Our statistical understanding of galaxy evolution is fundamentally driven by objects that lie above the surface-brightness limits of current wide-area surveys (µ ∼ 23 mag arcsec −2 ). While both theory and small, deep surveys have hinted at a rich population of lowsurface-brightness galaxies (LSBGs) fainter than these limits, their formation remains poorly understood. We use Horizon-AGN, a cosmological hydrodynamical simulation to study how LSBGs, and in particular the population of ultra-diffuse galaxies (UDGs; µ > 24.5 mag arcsec −2 ), form and evolve over time. For M * >10 8 M , LSBGs contribute 47, 7 and 6 per cent of the local number, mass and luminosity densities respectively (∼85/11/10 per cent for M * >10 7 M ). Today's LSBGs have similar dark-matter fractions and angular momenta to high-surface-brightness galaxies (HSBGs; µ < 23 mag arcsec −2 ), but larger effective radii (×2.5 for UDGs) and lower fractions of dense, star-forming gas (more than ×6 less in UDGs than HSBGs). LSBGs originate from the same progenitors as HSBGs at z > 2. However, LSBG progenitors form stars more rapidly at early epochs. The higher resultant rate of supernova-energy injection flattens their gas-density profiles, which, in turn, creates shallower stellar profiles that are more susceptible to tidal processes. After z ∼ 1, tidal perturbations broaden LSBG stellar distributions and heat their cold gas, creating the diffuse, largely gas-poor LSBGs seen today. In clusters, ram-pressure stripping provides an additional mechanism that assists in gas removal in LSBG progenitors. Our results offer insights into the formation of a galaxy population that is central to a complete understanding of galaxy evolution, and which will be a key topic of research using new and forthcoming deep-wide surveys.

Hydrodynamical cosmological simulations are increasing their level of realism by considering more physical processes and having greater resolution or larger statistics. However, usually either the statistical power of such simulations or the resolution reached within galaxies are sacrificed. Here, we introduce the NEWHORIZON project in which we simulate at high resolution a zoom-in region of ∼(16 Mpc)3 that is larger than a standard zoom-in region around a single halo and is embedded in a larger box. A resolution of up to 34 pc, which is typical of individual zoom-in, up-to-date resimulated halos, is reached within galaxies; this allows the simulation to capture the multi-phase nature of the interstellar medium and the clumpy nature of the star formation process in galaxies. In this introductory paper, we present several key fundamental properties of galaxies and their black holes, including the galaxy mass function, cosmic star formation rate, galactic metallicities, the Kennicutt–Schmidt relation, the stellar-to-halo mass relation, galaxy sizes, stellar kinematics and morphology, gas content within galaxies and its kinematics, and the black hole mass and spin properties over time. The various scaling relations are broadly reproduced by NEWHORIZON with some differences with the standard observables. Owing to its exquisite spatial resolution, NEWHORIZON captures the inefficient process of star formation in galaxies, which evolve over time from being more turbulent, gas rich, and star bursting at high redshift. These high-redshift galaxies are also more compact, and they are more elliptical and clumpier until the level of internal gas turbulence decays enough to allow for the formation of discs. The NEWHORIZON simulation gives access to a broad range of galaxy formation and evolution physics at low-to-intermediate stellar masses, which is a regime that will become accessible in the near future through surveys such as the LSST.

Massive black hole (MBH) coalescences are powerful sources of low-frequency gravitational waves. To study these events in the cosmological context we need to trace the large-scale structure and cosmic evolution of a statistical population of galaxies, from dim dwarfs to bright galaxies. To cover such a large range of galaxy masses, we analyse two complementary simulations: Horizon-AGN with a large volume and low resolution which tracks the high-mass (&gt;107 M⊙) MBH population, and NewHorizon with a smaller volume but higher resolution that traces the low-mass (&lt;107 M⊙) MBH population. While Horizon-AGN can be used to estimate the rate of inspirals for Pulsar Timing Arrays, NewHorizon can investigate MBH mergers in a statistical sample of dwarf galaxies for LISA, which is sensitive to low-mass MBHs. We use the same method to analyse the two simulations, post-processing MBH dynamics to account for time delays mostly determined by dynamical friction and stellar hardening. In both simulations, MBHs typically merge long after galaxies do, so that the galaxy morphology at the time of the MBH merger is no longer determined by the structural disturbances engendered by the galaxy merger from which the MBH coalescence has originated. These time delays cause a loss of high-z MBH coalescences, shifting the peak of the MBH merger rate to z ∼ 1 − 2. This study shows how tracking MBH mergers in low-mass galaxies is crucial to probing the MBH merger rate for LISA and investigate the properties of the host galaxies.

In the standard ΛCDM (Lambda cold dark matter) paradigm, dwarf galaxies are expected to be dark matter-rich, as baryonic feedback is thought to quickly drive gas out of their shallow potential wells and quench star formation at early epochs. Recent observations of local dwarfs with extremely low dark matter content appear to contradict this picture, potentially bringing the validity of the standard model into question. We use NewHorizon, a high-resolution cosmological simulation, to demonstrate that sustained stripping of dark matter, in tidal interactions between a massive galaxy and a dwarf satellite, naturally produces dwarfs that are dark matter-deficient, even though their initial dark matter fractions are normal. The process of dark matter stripping is responsible for the large scatter in the halo-to-stellar mass relation in the dwarf regime. The degree of stripping is driven by the closeness of the orbit of the dwarf around its massive companion and, in extreme cases, produces dwarfs with halo-to-stellar mass ratios as low as unity, consistent with the findings of recent observational studies. ∼30 per cent of dwarfs show some deviation from normal dark matter fractions due to dark matter stripping, with 10 per cent showing high levels of dark matter deficiency (Mhalo/M⋆ &lt; 10). Given their close orbits, a significant fraction of dark matter-deficient dwarfs merge with their massive companions (e.g. ∼70 per cent merge over time-scales of ∼3.5 Gyr), with the dark matter-deficient population being constantly replenished by new interactions between dwarfs and massive companions. The creation of these galaxies is therefore a natural by-product of galaxy evolution and their existence is not in tension with the standard paradigm.

Dwarf galaxies (M⋆ &lt; 109 M⊙) are key drivers of mass assembly in high mass galaxies, but relatively little is understood about the assembly of dwarf galaxies themselves. Using the NewHorizon cosmological simulation (∼40 pc spatial resolution), we investigate how mergers and fly-bys drive the mass assembly and structural evolution of around 1000 field and group dwarfs up to z = 0.5. We find that, while dwarf galaxies often exhibit disturbed morphologies (5 and 20 per cent are disturbed at z = 1 and z = 3 respectively), only a small proportion of the morphological disturbances seen in dwarf galaxies are driven by mergers at any redshift (for 109 M⊙, mergers drive under 20 per cent morphological disturbances). They are instead primarily the result of interactions that do not end in a merger (e.g. fly-bys). Given the large fraction of apparently morphologically disturbed dwarf galaxies which are not, in fact, merging, this finding is particularly important to future studies identifying dwarf mergers and post-mergers morphologically at intermediate and high redshifts. Dwarfs typically undergo one major and one minor merger between z = 5 and z = 0.5, accounting for 10 per cent of their total stellar mass. Mergers can also drive moderate star formation enhancements at lower redshifts (3 or 4 times at z = 1), but this accounts for only a few per cent of stellar mass in the dwarf regime given their infrequency. Non-merger interactions drive significantly smaller star formation enhancements (around two times), but their preponderance relative to mergers means they account for around 10 per cent of stellar mass formed in the dwarf regime.