Skip to content

phdenzel/skais-mapper

Repository files navigation

skais-mapper

https://img.shields.io/pypi/v/skais-mapper.svg https://img.shields.io/pypi/pyversions/skais-mapper.svg https://img.shields.io/badge/License-GPL%20v3-blue.svg

Table of Contents

skais-mapper is a tool for generating, plotting, and pre-processing hydrodynamics simulation (image) data for state-of-the-art generative AI models.

Requirements

skais-mapper is mostly built on python, but also includes some C extensions for the compute-intensive raytracing (building and visualizing datasets). Building from scratch thus requires cython, however skais ships with pre-compiled C files, making the minimal requirements

  • python >= 3.10
  • gcc (on linux) / clang (on macOS)

Also see pyproject.toml for the relevant python packages.

Install

It is recommended to install skais-mapper in a virtual environment via uv. For this, run

uv sync

Alternatively, you can simply run

python setup.py build_ext --inplace
pip install [-e] .

If you want to compile the C extension from the cython files directly, run in advance to the above

python setup.py build_c [-a]

On Nix(OS)

For Nix(OS) users, the repository includes a flake.nix file. It allows to create a development environment compatible with standard uv use.

Usage

skais-mapper implements a few sub-commands for generating and manipulating simulation data. Use the following to see what valid sub-commands exist:

[uv run] skais-mapper -h

skais-mapper sub-commands implement the hydra configuration management framework. For more information on sub-command usage, inspect the skais_mapper/configs/ directory, or use:

[uv run] skais-mapper [sub-command] -h

For instance, the command to generate 1000 images from snapshot 50 is as follows:

[uv run] skais-mapper generate +experiment=tng50-1-50-2D-0000-1000

Data

Currently, this framework is fully compatible with SPH data from the AREPO simulator, in particular the IllustrisTNG suite. It provides utility routines to fetch isolated halos from simulations snapshots and various raytracing algorithms for 2D column density projections of these halos and its galaxies. The framework generates HDF5 files with image datasets of various galactic properties, such as dark matter, star, or gas column density distributions.

License

skais-mapper is distributed under the terms of the GNU General Public License v3.0 or later license.

About

A framework for generating deep-learning SKA radio telescope & cosmological hydrodynamical simulation data

Topics

Resources

License

Stars

Watchers

Forks

Contributors 2

  •  
  •  

Languages