arXiv version:

MNRAS version (possibly paywalled):


If you saw our pre-print version (posted in June, when we submitted out paper; “v1” on the arXiv page), then a few things have changed since then. The most major changes have been:

  • Our conclusions regarding resolution and mixing have changed. We previously thought the difference between our results and some previous results could be explained by their simulations being unresolved.  We no longer think that; the difference is more likely due to differences in the amount of real, physical mixing (as opposed to artificial, numerical mixing due to resolution). This change is due to a few things, including a misreading of a few key lines in a paper, and seeing preliminary not-yet-published results of a group’s resolution study.
  • We’ve included much more discussion on the possibility that galactic-scale effects might change our conclusions. Usually you don’t have to worry about this for supernova feedback models; the supernova remnant only expands ~10 pc over 0.1 Myr, which is small and fast enough to be able to ignore things like the bubble breaking out of the galactic disk. But, since our model sometimes results in much larger, long-lived bubbles, it was useful to check the effects of things like disk breakout and disk shear.  We found those effects could substantially change the results of some simulated clusters, but these clusters tend to be the most massive clusters, which are also the rarest in nature. So, when you average over a realistic distribution of cluster masses, our final results don’t change much.

On a more general note, I’m really happy how feedback from our arXiv posting supplemented the formal pier review from the journal. By posting our article at submission-time, rather than a journal acceptance-time, we got a lot of useful comments, which we could incorporate into the final accepted version.