Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/upc-ghs/gpkde
Fortran Code for Grid Projected Kernel Density Estimation From Multidimensional Particle Distributions
https://github.com/upc-ghs/gpkde
concentration density-estimation fortran kernel particles three-dimensional
Last synced: about 1 month ago
JSON representation
Fortran Code for Grid Projected Kernel Density Estimation From Multidimensional Particle Distributions
- Host: GitHub
- URL: https://github.com/upc-ghs/gpkde
- Owner: upc-ghs
- License: mit
- Created: 2022-12-19T11:06:18.000Z (about 2 years ago)
- Default Branch: develop
- Last Pushed: 2024-08-08T06:50:27.000Z (5 months ago)
- Last Synced: 2024-08-08T08:20:28.105Z (5 months ago)
- Topics: concentration, density-estimation, fortran, kernel, particles, three-dimensional
- Language: Fortran
- Homepage:
- Size: 58 MB
- Stars: 3
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE.md
Awesome Lists containing this project
README
# gpkde
Fortran Code for Grid Projected Kernel Density Estimation of Discrete Particle Distributions
[](https://github.com/upc-ghs/gpkde/actions/workflows/ci.yml?branch=develop&kill_cache=1)
## Overview
The program performs Grid Projected Kernel Density Estimation (GPKDE) of a discrete dataset in one, two or three dimensional domains and is parallelized with the OpenMP library.
It works as a standalone program by reading an input simulation file, which configures the loading of a source file with data points and additional parameters for defining the reconstruction grid and the optimization for bandwidth selection.
## Build
### Get the code
Clone the repository
```
git clone https://github.com/upc-ghs/gpkde.git
```
### Makefile
Repository includes two makefiles at the folder `make`:
- `Makefile`: for the `gfortran` compiler, with compilation verified for `gfortran>=8.4.0`.
- `Makefile-ifort`: for the `ifort` compiler, with compilation verified for `[email protected]`.
**Note**: The preprocessor variable ``-DREAL32`` can be added to the compiler flags in order to build with single precision floating point. Compilation via makefiles create the folder `make/objtemp`. It is recommended to remove this folder while developing source code or integrating program changes for a consistent compilation.
### Meson
The project can also be built with the [meson](https://mesonbuild.com/) build system:
- Setup the project configuration
```
meson setup builddir -Ddebug=false --prefix=$(pwd) --libdir=bin
```
- Compile
```
meson install -C builddir
```
- Basic tests
```
meson test --verbose --no-rebuild -C builddir
```
**Note**: For building with single precision floating point, the option ``--Dreal32=true`` can be given to the setup instruction.
### Visual Studio and Windows
The folder `msvs` contains the project and solution files verified for Visual Studio 2019 and 2022 on a Windows system. An executable file and complementary `dll`'s providing the OpenMP library are available at the folder `wbin/`. Users can make easy use of these files at a system level by extending the `PATH` environment variable, adding their specific address to the `wbin/` folder (as [here](https://www.itprotoday.com/windows-server/how-can-i-add-new-folder-my-system-path) or [here](https://windowsloop.com/how-to-add-to-windows-path/)).
The necessary `dll`'s can also be installed from the Intel and Visual Studio redistributables (`x64`), respectively:
- [Intel Fortran Compiler Runtime for Windows](https://www.intel.com/content/www/us/en/developer/articles/tool/compilers-redistributable-libraries-by-version.html)
- [Microsoft Visual C++ Redistributable](https://learn.microsoft.com/en-US/cpp/windows/latest-supported-vc-redist?view=msvc-170)
**Note**: The windows executable `gpkdesp.exe` is built with single precision floating point (`/DREAL32`) which requires less memory specially for large three dimensional problems. This option is enabled by default in the Visual Studio solution/project files and can be disabled by modifying the project properties, leaving empty the field `Project>Properties>Fortran>Preprocessor>Preprocessor Definitions`. The executable `gpkde.exe` is built with double precision floating point.
## Command line interface
Some basic command line arguments have been implemented in order to control program execution. These can be requested as help with the instruction ``gpkde --help`` or ``gpkde -h``, which displays the following message in console:
```
GPKDE version *.*.*
Program compiled Apr 12 2023 19:44:24 with GFORTRAN compiler (ver. *.*.*)
Fortran code for Grid Projected Kernel Density Estimation of discrete particle distributions
usage:
gpkde [options] simfile
options:
-h --help Show this message
-l --logname Write program logs to
-nl --nolog Do not write log file
-np --nprocs Run with processes
-p --parallel Run in parallel
-v --version Show program version
For bug reports and updates, follow:
https://github.com/upc-ghs/gpkde
```
## Simulation file
For details about the configuration of input files please refer to the program [Documentation](doc/gpkde_IO_v100.pdf).
## Examples
A set of possible use cases of the reconstruction module are included as example files in this repository. The currently available simulations include:
- [1D Gaussian distribution](examples/ex01_1dnormal/)
- [2D Heterogeneous distribution](examples/ex02_2dhet/)
- [3D Heterogeneous distribution](examples/ex03_3dhet/)
## License
MIT License
## References
[Pérez-Illanes R., Fernàndez-Garcia, D., 2024, A General Purpose Parallel Fortran Code for Grid Projected Concentration Reconstruction from Multidimensional Particle Distributions, Environmental Modelling & Software, doi: 10.1016/j.envsoft.2024.106008.](https://doi.org/10.1016/j.envsoft.2024.106008)
## Resources
* [gfortran](https://gcc.gnu.org/wiki/GFortran)
* [OpenMP](https://www.openmp.org/)
* [Intel oneApi HPC toolkit](https://www.intel.com/content/www/us/en/developer/tools/oneapi/hpc-toolkit.html)
* [MIT License](https://mit-license.org/)