\chapter{Introduction} \section{Design philosophy} This software is a general purpose classical simulation package. It has been developed at Northwestern University (Evanston, USA; group of Prof.\ Randall Q.\ Snurr) during 2006-2009 in active collaboration with University Pablo de Olavide (Seville, Spain; group of Prof.\ Sofia Calero), and from 2010-2015 also at the University of Amsterdam (David Dubbeldam) and Technical University of Delft (group of Prof.\ T.J.H.\ Vlugt). It can be used for the simulation of molecules in gases, fluids, zeolites, aluminosilicates, metal-organic frameworks, and carbon nanotubes. Programs can be written in various ways, but often it is true that the fastest codes are probably the hardest to read, while programs strictly based on readability lacks efficiency. RASPA is based on the following ideas: \begin{itemize} \item{Correctness and accuracy}\\ For all the techniques and algorithms available in RASPA we have implemented the 'best' ones available in literature. For example, RASPA uses Configurational-Bias Monte-Carlo, it uses the Ewald summation for electrostatics, molecular dynamics is based on 'symplectic' integrators, all Monte-Carlo moves obey detailed balance etc. \item{Functional design}\\ Looking at the source, you will notice that there are not a lot of files. The program is split up according to its function: 'grid.c' contains the code to make and use a grid of a framework, 'ewald.c' handles all the electrostatic,'mc\_moves.c' contains all the moves to be used in Monte-Carlo,'potentials.c' contains all the VDW potentials etc. \item{Input made easy}\\ The requirements for the input files is kept as minimal as possible. Only for more advanced options extra commands in the input file are needed. Also the format of the input is straightforward. Default settings are usually the best ones. Fugacity coefficients and excess adsorption are automatically computed. \item{Integrated simulation environment}\\ The code is built up of many functions and routines which can be easily combined to do what you want. Molecular dynamics can be used in Monte Carlo and visa versa. Extension and modification of the code is relatively straightforward. \end{itemize} RASPA used three 'types' or 'groups' for the particles: 1) Framework atoms, 2) Adsorbates, and 3) Cations. The advantage is that all the energies are split and the interactions can be examined (also the energies are split in the Ewald Fourier part). Another example is when using thermostats in e.g. LTA5A where a different thermostat operates in the framework atoms, the adsorbates, and the cations. These all move at different length- and time scales. Note that it is not possible to exchange types during Identity-change moves (if defined they are ignored). \section{Units and conventions} \begin{itemize} \item{The standard units in RASPA from which all other units are derived are:}\\ \vskip 0.1cm \begin{tabularx}{\linewidth}{l|l|l|l} quantity & symbol & unit & value\\ \hline length & $l$ & \AA ngstrom & $10^{-10}$ m\\ temperature & $T$ & Kelvin & K\\ mass & $m$ & atomic mass & $1.6605402\times 10^{-27}$ kg\\ time & $t$ & pico seconds & $10^{-12}$ s\\ charge & $q$ & atomic charge & $1.60217733\times 10^{-19}$ C/particle\\ \hline \end{tabularx} \vskip 0.1cm \noindent Some examples of derived units:\\ \begin{tabularx}{\linewidth}{l|l|l|l} quantity & symbol & units & conversion value\\ \hline energy & $U$ & $J=\text{mass}\times\text{length}^2/\text{time}^2$ & $1.66054\times10^{-23}$ (=10 J/mol)\\ pressure & $p$ & $\text{Pa}=\text{mass}/(\text{length}\times\text{time}^2)$ & $1.66054\times10^7$\\ diffusion constant & $D$ & $D=\text{length}^2/\text{time}$ & $1\times10^{-8}$\\ force & $f$ & $f=\text{length}/\text{time}^2$ & $1.66054\times 10^{-13}$\\ \dots & \dots & \dots & \dots \\ \hline \end{tabularx}\\ A pressure input of 10 Pascal in the input file, is converted to 'internal units' by dividing by $1.66054\times10^7$. In the output any internal pressure is printed, multiplied by $1.66054\times10^7$. It is not necessary to convert units besides input and output, with a few exceptions. One of them is the Coulombic conversion factor \begin{equation} \frac{q_i q_j}{4\pi \epsilon_0}=\frac{\text{charge}^2}{4 \pi \times \text{electric constant}\times\text{length}\times\text{energy}}=138935.4834964017 \end{equation} with the electric constant as $8.8541878176\times10^{-12}$ in units of $\text{C}^2/(\text{N}.