Installing the software

Instructions for running code

  1. Clone the repo using

    git clone --recursive https://github.com/stochasticHydroTools/SlenderBody

    This will also clone the submodules in the dependency folder.

  2. Run

    cd SlenderBody
bash install.sh

    in the main directory. This will compile all of the dependencies and C++ modulues. It will also add the appropriate directories (where the python modulues are compiled) to your PYTHONPATH.

    If the compilation process fails this script will show an error. In that case fix the compilation issue and run

    bash install.sh

    again.

  3. Common mistakes:

    • UAMMD, which we use for hydrodynamic interactions, requires a GPU and a system with a working CUDA environment. You can still run the software without a GPU, but you cannot compute hydrodynamics between fibers.

    • When compiling UAMMD in the dependencies, the compiler may complain depending on what version of nvcc you use. If it complains that there is no file called cub, cd into SlenderBody/Python/Dependencies/UAMMD_PSE_Python/uammd/src/third_party and change the folder called “cub_bak” to “cub.” This should fix that issue. See also the UAMMD webpage for more compilation instructions.

    • Sometimes there is an issue with linking the fortran library (Heap) to C++ and then Python. If the commands in the makefile (in cppmodules) produce an error, this sequence typically works

      gfortran -shared -O3 ../../Fortran/MinHeapModule.f90 -o /($SLENDERROOT)/Fortran/libFortranHeap.so -fPIC
icpc -O3 -Wall -shared EndedCrossLinkers.cpp -o EndedCrossLinkedNetwork.so     -L/($SLENDERROOT)/Fortran/ -lFortranHeap -std=c++11 -fPIC -fopenmp -llapack -lblas -llapacke `python3 -m pybind11 --includes`

      where (SLENDERROOT) is the location of installation. Make sure that ($SLENDERROOT)/Fortran is also added to your LD_LIBRARY_PATH.

    • The file VectorMethods.cpp is set up to use LAPACKE and CBLAS. If you compile using Intel’s MKL, then simply comment lines 3 and 4 and uncomment line 18.

  4. Run the python scripts in Python/Examples. For example,

    python3 ThreeShearedFibs.py

    will run ThreeShearedFibs.py (see Examples page for documentation).

Parallelization

The number of OpenMP threads (environment variable) MUST be set to one to obtain good performance. In particular, you must use

export OMP_NUM_THREADS=1

in linux prior to running our code. The parallelization is then implemented in the C++ class FiberCollectionC.cpp which takes as an inpute the number of threads. The number of threads in these calculations can be set by passing an integer to the constructor of fiberCollection.py.

Uninstalling

  1. Run

    make clean

    inside the Python folder.

  2. The install.sh script modifies the .bashrc file (which are marked as written by the script). Simply remove these lines.