https://github.com/mdavezac/old-optimet
https://github.com/mdavezac/old-optimet
Last synced: about 1 month ago
JSON representation
- Host: GitHub
- URL: https://github.com/mdavezac/old-optimet
- Owner: mdavezac
- License: other
- Created: 2015-12-14T13:01:51.000Z (over 9 years ago)
- Default Branch: master
- Last Pushed: 2017-01-27T14:02:33.000Z (over 8 years ago)
- Last Synced: 2025-02-16T01:29:53.630Z (3 months ago)
- Language: C++
- Size: 1.17 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
Optimet
=======
Optimet is a simulation of multiple-scattering of light on a distribution of homogeneous spherical
nanoparticles embedded in a homogeneous medium. It can run in parallel on large clusters, using MPI.
It accepts several linear-system solvers, including a direct solver from
[Scalapack](http://www.netlib.org/scalapack/), and the iterative solvers from
[Belos](https://trilinos.org/packages/belos/). Optimet is a c++11 program.
Optimet is distributed under the GNU Public License.
Usage
=====
Example inputs are provided in the `example` subdirectory. The input is in XML format. It can
specify any arbitrary distribution of particles.
The compiled program is called with a single argument, the path to the XML input:
```
Optimet3D input.xml
```
Installation
============
Dependencies
------------
Optimet depends on the following external packages. In most cases, the build system will try to
download and install dependencies it cannot find on the system.
- [CMake](https://cmake.org/): The build system. Must be installed independantly.
- MPI: Required to run in parallel. Must be installed independantly.
- Scalapack: (optional) Parallel linear algebra. Must be installed independantly. Only usefull whhen
compiling with MPI.
- [Belos](https://trilinos.org/packages/belos/): (optional) A library of iterative solvers. Must be
installed independantly. Only usefull when compiling with MPI.
- [Boost](http://www.boost.org/): (required) A set of peer-reviewed c++ libraries. At this juncture,
Optimet only requires the [math special functions
module](http://www.boost.org/doc/libs/1_63_0/libs/math/doc/html/special.html). Automatically
downloaded if unavailable.
- [Eigen](http://eigen.tuxfamily.org/Findex.php?title=Main_Page): (required) A c++ linear algebra
library. Automatically downloaded if unavailable.
- [hdf5](https://support.hdfgroup.org/HDF5/): (required) A standard library for handling data.
Automatically downloaded if unavailable. Only the C bindings are used. The c++ bindings are not
necessary.
- [GSL](https://www.gnu.org/software/gsl/): (required) Numerical library for C and C++.
Automatically downloaded if unavailable.
- [F2C](http://www.netlib.org/f2c/): (required) Fortran to C library. Required by the Amos package.
- [Benchmark](https://github.com/google/benchmark): (optional) Micro-benchmarking library from
google. Automatically downloaded if unavailable *and* benchmarks are requested.
- [Catch](https://github.com/philsquared/Catch): (optional) A c++ testing framework. Downloaded
automatically if tests are required.
Installation
------------
The build system is [CMake](https://cmake.org/). It can generate a build environment from a fair
number of systems, from Unix makefiles to Xcode projects.
The cannonical usage is as follows:
```
cd /path/to/optimet_source
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release -Ddompi=ON -Ddotesting=OFF -Ddobenchmarks=OFF ..
make
```
The example above will *not* compile the tests and benchmarks. Leave these options out (or set them
to `ON`) if you would rather have them. Similarly, the example compiles with MPI parallelization
enabled. Remove or set to `ON` to compile a serial code only.
The executable `Optimet3D` should be directly in the build directory. We currently provide not
installation mechanism.
The benchmarks and unit-tests (when compiled) are in the `benchmarks` and `unittests` subdirectories
of the build directory. They consists in a number of executables which can be invoked manually.
Alternatively, the tests can be invoked from the build directory with `make test` or `ctest .`.
Supported Platforms
-------------------
Optimet is routinely tested on:
- Linux
- MacOS
It is known to compile with:
- gnu-g++ > 4.8
- clang
- intel > 13.0