Portable conda build

Prerequisites

Install Anaconda 3 if not already installed on your computer, download and follow the instructions at https://www.anaconda.com/

If you want to do spatial RDME simulations which require GPUs, install CUDA, download and instructions can be found at https://developer.nvidia.com/cuda-downloads

Add the following line at the end of /.bashrc so LM can find CUDA. If path to CUDA is different,

export path to where CUDA is installed:
export PATH="/usr/local/cuda/bin/:$PATH"

LM installation via conda

The conda environment can be created from the yml file included here. Go into the file and change "lmpretest" to whatever you want your environment name to be, then run:

replace envname with your choice of name for the environment.

    conda env create -n envname -f lm_precomp.yml
    
    conda activate envname

NOT RECOMMENDED(deprecated): You can also make your Anaconda environment with your choice of name "envname" using Python 3 through the following procedure:

    conda create -n envname -c conda-forge python=3.7.3
    conda activate envname
    conda install -c conda-forge -y git==2.33.1 gcc_linux-64==7.3.0 gxx_linux-64==7.3.0 binutils_linux-64==2.31.1 libstdcxx-ng==11.2.0 libgcc-ng==11.2.0
    conda install -c anaconda hdf5==1.10.4 h5py==2.10.0 protobuf==3.13.0.1
    conda install -c conda-forge -y numpy==1.19.2 cython==0.29.4 cmake==3.14.0 xlrd==1.2.0 swig==4.0.2 jupyter==1.0.0 matplotlib==3.4.3 ipywidgets==7.6.5 tqdm==4.62.3 pillow==8.3.2 jinja2==3.0.2 scipy==1.5.2 pybind11==2.8.1 pandas==1.3.4 pytables==3.5.2 biopython==1.79

Once you have created and activated your environment, go to a directory where you want to build Lattice Microbes and do the build running the following commands: (You do not need to run simulations where the build is located, lm will be installed in your environment as an executable)

cd .. (make sure you are in the directory which has Lattice-Microbes folder)

create a build directory:

mkdir build
cd build
cmake ../Lattice-Microbes/src/ -D MPD_GLOBAL_T_MATRIX=True -D MPD_GLOBAL_R_MATRIX=True
# The -D flags can be removed if you build with smaller transition matrix and reaction matrix size limits in the cmake options.
make && make install
# Run make install as root or sudo if possible

manually set CUDA_ARCHITECTURES if needed:

cmake ../Lattice-Microbes/src/ -D MPD_GLOBAL_T_MATRIX=True -D MPD_GLOBAL_R_MATRIX=True -D CUDA_ARCHITECTURES="60;70;75;80;86"
make && make install

Build for docker container

It is important to make sure the CUDA_ARCHITECTURES argument matches the architecture of the GPU you are using. If you are not sure what architecture your GPU is, you can run:

nvidia-smi

And check compute capability of your GPU. You can find the information in the last two digits of the GPU model. For example, RTX 3060 Ti has compute capability 86.

docker build --build-arg CUDA_ARCHITECTURES="70;75;80;86" -t lm2.5dev:tag .

Build for apptainer

For x86_64 architecture:

apptainer build your-container.sif apptainer.def

For aarch64 architecture:

apptainer build your-container.sif apptainer-dtai.def

at run time, specify the architecture:

   CUDA_ARCHITECTURES="75;80;86" apptainer run your-container.sif

Testing the installation

Check to make sure the installation worked, run:

lm --help
# Should return help commands for lm

There are example executable simulation files in example_rdme_files you can try running.

cd example_rdme_files
python run_simulation.py