Quick installation and execution
Here, we provide a brief description of the steps to download, compile and run a job with CosmoLattice. As an example, we load a model called lphi4, with input parameters specified in the file lphi4.in (details of this model and of its parameters can be found in Section My first model of (singlet) scalar fields, here we just use these files as a demonstration example).
An appropriate version of FFTW needs to be installed in order to handle the code's Fourier transforms, either for serial jobs run on a single processor, or for parallelized jobs run on multiple processors. For serial jobs, that is all that is needed. For parallelized jobs, in addition, the appropriate MPI compiler needs also to be installed. Optionally, if one wants to print two- or three-dimensional snapshots of a simulation, or simply to stop/restart a run at a certain time, the HDF5 library also needs to be installed. Also optionally, the simulations can be parallelized in more than one spatial dimension, to speed up jobs of long duration or to run massive simulations. This is handled by the ParaFaFT library (parallel Fast Fourier transforms), which is automatically built and enabled when the code is compiled with MPI. For further details on the installation and use of the code and the libraries, please see Installation.
Download: You can download CosmoLattice from http://www.cosmolattice.net, or alternatively use git clone as indicated below.
git clone https://github.com/cosmolattice/cosmolattice
cd cosmolattice # Enter into main code folder
mkdir build # Create a new directory
cd build # and go inside it.
cmake -DMODEL=lphi4 ../ # Selects model phi^4
# OR
cmake -DMODEL=lphi4 -DMPI=ON ../ # Selects model phi^4 and activates distributed parallelization
make cosmolattice # Compiling
./lphi4 input=../models/parameter-files/lphi4.in # Executes a run (input parameter file 'lphi4.in')
# OR
mpirun -n 8 lphi4 input=../models/parameter-files/lphi4.in # Parallelized run on 8 MPI ranks (input file 'lphi4.in')
Assuming it uses Environment Modules, otherwise one needs to see how to load the required libraries.
git clone https://github.com/cosmolattice/cosmolattice
cd cosmolattice # Enter into main code folder
mkdir build # Create a new directory
cd build # and go inside it.
module load CMAKE # Here we will call C++, FFTW3, CMAKE and MPI.
module load C++ # The order is important, C++ before MPI before FFTW3
module load MPI # Needed to run parallelized simulations
module load FFTW3 # Needed for Fourier transforms in parallelized simulations
cmake -DMODEL=lphi4 ../ # Selects model phi^4 for serial runs
# OR
cmake -DMODEL=lphi4 -DMPI=ON ../ # Selects model phi^4 and activates distributed parallelization
make cosmolattice # Compiling
# Now you can run lphi4. How you do it depends on the cluster.
module spider <library name> in the command line. You can also check the cluster's documentation for more information.
Optional: Install HDF5 and simply add the flag -DHDF5=ON when you call CMake:
cmake -DMODEL=lphi4 -DHDF5=ON ../ # For serial use
cmake -DMODEL=lphi4 -DMPI=ON -DHDF5=ON ../ # For parallel use (also works with ParaFaFT)
Installing fftw3 and HDF5: These libraries can be installed with your system's package manager, or built automatically by CosmoLattice by adding the corresponding flag when you call CMake:
cmake -DMODEL=lphi4 -DAUTOBUILD_FFTW=ON ../ # Build fftw3 automatically
cmake -DMODEL=lphi4 -DHDF5=ON -DAUTOBUILD_HDF5=ON ../ # Build HDF5 automatically
Note
Some of the above commands are only indicative, as they can change from machine to machine. For more detailed explanations about the installation of CosmoLattice and the libraries it uses (or can use), see Installation.