yinggwan / gauss

Gaggle of Algorithms and Utilities for Simulating Stuff

OSX (el capitan)/Ubuntu (14.04): Windows (Visual Studio 2017):

External Dependencies

1. Qt >= 5.8 (https://www.qt.io, version 5.9.0 only for Windows build)
2. OPTIONAL: Pardiso Solver (http://www.pardiso-project.org)
3. OPTIONAL: Gurobi Solver (http://www.gurobi.com)

Included Submodules (Installed Automatically)

1. Libigl https://github.com/libigl/libigl
2. Eigen >= 3.2 (By Default GAUSS uses the libigl Eigen install)
3. Spectra Eigenproblem Solver (https://spectralib.org)
4. The Flexible Collision Library (https://github.com/flexible-collision-library/fcl)
5. Eigen-Gurobi Interface (https://github.com/jrl-umi3218/eigen-gurobi/)

New Sample Application

Use this sample application to get a fast start on developping applications using GAUSS. It provides a working (I think ?) CMake file and includes GAUSS as a submodule.

"Easy" Install Scripts

Clone this repository with

git clone --recursive https://github.com/dilevin/GAUSS.git

Install Instructions OS X (Basic install, no bells or whistles)

1. Install Homebrew (https://brew.sh)
2. At command line: chmod a+x ./InstallGAUSS_OSX.sh
3. Run InstallGAUSS_OSX.sh which does the following:
   • downloads and installs Qt 5.9 in ~/Qt
   • installs CMake using homebrew (upgrades currently installed version)
   • installs Eigen 3 using home brew (upgrades currently installed version)
   • installs libigl in ./build/libigl
   • installs Gauss in ./build

Install Instructions Ubuntu (Basic install, no bells or whistles)

1. This install procedure requires gcc and g++ version 5 or greater setup as the default c/c++ compilers
2. At command prompt: chmod a+x ./InstallGAUSS_Ubuntu.sh
3. Run InstallGAUSS_Ubuntu.sh which does the following
   • downloads and installs Qt 5.9 in ~/Qt
   • install libigl into ./build/libigl
   • installs CMake using apt-get
   • installs Eigen 3 using apt-get
   • installs Gauss in ./build

Warning

These scripts are hand maintained to make GAUSS Easy to build on certain platforms. If they fail please resort to custom install using cmake below.

Generic Build Instructions

Setup

Ensure that Qt version >= 5.8 is installed. On Ubuntu 16.04 you may have to change the paths to be consistent in ccmake. Something like the following. /home//Qt/5.9.2/gcc_64/lib/cmake/

Download (and compile?) the igl library into the directory above ${GAUSS_DIR} or choose a custom path in config.cmake.

If you are on OSX and you wish to enable OpenMP, install llvm via homebrew and follow the instructions in config.cmake

Likewise, if you wish to use the pardiso solver, download the pardiso library from the website and set the path in config.cmake. If you get errors referring to libgfortran, you may need to follow this guide to fix the library for your paths.

Build

In the root of this repository do the following

mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release -C ../config.cmake ../
make

To see if it worked, run the tests and examples

./bin/Tests
./bin/Example1
./bin/Example2
./bin/Example3
./bin/Example4 # If OpenMP was enabled

A Note About the Examples

To play examples in Gauss press 'p' once the Qt window appears.

MATLAB Interface

GAUSS Includes a rudimentary MATLAB interface, tested using MATLAB 2015b and 2017a on OSX. To enable the MATLAB interface, build the project Gauss_MATLAB, then open MATLAB and issue the following commands:

addpath('{Gauss_Root_Dir}/src/MATLAB/')
addpath('{Gauss_Root_Dir}/build/lib/{Build_Mode_of_Gauss_MATLAB}/')
savepath

An example of using the MATLAB interface is given in {Gauss_Root_Dir}/src/Examples/example8.m