Abstract
How mergers affect galaxy formation depends on both feedback processes, and on the geometry and strength of the mergers themselves. We introduce the PARADIGM project, where we study the response of a simulated Milky-Way-mass galaxy (M-200c similar to 10(12) M-circle dot at z=0) forming in a cosmological setting to differing merger histories, using genetically modified initial conditions, each simulated with the VINTERGATAN and IllustrisTNG codes. While VINTERGATAN has been developed with an emphasis on resolving the cold interstellar medium, IllustrisTNG uses a subgrid two-phase model and consequently scales to large volume simulations, making them ideal to examine complementary views on how merger histories and feedback interact. Our genetic modifications alter the mass ratio of an important z approximate to 2 merger while maintaining the halo's z=0 mass. Whether simulated with VINTERGATAN or IllustrisTNG, smaller mass ratios for this early merger result in larger galaxies at z=0, due to a greater build-up of a kinematically cold disc. We conclude that such broad trends are robustly reproducible; however, the normalization of the resulting stellar sizes is substantially different in the two codes (ranging between 0.5 and,1.7 kpc for VINTERGATAN but 1.3-7.0 kpc for IllustrisTNG). The VINTERGATAN galaxies systematically form stars earlier, leading to a larger bulge component. Despite the difference in size normalization, both simulation suites lie on the observed size-mass relation for their respective morphological types. In light of these results, we discuss the interplay between internal processes and large-scale gravitational interactions and gas accretion, and how the two galaxy models converge on similar emergent trends but along different evolutionary pathways.