A physically motivated framework for measuring the mass and redshift dependence of galaxy pair fractions across cosmic time

Low mass galaxy pair fractions are under-studied across cosmic time. In the era of JWST, Roman, and Rubin, a self-consistent framework is needed to select both low and high mass galaxy pairs to connect observed pair fractions to cosmological merger rates across all mass scales and redshifts. We use the Illustris TNG100 simulation to identify physically associated pairs between $z=0-4.2$. Our sample includes low mass ($\rm 10^8<M_*<5\times10^9\,M_{\odot}$) and high mass ($\rm 5\times10^9<M_*<10^{11}\,M_\odot$) isolated subhalo pairs, with stellar masses from abundance matching. The low mass pair fraction, i.e. the fraction of galaxies in pairs, increases from $z=0-2.5$, while the high mass pair fraction peaks at $z=0$ and is constant or slightly decreasing at $z>1$. At $z=0$ the low mass major (1:4 mass ratio) pair fraction is 4$\times$ lower than high mass pairs, consistent with findings for cosmological merger rates. Our results indicate that pair fractions can faithfully reproduce trends in merger rates if galaxy pairs are selected appropriately. Specifically, static pair separation limits applied equivalently to all galaxy pairs do not recover the evolution of low and high mass pair fractions. Instead, we advocate for separation limits that vary with the mass and redshift of the system, such as separation limits scaled by the virial radius of the host halo ($r_{\mathrm{sep}}< 1 R_{\rm vir}$). Finally, we place isolated mass-analogs of Local Group galaxy pairs (i.e., MW--M31, MW--LMC, LMC--SMC) in a cosmological context, showing that isolated analogs of LMC--SMC-mass pairs, and low separation ($<50$kpc) MW--LMC-mass pairs, are $2-3\times$ more common at $z\gtrsim2-3$.