Python 3 version of superFATBOY GPU accelerated data pipeline for IR and optical astronomical data
- numpy
- scipy
- astropy
- matplotlib (optional)
- sep or sextractor (optional)
- CUDA, PyCUDA, and CuPy (optional)
- deepCR (optional)
- To install optional GPU libraries, first use Makefile to build CUDA code, first make sure that you set environment variable
CUDA_HOME
to point to your install directory for CUDA (e.g./usr/local/cuda
). Optionally you may also setPYTHON3_INCLUDE
to e.g./usr/include/python3.8
but this should be auto-detected and not necessary to set manually unless it complains that it can't find Python.h. Then
cd superFATBOY3/superFATBOY
make gpu3
- To install as sudo:
cd superFATBOY3
sudo python3 setup.py install
- To install and run superFATBOY3 on HiperGator or other HPC environments, try this guide.
- YMMV on versions of GCC, CUDA, and Python depending on your particular HPC environment but the below works on HiperGator Nov 2023. Note that you must choose a compatible GCC and CUDA (sometimes the newest GCC is too new for the newest CUDA).
module load conda gcc/9.3.0 cuda/11.4.3
conda create --name sFB3 python=3.9 cupy numpy scipy astropy matplotlib pycuda
conda activate sFB3
cd superFATBOY3/superFATBOY/
make gpu3
cd ..
python setup.py install
module load conda gcc/9.3.0 cuda/11.4.3
conda activate sFB3
superFatboy3.py [-list] [filename.xml]