Xiaolong Du's work focuses on cosmological structure formation in different dark matter models and on constraining these models from gravitational effects such as substructure lensing. Using numerical simulations and semi-analytic models, he tries to find an accurate, but fast, way to study the behaviors of dark matter on sub-galactic scales, where different dark matter candidates may behave very differently.

Du actively works on developing numerical tools for simulating one of the promising dark matter candidates called Fuzzy Dark Matter (FDM) or Ultralight axions (ULAs). In this model, dark matter behaves like waves instead of collisionless particles, presenting many interesting phenomena on galactic scales and providing one potential way to solve the small-scale problems faced by the standard Cold Dark Matter. He developed numerical codes, both spectral code and SPH code, to simulate the structure formation with FDM. Together with Andrew Benson, Fangzhou Jiang and Shengqi Yang, he also works on implementing the FDM physics into the public semi-analytic code for galaxy formation, Galacticus.

Du is also interested in compact astrophysical objects, e.g. the formation of boson stars, radio signal from the encounter of boson stars with a neutron star. These astrophysical objects may help us to understand the properties of the invisible dark matter particles in the universe.