Follow the baryons

Galaxies falling into groups and clusters lose their gas and stop forming stars. I develop ways to measure the local environment of every galaxy and to model ram-pressure stripping in semi-analytical models, to find out when, where and how fast environment quenches galaxies, and how far its influence reaches.

• The gas around satellites shows no abrupt change at the virial radius, so stripping should begin before galaxies cross the halo boundary (Ayromlou et al. 2019).

• Modeling ram-pressure stripping for all galaxies improves how well L-Galaxies reproduces observed quenched fractions and star formation rates, across environment, stellar mass and redshift (Ayromlou et al. 2021).

• Galactic conformity (neighbors sharing the star-forming state of their central galaxy) reaches at least 5 Mpc in models and in SDSS and DESI data. Much, but not all, of it comes from galaxies near massive systems (Ayromlou et al. 2023).

JWST has revealed galaxies that are massive, UV-bright, compact and, in some cases, already quenched within the first two billion years, and most galaxy-formation models struggle to make them. With my students and collaborators, I test which physics can: star formation and feedback in dense gas, mergers, and environmental quenching at the highest redshifts.

Spoiler: not yet, but star formation and feedback regulated by the local gas density close much of the gap.

• Three versions of L-Galaxies broadly match observed scaling relations up to z ≈ 10, but they underpredict massive quenched galaxies, by up to a factor of 60 at z ≈ 3, and make them several times too large (Vani et al. 2025).

• Making star formation efficient and stellar feedback ineffective in dense gas, with dissipative gas-rich mergers, makes massive UV-bright galaxies at z ≳ 9 and massive quenched galaxies at z ≈ 8 to 3 up to a hundred times more abundant than in the previous model, without changing the AGN physics or the initial mass function. Its compact galaxies follow the UV size-luminosity relation JWST observes at z ≈ 13 to 10 (Vani et al. 2026).

• Among more than half a million simulated galaxies at z = 5, quenched galaxies are overwhelmingly satellites, although satellites are fewer than 10% of all galaxies. The low-mass quenched galaxies JWST finds may be quenched by their environment near massive halos, and often only temporarily (Döven et al. 2026).

• A semi-empirical model built directly on observed UV luminosity functions predicts star formation efficiencies of 0.8 to 0.9 at z > 9, falling to 0.2 to 0.3 at later times; dust raises them further, while top-heavy initial mass functions lower them (Fu et al. 2026).