This repository hosts Fortran90 programs tailored for serial and parallel molecular dynamics simulations.
Data analysis of the generated data is conducted utilizing gnuplot.
In order to run the program, the user must download the prerequisites listed below.
- gfortran: The GNU Fortran compiler is required to compile and execute Fortran 90 code.
-
Install gfortran in Ubuntu with:
sudo apt-get install gfortran
-
- gnuplot: Gnuplot is a command-line driven plotting tool. It's used for generating plots.
-
Install gnuplot in Ubuntu with:
sudo apt-get install gnuplot
-
Important
Parallel code uses MPI. In this project we used the version available at the CERQT2 cluster
To run the program, type:
make main
make run
Compiles the modules and generates the main program. Then executes the main program.
make plot
Runs plots from data in datafile and performs some folder organization.
Tip
For more commands type make help
Main.f90 contains the main simulation loop. It requires the use of the modules detailed in the subsequent section.
Includes the system initialization, thermalization, and production run processes.
Data files 'thermodynamics.dat' and 'resultsrdflong_def.dat' will be generated containing the simulation results.
The modules required to compile the main program are:
- forces.f90: Allows for the calculation of forces acting on the particles.
- initialize.f90: Contains subroutines for system initialization and applying periodic boundary conditions.
- integrate.f90: Contains subroutines for the Velocity Verlet integrator, Andersen thermostat, and Gaussian distribution.
- thermodynamics.f90: Contains subroutines for calculating different observables of the simulation.
Additionally, the following files are used for proper functioning:
- input.txt: Includes input simulation parameters such as the number of particles, timestep, cutoff, temperature, among others.
- data2plots.gnu: Generates figures from the data generated in the simulation.
- data2plots_adSI.gnu: Generates figures in SI units.
- binning_gestor.f90: Performs block average method and calculates the mean and standard deviation of observables.
It also incorporates some VMD plots for a set of precalculated results.
It contains the same functionalities as in serie, but the code is adapted for parallelization. In addition, it also calculates paralelization scalability.
To run the parallel program, sending jobs to the cluster, type:
make cluster_compile
make cluster_ALLRUN
Compile the modules and generates the main program. Then executes the main program.
make plot
make paralel_plot
Runs plots from data in datafile and moves them to a new folder. Organizes directory in folders.
It also supports running jobs in interactive mode:
make enter_interactive
cd [working directory]
module load openmpi/4.1.4_ics-2021.3
make runX
Where X indicates the number of desired processors. 1, 2, 4, 8, 16, 32 and 40 processors are supported through the Makefile.
Tip
For more commands type make help
The generated plots should look like:
| Serie | Parallel |
|---|---|
 |
 |
 |
 |
 |
 |
Scalability of parallelization
In this simulation, the system was modeled at a temperature T = 300K, mirroring the conditions typical of Kr gas.
The interaction between particles was defined using the Lennard-Jones potential, with parameters ε/kB = 162.58 K and σ = 3.6274 angstrom.
The parameters were extracted from Rutkai et al. (Rutkai G.; Thol M.; Span R.; Vrabec J. How Well Does the Lennard-Jones Potential Represent the Thermodynamic Properties of Noble Gases?. Mol. Phys. 2017, 115, 1104–1121.)
Pau Franquesa (Initial conditions and pbc)
Pau Eritja (Forces)
Safae el Jelloui (Statistics and visualization of results)
Marcos Carrera (Integration)
Berta Bori (Coordination and integration of the code)