Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
SYMPLER: SYMbolic ParticLE simulatoR
Fetching latest commit…
Cannot retrieve the latest commit at this time.
/* * This file is part of the SYMPLER package. * https://github.com/kauzlari/sympler * * Copyright 2002-2017, * David Kauzlaric <email@example.com>, * and others authors stated in the AUTHORS file in the top-level * source directory. * * SYMPLER is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation, either version 3 of the License, or * (at your option) any later version. * * SYMPLER is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with SYMPLER. If not, see <http://www.gnu.org/licenses/>. * * Please cite the research papers on SYMPLER in your own publications. * Check out the PUBLICATIONS file in the top-level source directory. * * You are very welcome to contribute extensions to the code. Please do * so by making a pull request on https://github.com/kauzlari/sympler * * Funding: * DFG (Deutsche Forschungsgemeinschaft) in the framework of * SFB499 TPA4 * Project KA 3482/2 * Projects LI 1831/1-1,2 * */ SYMPLER: SYMbolic ParticLE simulatoR ______________________________________________________________________ Contents of this file: - Field of application - Main features of SYMPLER - Compiling SYMPLER - Running SYMPLER - Postprocessing - Contributing to SYMPLER ______________________________________________________________________ Field of application: The need for a unified flexible and modular simulation tool allowing for the investigation of structural, thermodynamic, and dynamical properties of fluids and solids from microscopic over mesoscopic up to macroscopic time and length scales with suitable particle based simulation methods such as molecular dynamics, dissipative particle dynamics or smoothed particle hydrodynamics. The user should be enabled to define her/his own physical models without the need for recoding or code extensions. ______________________________________________________________________ Main features of SYMPLER: + Modular object oriented structure that is passed to the user level and allows to easily switch among different integration algorithms, particle interaction forces, boundary conditions, etc. + Arbitrary number of particle-species for the simulation of complex multi-component systems + Arbitrary number of additional user-defined degrees of freedom per particle-species + Symbolic definition of runtime compiled mathematical expressions for particle interactions + Import of CAD-geometries in the STL-format ______________________________________________________________________ Compiling SYMPLER: + Required tools for configuration and compilation: ++ the GNU autotools (aclocal, libtool, autoconf, automake, etc.) ++ a C++ compiler such as GNU g++ ++ a C compiler such as GNU gcc + Required libraries: ++ GNU scientific library (GSL) ++ libxml2 ++ CppUnit (provides unit tests; on Ubuntu: sudo apt install libcppunit-dev) NOTE: You usually also have to install the libraries for development. They have names such as libgsl*-dev, libsdl*-dev, and so on. + Optional libraries: ++ liblzma-dev: For the configure-option --enable-all-static (see below) ++ libsdl (1.2.0 or newer but not 2.0 or newer): If you use configure option --enable-meter-lives (see below). Gives simple runtime visualisation with the modules MeterLive and MeterLiveColoured. If the library is installed it will usually be found and the modules will be compiled. If not, you will have to modify some include and/or library paths by hand. ++ OpenMP: For partial but still quite inefficient shared memory parallelisation. Use the option -fopenmp during configuration (CXXFLAGS="... -fopenmp"). ++ libtnt, libjama, libsuperlu: Use the options --with-tnt --with-superlu during configuration. This allows for: +++ coupling of particles to a linear system of equations solver with module IntegratorLSE +++ the module for velocity constraints VelConstraints +++ the integrator IntegratorVelocityVerletPressure +++ running successfully the testsuite (see also section "Contributing to SYMPLER" below) + Configuration and Compilation: All the commands invoked in the following have their own documentation. Please refer to it for more information. It follows first a standard method working on standard Linux systems such as Ubuntu 12.04 or newer. Below you find special instructions for MacOS and Intel compilers. ++ Configuration and Compilation for standard Linux-systems: Assuming the complete path to the directory where you downloaded the SYMPLER source-code is given by $SOURCE-DIR, then the series of commands that usually works is: $ cd $SOURCE-DIR $ autoreconf --install --force Then, assuming the complete path to the directory where you want to compile SYMPLER is given by $BUILD-DIR (it might be the same as or different from $SOURCE-DIR): $ cd $BUILD-DIR $ $SOURCE-DIR/configure [options] CXXFLAGS="[cxxflags]" $ make -j[1 to roughly the number of available cores] Useful [cxxflags] are (for more see the g++ documentation): -O3: for optimised compilation with g++ -fopenmp: For compilation with OpenMP parallelisation (currently still inefficient) and useful [options] are: --with-tnt: use tnt library and jama libraries --with-superlu: use superlu --enable-meter-lives: Compile additional MeterLive modules for runtime visualisation. If you enable this option and get an error message such as "...SDL/SDL.h: No such file or directory", then you have not installed libsdl at all (see optional requirements above) or you must give the correct include and library paths in the CXXFLAGS or elsewhere. Note: It is NOT recommended to use this option on HPC machines. Anyway, most