\text{m}^2)$. This factor is needed to convert the electrostatic energy to the internal units at every evaluation. \noindent The Boltzmann's constant $k_B$ is \begin{equation} k_B=\text{Boltzmann constant}/\text{energy}=0.8314464919 \end{equation} with the Boltzmann constant as $1.380650324\times10^{-23}$ in units of J/K, and $k_B=0.8314464919$ in internal units. \item{Numbering is based on the C-convention, i.e. starting from zero.} \item{Files in the current directory always have preference.}\\ Sometimes one would like to try various parameters for force field fitting for example. In order to avoid making a lot of directories for each force field it is more convenient to have the 'pseudo\_atoms.def', 'force\_field\_mixing\_rule.def' and 'force\_field.def' files in the \emph{current} directory. \end{itemize} \section{Compiling and installing RASPA} \subsection{Requirements} RASPA needs a C compiler, like 'gcc' or intel's 'icc' compilers, and optionally the libraries 'fftw', 'blas', and 'lapack'. \subsection{RASPA from 'git'} Working with 'git' and a remote repository means that you will have to distinguish between two locations of the code: \begin{enumerate} \item{The repository (visible to everyone)} \item{your local copy (only visible to you)} \end{enumerate} To check-out the code for the first time do: \begin{verbatim} git clone https://github.com/iraspa/RASPA2 \end{verbatim} After that, you can update the code by using \begin{verbatim} git pull \end{verbatim} \subsection{installing RASPA} The \emph{RASPA\_DIR} environment variable should be set to where you would like to install RASPA. A common way of defining it is using the bash-shell \begin{quote} export RASPA\_DIR=\$\{HOME\}/RASPA/simulations/ \end{quote} or \begin{quote} setenv RASPA\_DIR "\$\{HOME\}/RASPA/simulations/" \end{quote} for 'csh' and 'tcsh' shells. It is possible to add this line to ".bashrc", "/etc/bashrc", "/etc/profile" etc, depending on the unix-version and shell version to automatically have the environment variable set at login. Note that the source-code of RASPA is kept separate from the installation data. RASPA needs the environment variable to locate various files it needs, e.g. molecule definitions, framework definitions, force and field definitions. It looks for these files relative to the RASPA\_DIR directory. Before installing RASPA with \begin{verbatim} make install \end{verbatim} from the top-directory, the code needs to be compiled. \subsection{compiling RASPA} RASPA uses the standard 'configure' utilities (autoconf, automake, libtool, and make). The steps to install from scratch, i.e. after a 'make distclean' or 'git clone' are \begin{enumerate} \item{\verb=rm -rf autom4te.cache=} \item{\verb=mkdir m4=} \item{\verb=aclocal=} \item{\verb=autoreconf -i=} \item{\verb=automake --add-missing=} \item{\verb=autoconf=} \item{\verb,./configure --prefix=${RASPA_DIR}, \qquad or\\ \verb=./scripts/CompileScript/make-gcc-local=} \item{\verb=make=} \end{enumerate} where '\$\{RASPA\_DIR\}' is the directory you would like to install RASPA, and the commands are executed in the top directory. Usually (when recent automake and autoconf versions are installed), it is enough to do \begin{enumerate} \item{\verb=make clean=} \item{\verb,./configure --prefix=${RASPA_DIR},} \item{\verb=make=} \end{enumerate} You can use the 'CFLAGS' environment variable to set compiler options and 'CC' to set the compiler. For example, for a gcc compiler one could use \begin{verbatim} export CFLAGS="-Wall -O3 -ffast-math" export CC="gcc" \end{verbatim} \subsection{Running RASPA\label{Introduction: running RASPA}} Running RASPA is based on two files: \begin{itemize} \item{A 'run' file to execute the program}\\ an example file is: \begin{verbatim} #! /bin/sh -f export RASPA_DIR=${HOME}/Research/simulations/ $RASPA_DIR/bin/simulate \end{verbatim} This type of file is know as a 'shell script'. RASPA needs the variable 'RASPA\_DIR' to be set in order to look up the molecules, frameworks, etc. The