Installation

Requirements

Python Version

Goku-ELG requires Python 3.8 or higher. Python 3.12 is recommended for optimal performance.

Dependencies

Core dependencies:

numpy
scipy >= 1.15.1
matplotlib
scikit-learn
h5py
emukit
GPflow (for Gaussian Process models)

Installation Steps

Step 1: Create a Conda Environment

We recommend using a dedicated conda environment:

conda create -n gal_goku python=3.12 numpy scipy=1.15.1 matplotlib
conda activate gal_goku

Step 2: Install Additional Dependencies

Install the required Python packages:

python -m pip install emukit
python -m pip install scikit-learn
python -m pip install h5py

Step 3: Install Goku-ELG

There are two main packages to install:

Option A: Install both packages

# Navigate to the gal_goku package
cd src/gal_goku
python -m pip install -e .

# Navigate to the gal_goku_sims package
cd ../gal_goku_sims
python -m pip install -e .

Option B: Install from the root directory

If you have a setup script in the root:

python -m pip install -e .

Special Dependencies

ClassyLSS (Optional but Recommended)

For computing linear power spectra, the package uses classylss. The installation script will automatically clone and set up the required repositories:

These will be cloned automatically during installation of the gal_goku package.

Manual Installation

If you need to install classylss manually:

# Clone repositories
git clone https://github.com/sbird/classylss.git
git clone https://github.com/lesgourg/class_public.git

# Copy external data
cp -r class_public/external/bbn classylss/classylss/data/

# Install classylss
cd classylss
python -m pip install -e .

Verification

To verify your installation, try importing the packages:

import gal_goku
from gal_goku import emus, gal
from gal_goku_sims import hmf, xi

print("Installation successful!")

Running Tests

You can verify the installation by running a simple test:

import numpy as np
from gal_goku import emus

# This should run without errors if installation is correct
print("Goku-ELG is ready to use!")

Troubleshooting

Common Issues

Import Errors

If you encounter import errors, make sure you’ve activated the correct conda environment:

conda activate gal_goku

Missing Dependencies

If specific dependencies are missing, install them individually:

pip install <package-name>

ClassyLSS Issues

ClassyLSS requires a C compiler. On Linux, ensure you have gcc installed:

sudo apt-get install gcc  # Ubuntu/Debian
sudo yum install gcc       # CentOS/RHEL

Development Installation

For developers who want to contribute:

# Clone the repository
git clone https://github.com/qezlou/private-gal-emu.git
cd private-gal-emu

# Create development environment
conda create -n gal_goku_dev python=3.12
conda activate gal_goku_dev

# Install in editable mode with development dependencies
cd src/gal_goku
pip install -e ".[dev]"

GPU Support (Optional)

For accelerated Gaussian Process training with GPUs, install TensorFlow with GPU support:

pip install tensorflow-gpu

# Verify GPU is available
python -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"

Next Steps

After installation, check out the Quick Start Guide guide to learn how to use Goku-ELG.