Ultra-faint dwarf galaxies like Leo P and Pisces A have low stellar masses and a large neutral gas content. Simulating the formation of this type of galaxies has proven to be very difficult, since simulated galaxies with comparable stellar masses hardly contain any neutral gas and do not have ongoing star formation, contrary to what is observed. These systems are close to the lower mass limit for halos that are able to retain gas in a reionizing Universe, making them very sensitive to details in the physical model of the simulations.
We ran a large set of simulations of isolated dwarf galaxy halos to investigate the influence of different physical processes on the resulting star formation histories and baryonic contents of the galaxy. To this end, we made mock observations of our simulation results, to be able to compare them with the baryonic Tully-Fisher relation (BTFR) that was observed by McGaugh (2012).
The simulations were run using our in-house adapted version of the N-body/SPH-code Gadget2, extended with metallicity dependent gas cooling and heating in the presence of a time dependent UV background (UVB), an advanced equation of state and prescriptions for star formation using sink particles.
We also tested the effect of a novel implementation of the feedback delivered by metal poor Population III (Pop III) stars, based on Nomoto (2013) and Susa (2014). These stars have a significantly higher energy output compared to metal rich(er) Pop I and Pop II stars, which adds a time-dependence to the stellar feedback.
In total, 235 simulations were run with various initial conditions and code parameters. The name of the simulation consists of two parts: a 4 character code identifier (e.g. C1P1 and also C1P1bis) and a 6 character initial condition identifier. The latter specifies the total mass, initial rotation velocity and resolution of the initial conditions according to Table 1. The former specifies the detailed model that was used to evolve the initial conditions in time, see Table 2. Model C1P1bis is a special case, since this is exactly the same as C1P1, but with different random seeds for the initial conditions. This model was run to quantify the effect of stochastic differences on the outcome of the simulations.
| Mass parameter | Physical total mass (M\(_\odot\)) |
|---|---|
| M1 | \(1.0 \times 10^9\) |
| M3 | \(3.0 \times 10^9\) |
| M5 | \(5.0 \times 10^9\) |
| M7 | \(7.0 \times 10^9\) |
| M9 | \(9.0 \times 10^9\) |
| Rotation parameter | Physical rotation velocity (km/s) |
| R00 | \(0.0\) |
| R05 | \(5.0\) |
| R10 | \(10.0\) |
| Resolution parameter | Number of particles |
| L | \(2 \times 50,000\) |
| H | \(2 \times 200,000\) |
| Code name | UVB model | Stellar feedback strength | Pop III model |
|---|---|---|---|
| C1P1 | low and late | \(0.7\) | no Pop III stars |
| C2P1 | full and late | \(0.7\) | no Pop III stars |
| C3P1 | full and early | \(0.7\) | no Pop III stars |
| C3P2 | full and early | \(1.0\) | no Pop III stars |
| C3P3 | full and early | \(2.0\) | no Pop III stars |
| C4P2 | no UVB | \(1.0\) | no Pop III stars |
| C7P4 | full and early | \(0.7\) | Pop III model 1A: low feedback |
| C7P6 | full and early | \(0.7\) | Pop III model 1A: high feedback |
| C7P7 | full and early | \(0.7\) | Pop III model 1A: intermediate feedback |
| C9P8 | full and early | \(0.7\) | Pop III model 1B: low feedback |
| C9P9 | full and early | \(0.7\) | Pop III model 1B: high feedback |
| CaPa | full and early | \(0.7\) | Pop III model 2: high SW, low SN |
| CbPc | full and early | \(0.7\) | Pop III model 2: low SW, high SN |
| CcPd | full and early | \(0.7\) | Pop III model 3 |
The plots below are interactive: you can select between various code versions and initial condition parameters, and change the colors and symbols. You can also change between different plots: the BTFR plot (with circular velocities derived from the neutral gas or derived from the stellar velocity dispersion) and various scaling relations.
We also show some observational relations for comparison. The full blue line represents a fit to the observed BTFR from McGaugh (2012). The full black lines represent least squares fits to the simulation data. The full symbols are the low resolution simulations, while the open symbols have a 4 times higher resolution. The size of the symbols is related to the mass of the underlying halo: the lower the mass, the smaller the symbol.
Not all simulations are included. 5 simulations were discarded because they yielded totally unphysical results and are excluded from all plots. Furthermore, for some simulations it proved impossible to fit a Sérsic profile to the stellar luminosity profile, so that these simulations are not shown on the scaling relations. For simulations with no neutral gas content, it is impossible to derive a circular velocity from the neutral gas, which means about half the simulations are not shown on the BTFR with neutral gas derived velocities.
Hover over an element to see more information