Troubleshooting¶
RuNNer 2.0 Issues¶
Library not found¶
Error: ERROR: No lapack library found
Solution: Install BLAS/LAPACK libraries:
# On Ubuntu/Debian
sudo apt-get install liblapack-dev libopenblas-dev
# On RHEL/Rocky/CentOS
sudo dnf install lapack-devel openblas-devel
Or point to an MKL installation:
export MKLROOT=/opt/intel/mkl
./configure
Static linking issues¶
Error: WARNING: SWITCHING TO A NON-STATIC BUILD
Solution: This is usually fine. Static builds require all libraries in static form. To explicitly force dynamic linking:
./configure STATIC=OFF
FFTW3 not found¶
Error: WARNING: SWITCHING TO A NON-FFTW3 BUILD
Solution: Install FFTW3 or specify the path:
# Install FFTW3
sudo apt-get install libfftw3-dev # Ubuntu/Debian
sudo dnf install fftw-devel # RHEL/Rocky
# Or specify custom path
./configure FFTW_ROOT=/path/to/fftw3
Note: FFTW is only required for using PW or PPPM solvers. For 3G and 4G for small systems the Ewald summation is faster. Thus, you only need FFTW if you are interessted in large systems and are using 3G or 4G potentials.
MPI not detected¶
Error: WARNING: SWITCHING TO NON-MPI BUILD
Solution: Use an MPI compiler wrapper:
./configure FC=mpiifort # Intel
./configure FC=mpif90 # GNU
Reconfiguration¶
If you need to change settings after an initial configure:
# Reconfigure with new options
./configure FC=gfortran MKL=OFF
# Clean build directory
make clean
# Rebuild
make -j
RuNNerASE Issues¶
Import error: Module not found¶
Error: ModuleNotFoundError: No module named 'runnerase_prediction'
Solutions:
- Ensure virtual environment is activated:
source .venv/bin/activate - Verify installation:
pip list | grep runnerase - Reinstall:
uv pip install -e .(with UV)
Dependency resolution issues¶
Problem: Packages can't resolve local dependencies
Solutions:
- Use the UV method (handles dependencies automatically)
- If using pip, install in correct order: runner-interface → core → prediction → training → active-learning
- Clear pip cache:
pip cache purge
libRuNNer.so not found¶
Error: OSError: libRuNNer.so: cannot open shared object file
Solution: Add the RuNNer build directory to your library paths:
export LIBRARY_PATH=/path/to/runner2:$LIBRARY_PATH
export LD_LIBRARY_PATH=/path/to/runner2:$LD_LIBRARY_PATH
# Make permanent by adding to ~/.bashrc:
echo 'export LD_LIBRARY_PATH=/path/to/runner2:$LD_LIBRARY_PATH' >> ~/.bashrc
Python version incompatibility¶
Error: ERROR: Python version >= 3.10 required
Solutions:
- Check Python version:
python --version - Use Python 3.10+, preferably 3.12
- Create environment with specific version:
uv venv --python 3.12
Getting Help¶
RuNNer 2.0 (Fortran)¶
- Bug Reports: https://gitlab.com/runner-suite/runner2
- Documentation: https://runner-suite.gitlab.io/runner2
RuNNerASE (Python)¶
- Bug Reports: https://gitlab.com/runner-suite/runnerase-2.0/issues
- Documentation: https://runner-suite.gitlab.io/runnerase-2.0