We study cosmic voids in the normal-branch Dvali-Gabadadze-Porrati (nDGP) braneworld models, which are representative of a class of modified gravity theories where deviations from General Relativity are usually hidden by the Vainshtein screening in high-density environments. This screening is less efficient away from these environments, which makes voids ideally suited for testing this class of models. We use N-body simulations of $\Lambda$-cold dark matter ($\Lambda$CDM) and nGDP universes, where dark matter haloes are populated with mock galaxies that mimic the clustering and number densities of the BOSS CMASS galaxy sample. We measure the force, density and weak lensing profiles around voids identified with six different algorithms. Compared to $\Lambda$CDM, voids in nDGP are more under-dense due to the action of the fifth force that arises in these models, which leads to a faster evacuation of matter from voids. This leaves an imprint on the weak lensing tangential shear profile around nDGP voids, an effect that is particularly strong for 2D underdensities that are identified in the plane-of-the-sky. We make predictions for the feasibility of distinguishing between nDGP and $\Lambda$CDM using void lensing in upcoming large-scale surveys such as LSST and Euclid. We compare with the analysis of voids in chameleon gravity theories and find that the weak lensing signal for 3D voids is similar to nDGP, whereas for 2D voids the differences with $\Lambda$CDM are much stronger for the chameleon gravity case, a direct consequence of the different screening mechanisms operating in these theories.

We present a systematic comparison of several existing and new void finding algorithms, focusing on their potential power to test a particular class of modified gravity models - chameleon $f(R)$ gravity. These models deviate from standard General Relativity (GR) more strongly in low-density regions and thus voids are a promising venue to test them. We use Halo Occupation Distribution (HOD) prescriptions to populate haloes with galaxies, and tune the HOD parameters such that the galaxy two-point correlation functions are the same in both f(R) and GR models. We identify both 3D voids as well as 2D underdensities in the plane-of-the-sky to find the same void abundance and void galaxy number density profiles across all models, which suggests that they do not contain much information beyond galaxy clustering. However, the underlying void dark matter density profiles are significantly different, with f(R) voids being more underdense than GR ones, which leads to f(R) voids having a larger tangential shear signal than their GR analogues. We investigate the potential of each void finder to test f(R) models with near-future lensing surveys such as EUCLID and LSST. The 2D voids have the largest power to probe f(R) gravity, with a LSST analysis of tunnel (which is a new type of 2D underdensity introduced here) lensing distinguishing at 80 and 11$\sigma$ (statistical error) f(R) models with $|f_{R0}|=10^{-5}$ and $10^{-6}$ from GR.

In theories of modified gravity with the chameleon screening mechanism, the strength of the fifth force depends on environment. This induces an environment dependence of structure formation, which differs from $\Lambda$CDM. We show that these differences can be captured by the marked correlation function. With the galaxy correlation functions and number densities calibrated to match between $f(R)$ and $\Lambda$CDM models in simulations, we show that the marked correlation functions from using either the local density or halo mass as the marks encode extra information, which can be used to test these theories. We discuss possible applications of these statistics in observations.