scripts sets the variable and runs RASPA. RASPA can then be run from any directory you would like. \item{An 'input'-file describing the type of simulation and the parameters}\\ In the same directory as the 'run'-file, there needs to be a file called 'simulation.input'. An example file is: \begin{verbatim} SimulationType MonteCarlo NumberOfCycles 100000 NumberOfInitializationCycles 10000 PrintEvery 1000 Box 0 BoxLengths 30 30 30 ExternalTemperature 300.0 component 0 methane TranslationProbability 1.0 CreateNumberOfMolecules 100 \end{verbatim} This tells RASPA to run a Monte-Carlo simulation of 100 methane molecules in a $30\times30\times30$ \AA\ cubic box (with 90$^\circ$ angles) at 300 Kelvin. It will start with 10000 cycles to equilibrate the system and will use 100000 cycle to obtain thermodynamic properties of interest. Every 1000 cycles a status-report is printed to the output. The Monte-Carlo program will use only the 'translation move' where a particle is given a random translation and the move is accepted or rejected based on the energy difference. \end{itemize} In order to run it on a cluster using a queuing system one needs an additional file 'bsub.job' (arbitrary name) \begin{itemize} \item{'gridengine'} \begin{verbatim} #!/bin/csh # Serial sample script for Grid Engine # Replace items enclosed by {} #$ -S /bin/csh #$ -N Test #$ -V #$ -cwd echo $PBS_JOBID > jobid setenv RASPA_DIR ${HOME}/RASPA/simulations/ $RASPA_DIR/bin/simulate \end{verbatim} The job can be submitted using 'qsub bsub.job'. \item{'torque'} \begin{verbatim} #!/bin/bash #PBS -N Test #PBS -o pbs.out #PBS -e pbs.err #PBS -r n #PBS -V #PBS -mba cd $PBS_O_WORKDIR echo $PBS_JOBID > jobid export RASPA_DIR=${HOME}/RASPA/simulations ${RASPA_DIR}/bin/simulate \end{verbatim} The job can be submitted using 'qsub bsub.job'. \item{'slurm'} \begin{verbatim} #!/bin/bash #SBATCH -N 1 #SBATCH --job-name=Test #SBATCH --export=ALL echo $SLURM_JOBID > jobid valhost=$SLURM_JOB_NODELIST echo $valhost > hostname module load slurm ${RASPA_DIR}/bin/simulate \end{verbatim} The job can be submitted using 'sbatch bsub.job'. \end{itemize} \section{Output from RASPA} RASPA generates output from the simulation. Some data is just information on the status, while other data are written because you specifically asked the program to compute it for you. The output is written to be used with other programs like: \begin{itemize} \item{gnuplot} \item{VTK} \item{iRASPA} \item{VMD} \end{itemize} The main output is written to the directory 'Output/System\_0/', 'Output/System\_1/', \dots for each of the simulated systems. Usually one simulates only a single system. However, the Gibbs ensemble requires 2 systems, one for vapor phase and one for the liquid phase, while $n$ systems are used by the (hyper-) parallel-tempering technique(s). \section{Citing RASPA} If you are using RASPA and would like to cite it in your journal articles or book-chapters, then for RASPA: \begin{quote} D. Dubbeldam, S. Calero, D.E. Ellis, and R.Q. Snurr, \newblock RASPA: Molecular Simulation Software for Adsorption and Diffusion in Flexible Nanoporous Materials, \newblock {\em Mol. Simulat.}, \url{http://dx.doi.org/10.1080/08927022.2015.1010082}, 2015. \end{quote} For the inner workings of Monte Carlo codes: \begin{quote} D. Dubbeldam, A. Torres-Knoop, and K.S. Walton, \newblock On the Inner Workings of Monte Carlo Codes, \newblock \url{http://dx.doi.org/10.1080/08927022.2013.819102} \newblock {\em Mol. Simulat.}, 39(14-15), 1253-1292, 2013. \end{quote} For the description of Molecular Dynamics and diffusion: \begin{quote} D. Dubbeldam and R.Q. Snurr, \newblock Recent Developments in the Molecular Modeling of Diffusion in Nanoporous Materials, \newblock \url{http://dx.doi.org/10.1080/08927020601156418}, \newblock {\em Mol. Simulat.}, 33(4-5), 305-325, 2007. \end{quote} For the description of the implementation of force fields: \begin{quote} D. Dubbeldam, K.S. Walton, T.J.H. Vlugt, and S. Calero, \newblock Design, Parameterization, and Implementation of Atomic Force Fields for Adsorption in Nanoporous Materials, \newblock \url{https://doi.org/10.1002/adts.201900135}, \newblock {\em Adv. Theory Simulat.}, 2(11), 1900135, 2019. \end{quote}