velocity / . The error bars represent the sampling error on the data in each bin.

As described in section 3, we track the halos containing galaxies in the Cooling and Feedback runs to z=0 using the Adiabatic run merger tree. We find that the most important criterion for a halo to survive to z=0 is its redshift of capture, with halos captured before z=3 being completely disrupted by z=0. However, this redshift value may be dependent on the host mass.

In figure 3, we compare the observed stellar velocity dispersions to our simulated galaxies’ dark matter velocity dispersions at z=0 in the Adiabatic run, since estimates of virial mass from observations have typically large errors [33]. For all sampled redshifts (z=11 to z=8), we find too few satellites when compared with the observed MW satellites, with better agreement if we wait until lower redshift before sampling the galaxies. Assuming this trend continues, this suggests that satellite galaxy formation is not complete by reionisation. Indeed, we find one galaxy formed between z=8.5 and z=8, after our implemented epoch of reionisation. Furthermore, [14, 29] suggest that the current sample of observed MW satellites is incomplete. Since our runs containing star formation have not reached z=0, we cannot comment on the completeness of our simulations.

6 Conclusions

We find that the velocity dispersions of satellites we produce underproduces those of observed Milky Way satellites, suggesting that satellite progenitor formation is not complete by reionisation. Star and galaxy formation continue past reionisation, with little sign of reionisation immediately halting the star formation rate. Supernova feedback reduces the star formation in halos with total mass M , but not the number of galaxies formed. Finally, we find only halos captured after z=3 survive to z=0 as Milky Way satellites.

Acknowledgements The authors would like to thank Romain Teyssier for useful comments and discussions during the work presented here. The simulations were performed on the JADE supercomputer at CINES, France.


  1. (1) Bekki K., Chiba M., 2005, Monthly Notices of the Royal Astronomical Society, 356, 680
  2. (2) Bekki K., Stanimirovic S., 2009, Monthly Notices of the Royal Astronomical Society, 395, 342 (2009)
  3. (3) Benson A. J., Bower R. G., Frenk C. S., Lacey C. G., Baugh C. M., Cole S., 2003, The Astrophysical Journal, 599, 38
  4. (4) Benson A. J., Lacey C., Baugh C., Cole S., Frenk C., 2002, Extragalactic Gas at Low Redshift, 254, 354
  5. (5) Bertschinger E., 2001, The Astrophysical Journal Supplement Series, 137, 1
  6. (6) Bullock J. S., Kravtsov A. V., Weinberg D. H., 2000, The Astrophysical Journal, 539, 517
  7. (7) Dekel A., Silk J., 1986, Astrophysical Journal, 303, 39
  8. (8) Dubois Y., Teyssier R., 2008, Astronomy and Astrophysics, 477, 79
  9. (9) Dunkley J., Komatsu E., Nolta M. R., Spergel D. N., Larson D., Hinshaw G., Page L., Bennett C. L., Gold B., Jarosik N., Weiland J. L., Halpern M., Hill R. S.,Kogut A., Limon M., Meyer S. S., Tucker G. S., Wollack E., Wright E. L., 2009, The Astrophysical Journal Supplement, 180, 306
  10. (10) Geen S., Slyz A., Devriendt J., 2011, in preparation
  11. (11) Gnedin N. Y., 2000, The Astrophysical Journal, 542, 535
  12. (12) Haardt F., Madau P., 1996, Astrophysical Journal 461, 20
  13. (13) Klypin A., Kravtsov A. V., Valenzuela O., Prada F., 1999, The Astrophysical Journal, 522, 82
  14. (14) Koposov S., Belokurov V., Evans N. W., Hewett P. C., Irwin M. J., Gilmore G., Zucker D. B., Rix H.-W., Fell- hauer M., Bell E. F., Glushkova E. V., 2008, The Astro- physical Journal, 686, 279
  15. (15) Mateo M. L., 1998, Annual Review of Astronomy and Astrophysics, 36, 435
  16. (16) Moore B., Diemand J., Madau P., Zemp M., Stadel J., 2006, Monthly Notices of the Royal Astronomical Society, 368, 563
  17. (17) Moore B., Ghigna S., Governato F., Lake G., Quinn T., Stadel J., Tozzi P., 1999, The Astrophysical Journal, 524, L19
  18. (18) Murante G., Monaco P., Giovalli M., Borgani S., Diaferio A., 2010, Monthly Notices of the Royal Astronomical Society, 405, 1491
  19. (19) Okamoto T., Frenk C. S., 2009, Monthly Notices of the Royal Astronomical Society: Letters, 399, L174
  20. (20) Powell L., Slyz A., Devriendt J., 2010, in preparation
  21. (21) Primack J. R., 2009, New Journal of Physics, 11, 5029
  22. (22) Prunet S., Pichon C., Aubert D., Pogosyan D., Teyssier R.,Gottloeber S., 2008, The Astrophysical Journal Supplement Series, 178, 179
  23. (23) Scannapieco C., Tissera P. B., White S. D. M., SpringelV., 2005 , Monthly Notices of the Royal Astronomical Society, 364, 552
  24. (24) Scannapieco C., Tissera P. B., White S. D. M., Springel V., 2006, Monthly Notices of the Royal Astronomical Society, 371, 1125
  25. (25) Somerville R. S., 2002, The Astrophysical Journal, 572,L23
  26. (26) Strigari L. E., Bullock J. S., Kaplinghat M., Diemand J.,Kuhlen M., Madau P., 2007, The Astrophysical Journal,669, 676
  27. (27) Taylor J. E., Babul A., 2004, Monthly Notices of the Royal Astronomical Society, 348, 811
  28. (28) Teyssier R., 2002, Astronomy and Astrophysics, 385, 337
  29. (29) Tollerud E. J., Bullock J. S., Strigari L. E., Willman B., 2008, The Astrophysical Journal, 688, 277
  30. (30) Tweed D., Devriendt J., Blaizot J., Colombi S., Slyz A., 2009, Astronomy and Astrophysics, 506, Issue 2, pp.647-660
  31. (31) Wadepuhl M., Springel V., 2010, eprint arXiv, 1004, 3217
  32. (32) Wilkins S. M., Bunker A. J., Lorenzoni S., Caruana J., 2010, eprint arXiv, 1002, 4866
  33. (33) Wolf J., Martinez G. D., Bullock J. S., Kaplinghat M.,Geha M., Munoz R. R., Simon J. D., Avedo F. F., 2010, Monthly Notices of the Royal Astronomical Society, p. 778
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minimum 40 characters and the title a minimum of 5 characters
Add comment
Loading ...
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description