The topic of approximate methods is a growing area in which alternative techniques are used to generate simulated catalogs instead of running computationally expensive full N-body simulations (the latter taking typically thousands of CPUs for many days). Approximate methods allow running massive ensembles of simulations. In this line, during my PhD I developed ICE-COLA (Izard et al. 2016, 2018), a tool for producing fast cosmological simulation with some unique capabilities.
ICE-COLA is a parallel code for fast cosmological simulations, that reduces both the CPU-time and the storage requirements by 2-3 orders of magnitude with respect to traditional methods based on full N-body simulations and semi-analytic methods. It has the advantage of producing at run time catalogs in the light cone format directly, in which distant objects are seen as they were in the past, and in this way a huge compression factor in the data volume is achieved. I demonstrated that an approximate method can reach enough accuracy to simulate weak gravitational lensing experiments, enabling multi-probe analysis of galaxy clustering and weak gravitational lensing.
I’ve also been working on a pipeline to pin galaxies into simulated dark matter distributions. All my tools together constitute a very efficient methodology to connect fundamental theory to observations and thereby to sample very large cosmological volumes and explore model parameters. The final product are mock galaxy catalogs, very valuable in the preparation of upcoming galaxy surveys.
In my most ambitious simulation effort, I generated 500 ICE-COLA simulations covering very large volumes. I used them to develop an analytical model for the observational systematics caused by galaxy neighbors (one example of this being the effect called blending). Mitigating such systematics is imperative for the next generation of experiments that will have high number densities, such as the projects that I am involved in: Euclid space mission, the Wide Field Infrared Survey Telescope (WFIRST), and the Large Synoptic Survey Telescope (LSST)
More to come on: how to mitigate the impact of observational systematic effects, and on estimating covariance matrices of observables from simulations!