General Remarks

DNN reco is intended to be a highly modularized software framework for neural network applications in IceCube. As such, there are many modules that you can easily modify or switch out, depending on your needs. A central configuration file defines which modules and which settings will be used. This configuration file is necessary for each of the individual steps. There are many options that can be configured. Here we will only briefly touch a few basic concepts. An extensive documentation of the configuration file options is provided in Configuration Options.

In this tutorial we will use the directory and virtual environment that we created during the installation steps (see Installation and Requirements).

# Redefine the environment variable
export DNN_HOME=/data/user/${USER}/DNN_tutorial

# load icecube environment
eval $(/cvmfs/icecube.opensciencegrid.org/py3-v4.3.0/setup.sh)
/cvmfs/icecube.opensciencegrid.org/py3-v4.3.0/RHEL_7_x86_64/metaprojects/icetray/v1.12.0/env-shell.sh

Now that the icecube environment is loaded, we can activate our virtual environment.

# activate python virtual environment
source ${DNN_HOME}/py3-v4.3.0_tensorflow2.14/bin/activate

# set CUDA environment variables
export CUDA_HOME=/data/user/mhuennefeld/software/cuda/cuda-11.8
export PATH=$PATH:${CUDA_HOME}/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:${CUDA_HOME}/lib64

Note

It is important to first load the icecube environment and then activate the virtual environment. If done the other way around, the environment variables defined by the virtual environment will be overwritten by the icecube env-shell.sh. As a result, a different python version will be used than the one from the virtual environment.