HPC admins will refrain from installing the required SDL-library in an HPC environment. --enable-all-static: for a statically linked stand alone executable not requiring any dynamical linking of libraries (except for the glibc used during linking after compilation of SYMPLER; and the runtime compiled expressions of course). NOTE the required library liblzma-dev (listed under optional libraries above). Also libicu should be installed, but this is usually the case by default (for example libicu-dev and libicu55 on Ubuntu 16.04). Further note that on older linux-distributions (for example Ubuntu around 12, 13, 14), this option causes SYMPLER to crash very often. On newer distributions this issue has vanished. Further bug reports on this issue are welcome. After successful compilation the executable "sympler" will be created in the directory src/. ++ Configuration and Compilation on MacOS The following instructions were last successfully tested with Mac OS X 10.11.6 (based on older instructions by Andreas Greiner, Tobias Wondratsch, Oliver Grünert). The libxml2 library comes already preinstalled by Apple but caused trouble during the SYMPLER installation. You may give the preinstalled libxml2 a try or immediately go and get an alternative version and the two remaining libraries by using one of the packaging-systems for Mac OS. One of the best packaging-systems is homebrew. How to install homebrew is described on https://brew.sh. Two nice properties of homebrew are, that it is easy to install and very easy to UNINSTALL, meaning that your operating system will be the same as before, without any old header or binary files lying around. After homebrew is installed, you are ready to install the GSL and the SDL library. But first, on a fresh system, you might still be lacking $ brew install gcc $ brew install libtool $ brew install autoconf $ brew install automake To install GSL type $ brew install gsl into your terminal. Then, to install SDL type $ brew install sdl into your terminal. If you have to, also install libxml2 $ brew install libxml2 Similarly for further libraries which you might need (see Linux section). Make sure the symbolic link used to find the libraries is set to your version installed by brew. Usually you use a command $ brew link ... for that. Try first to skip this paragraph because the next step is only necessary if you run into trouble with your already preinstalled command-line tool gm4 (a macro processor used by aclocal). Then you need to replace it with a newer version, since the old version which came from Apple was broken. This was the case for example in MacOS 10.6.2. You can grab the most recent version from http://ftp.gnu.org/gnu/m4/. Now drag the old gm4 tool in /usr/bin to a backup folder (for example your Desktop) and replace it with the newer version (you will be asked to enter your root password). Now you are ready to compile SYMPLER! So grab the SYMPLER source code into a code directory (called $CODE-DIR here) and create a new build directory (called $BUILD-DIR here) into which you will compile. Then in your terminal (adapt the paths to your needs): $ cd $CODE-DIR $ autoreconf --install --force $ cd $BUILD-DIR $ $CODE-DIR/configure CXXFLAGS="-O3" $ make -j[1 to roughly the number of available cores] After successful compilation the executable "sympler" will be created in the directory src/ For further compile options see the Linux section. If you get a compile error related to cppunit, then try to switch in the CXXFLAGS to a different C++ standard than the default one of your compiler. For example $ $CODE-DIR/configure CXXFLAGS="-O3 -std=c++14" A known problem that may occur is a conflict with a libxml2 version previously installed in MacOS, for example due to a previous installation of Xcode. Just find a way to clean your system from this old libxml2 version and the SYMPLER installation should work fine by using the brew version of libxml2. Or try the command $ brew link ... mentioned above. ++ SYMPLER and Intel compilers: A few hints when using icc (the Intel C++ compiler) for SYMPLER: ++ Set the environment variable LANG=C ++ If using autoparallelization, set LD_LIBRARY_PATH to the lib directory of icc ++ -ipo gave problems with linking Otherwise, compilation worked fine. A good configure invocation might look like (adapt the paths to your needs): export LANG=C; export LD_LIBRARY_PATH=/simstorage/software/intel/cpp11/lib/intel64; ../sympler/configure CXX=/simstorage/software/intel/cpp11/bin/intel64/icc \ CC=/simstorage/software/intel/cpp11/bin/intel64/icc LDFLAGS='-lstdc++ -lz' \ CXXFLAGS="-O3 -ip -parallel -funroll-loops -complex-limited-range \ -IPF-fma -IPF-fltacc -IPF-fp-relaxed -fp-model=fast -no-debug -xSSE4.1 \ -mcmodel=medium -falign-functions -opt-calloc -unroll-aggressive" ______________________________________________________________________ Running SYMPLER A static executable will work, if your glibc is roughly the same as the one, which was used for compilation. Additionally executables are specific to 32 and 64 bit platforms, respectively. The modules MeterLive and MeterLiveColoured are not activated in a static version. Alternatively you can use a dynamically linked version. A dynamic executable needs at least the following libraries installed: gsl (GNU scientific library) libxml2 (for parsing the input files) + Starting the executables Assuming you have an executable with name sympler in the directory $BUILD-DIR, just type $ $BUILD-DIR/sympler and you will be told the further options you may try in order to really get s.th. useful out of the code. For example $ $BUILD-DIR/sympler --help gives a general help text. $ $BUILD-DIR/sympler --help all lists the help text for all available modules $ $BUILD-DIR/sympler --help [UNKNOWN] gives you a list of all groups of modules you may request help for if [UNKNOWN] is an argument that sympler does not understand. For example you may get help for the Integrators with $ $BUILD-DIR/sympler --help Integrators To really start a simulation type $ $BUILD-DIR/sympler [INPUT-FILE] where [INPUT-FILE] is an input file in xml-format. + Input files The code has a modular structure. This is reflected in the input files by using the XML-format (for further information on XML see: http://www.w3schools.com/xml/default.asp). A quick glance at the input files and the online-help text of the executable should be enough to understand how it works. All modules and their attributes are described in the online-help. A bit of definition: The XML-expression <ModuleName attributeName = "value" /> shows you the difference between a module name, an attribute name and the value assigned to the latter. Compare this with the example given below. If you play around with parameters, be aware of possible simulation instabilities due to a timestep being to large for your chosen parameters. In this case you will obtain some message that a particle flew to far. The timestep can be modified in the module <Controller> with the attribute dt. The number of timesteps can be changed with the attribute timesteps. All quantities have to be given in units of your choice for instance in such a way that the Boltzmann constant may be set to 1. All simulations are 3D. Quasi-2D-simulations can be performed by applying periodic boundary conditions. Real 2D simulations can be performed by enforcing interactions to happen only in 2D. Here is a small example. The newest version of SYMPLER might complain because some modules or attributes are out-dated, but the general idea should become clear: <?xml version="1.0"?> <!DOCTYPE Simulation SYSTEM "/home/lieneman/Projekte/NanoSAVE/Moldyn/moldyn.dtd"> <Simulation simName="MDPD simulation" inputFromResults="yes"> <Square cutoff="1" name="L1" /> <Square cutoff="0.75" name="L075" /> <Controller timesteps="100000" statusEvery="10" dt="0.001"> <IntegratorVelocityVerlet species="H" /> </Controller> <ValCalculatorRho symbol="n" weightingFunction="L1" species1="H" species2="H" selfContribution="false"/> <!-- Fluid-fluid, attractive --> <FPairVels species1="H" species2="H" pairFactor="-1/4.7746*[rij]/rij*750" weightingFunction="L1" /> <!-- Fluid-fluid, repulsive --> <FPairVels species1="H" species2="H" pairFactor="3.14159265*0.75^4/15*(ni+nj)*[rij]/rij*40" weightingFunction="L075" /> <ThermostatPetersIso temperature = "15" dissipation = "5" species1 = "H" species2 = "H" weightingFunction = "L1" /> <Phase> <LinkedListCreator/> <BoundaryCuboid boxX="32" boxY="32" boxZ="32" periodicX="yes" periodicY="yes" periodicZ="no"> <ReflectorThermalize temperature="15" oneHit="true"/> <ParticleCreatorStatic density="20" species="H" corner1 = "(10,10,18)" corner2 = "(22,22,6)" temperature = "5" ellipsoid = "true" /> </BoundaryCuboid> </Phase> <MeterPosVel measureEvery="1000" species="H"> <OutputVTK fileName="results/posvel_prepare.vtk" format="ascii"></OutputVTK> </MeterPosVel> <GridAveragerStructured avgOver="50" measureEvery="1000" nX="16" nY="16" nZ="16"> <Pressure species="H"/> <OutputFile fileName="results/gas_pressure.dat" multipleFiles="yes" columns = "H_pressure_mean" /> </GridAveragerStructured> </Simulation> Besides reading the help text, another way to quickly find out which attributes are allowed for a specific module, is to type some non-sense which most likely does not correspond to any attribute. + Runtime compiled expressions In the example file you see some of them. For example "...(ni+nj)*[rij]/rij*40 where 'n' is a user defined scalar symbol, [rij] is the distance vector [ri]-[rj] for particle pair (i,j) and 'rij' is its absolute value. Single indices 'i', 'j' denote a particle property while double indices 'ij' denote the property of the pair of particles. Type $ $BUILD-DIR/sympler --help expressions for a complete overview of possible elements of runtime compiled expressions. Help for the two module-groups Forces and Symbols that use runtime-compiled expressions the most you can get with $ $BUILD-DIR/sympler --help Forces $ $BUILD-DIR/sympler --help Symbols Further check the PUBLICATIONS file for publications explaining further details. ______________________________________________________________________ Postprocessing SYMPLER produces output in various ASCII and binary formats, for example VTK, simple column data, or binary arrays of data. First of all you need Meters in your input file. Check $ $BUILD-DIR/sympler --help Meters In the above example we have two Meters, namely MeterPosVel and GridAveragerStructured. The thing is that a Meter itself usually does not create any output. It just collects data. GridAveragerStructured is even lazier and lets GridMeters do the work. Check $ $BUILD-DIR/sympler --help GridMeters GridMeters are also Meters. Hence they also collect data and nothing else. For output you have to attach PostProcessors to the Meters. In the example above you can find two of them: OutputVTK and OutputFile. Check $ $BUILD-DIR/sympler --help Postprocessors ______________________________________________________________________ Contributing to SYMPLER Feel free to contribute code-extensions by making a pull request on https://github.com/kauzlari/sympler You should follow these rules before making a pull request: - At the very least, run the test-script in the testsuite directory. - Think about adding an own test of your new features to the testsuite