C++ |

GalaPy

Wed, 01 May 2024 00:00:00 +0000

Role: lead author & maintainer. Stack: C++ compute core exposed through a Python API; hybrid parallelisation (vectorised array programming, shared-memory concurrency, distributed-memory message passing).

GalaPy simulates the panchromatic emission of galaxies from X-rays to radio and performs Bayesian spectral-energy-distribution fitting. Models are generated on the fly rather than interpolated from a template library, which is what allows a peak throughput of almost 1000 SEDs per second on a single core — the fastest SED-generation tool of its kind.

The library is designed around an object-oriented architecture that keeps it portable with minimal memory overhead, and it scales from a laptop to an HPC allocation without changing the user-facing API. Recent additions include parallel sampling for Bayesian inference and a fully analytical, panchromatic AGN component.

Since release, GalaPy has been adopted by groups across Europe, Asia and South America, for work ranging from nearby resolved galaxies to JWST high-redshift candidates. Its role in my wider programme is described in my .

Presented in Astronomy & Astrophysics 685, A161 (2024) — see — with the computational design detailed in Astronomy and Computing 55, 101079 (2026): .

SCAMPy

Sat, 01 Aug 2020 00:00:00 +0000

Role: author & maintainer. Stack: C++ core (object-oriented, polymorphic) wrapped in Python.

SCAMPy implements the Sub-halo Clustering and Abundance Matching (SCAM) scheme, which combines the classical Halo Occupation Distribution with sub-halo abundance matching. The procedure runs in two steps: the HOD prescription selects which sub-haloes host galaxies, then SHAM assigns an observable property of choice to each selected sub-halo. It requires only the 1- and 2-point statistics of the target population as input — typically an observed luminosity function and its clustering — which makes the method fully data-driven rather than dependent on a physical galaxy-formation model.

The core functionalities are written in C++ and make wide use of polymorphism for flexibility and computational efficiency; the whole library is wrapped in Python so it can be driven from a scripting interface or embedded in a larger pipeline. Mock catalogues produced this way reproduce both the abundance and the clustering of the target sample, and the approach applies across a wide redshift range, from high-redshift Lyman-break galaxies to low-redshift radio sources.

SCAMPy came out of my Ph.D. thesis and has since become a core ingredient in simulation pipelines used for SKAO preparatory science, including the full-sky radio mock catalogues described in my .

Presented in Monthly Notices of the Royal Astronomical Society 498, 2095 (2020) — see . Also indexed in the Astrophysics Source Code Library as ascl:2002.006.

CosmoBolognaLib

Wed, 01 Nov 2017 00:00:00 +0000

Role: contributor — void size function modelling and void-catalogue cleaning. Stack: C++ libraries with a SWIG-generated Python wrapper. Author & maintainer: Federico Marulli (Università di Bologna).

CosmoBolognaLib (CBL) is a living project that provides a common numerical environment for cosmological investigations of the large-scale structure of the Universe. Its focus is handling astronomical catalogues — real and simulated — measuring one-, two- and three-point statistics in configuration space, and running cosmological analyses.

My contribution is the set of tools for cosmic voids: the algorithm that redefines void ridges and therefore their radii, together with the model for the void size function. The cleaning procedure takes a void catalogue produced by a void finder and returns a catalogue of non-overlapping spheres, each embedding a fixed density contrast in the tracer density field — which is what makes the measured size function comparable with theoretical predictions. These functions were released in CBL v3.2 and are credited in the library’s changelog to Ronconi & Marulli (2017).

The tooling underpins the void work described in my , and was subsequently adopted in Key Projects within the Euclid Collaboration to forecast the cosmological constraining power of void statistics.

The void tools are presented in Astronomy & Astrophysics 607, A24 (2017) — see . The library itself is described in Marulli, Veropalumbo & Moresco (2016).