The RULES Project

RULES (Reconstructing the Universe from empiricaL knowledgE of galaxieS) builds a mock Universe directly from the empirical knowledge of galaxies, rather than from physical models of galaxy formation. Following an “observations-to-observations” methodology, RULES translates key observational relations — such as the stellar mass function and quiescent fraction derived from deep JWST surveys — into mock catalogs and sky images spanning z = 0 to z ~ 8, with a comprehensive set of galaxy properties.

The project chains three independently upgradable components in sequence: empirical knowledge, property generation — currently powered by the Empirical Galaxy Generator (EGG) — and image generation. Instead of relying on resolution-limited subgrid models and uncertain theoretical prescriptions, RULES produces a representative galaxy population that matches observations across all redshifts and over a wide range of physical properties, including stellar masses, galaxy sizes, and multi-band fluxes. Its goal is not to explain the underlying physics, but to provide a precise observational benchmark from which a variety of derived data products — custom sky catalogs and realistic mock images — can be generated rapidly without expensive computations.

Figure 1. Schematic diagram of RULES. From deep JWST observations we derive empirical relations of galaxies (e.g., the stellar mass function), use them to generate mock galaxy properties (e.g., redshift, stellar mass, SED), and then build mock images, producing a self-consistent mock Universe.

VISION — The First Application of RULES

VISION (Versatile Integrated Simulator for Inquiry into Observational Needs) is built as the first application of RULES: an automated pipeline designed to evaluate or predict the performance of arbitrary photometric surveys. Building on the RULES mock Universe, VISION inherits its empirical accuracy and rapid generation, and automatically benefits from its future improvements. It generates mock catalogs of shallower surveys that mimic actual observations by adding instrumental effects, and then predicts their performance.

VISION focuses on two key metrics: mass completeness limits, which define the stellar mass above which a survey is complete for a given fraction of the true galaxy population, and photometric redshift accuracy, which underpins cosmological probes such as weak gravitational lensing. It incorporates a realistic noise model and an SNR cut, performs photo-z fitting with EAZY, and includes a built-in library of about 170 filters from major ground- and space-based facilities. VISION has been validated against COSMOS-Web observations, showing good agreement for the same filter set, and enables rapid iteration over observational designs for next-generation facilities such as CSST, Euclid, and Roman — without the need for expensive cosmological simulations.

Figure 2. Schematic diagram of VISION. Based on the RULES mock Universe, VISION takes user-defined observational parameters (sky region, filter set, SNR cut, etc.) as input, and then rapidly computes mass completeness limits and photometric redshift performance, along with the corresponding mock catalogs and images, for any specific survey, such as CSST, Euclid, and Roman.

Website maintained by: Xi Wang
Questions: 221840241@smail.nju.edu.cn