Code and data of the article:
Random Auxetic Porous Materials from Parametric Growth Processes
Jonàs Martínez
Computer Aided Design (2021)
- We introduce an algorithmically-defined growth process that is able to generate a random porous material with negative Poisson's ratio.
- The growth process involves randomly placed non-overlapping cells that grow according to a parameterized growth law.
- The growth process can be computed very efficiently and enables the generation of large-scale random auxetic materials in commodity computers.
The data from the article is at data.tar.gz
Tested on Ubuntu 20.04.2 LTS
- Anaconda Python 3 distribution (https://www.anaconda.com/)
Type on a terminal:
sudo apt-get install g++ cmake libcairo2-dev octave texlive-latex-extra dvipng cm-super python3-pip
conda install -c anaconda numpy pandas seaborn
conda install -c conda-forge scikit-optimize
pip3 install pdfCropMargins --user --upgrade
To compile type on a terminal:
cd code/
cmake -DCMAKE_BUILD_TYPE=Release .
make
If everything goes well, an executable file called growthprocess2d will be created in the folder code/
The executable growthprocess2d has a command line parameter that specifies the path to the text file with the input parameters. The input parameters file has a simple format, consisting of attribute-value pairs in each line, separated by the symbol '=' The following table gives a brief description of each attribute and the expected value.
| Attribute | Value | Description |
|---|---|---|
| name | string | Path where the results (adding the extension, e.g., png, ppm, txt) are going to be saved |
| point_process | [Random, RSA] | Which point process is used: Poisson point process (Random) or Random Sequential Adsorption (RSA) |
| image_size | integer | Width and height of the porous material image |
| symmetry_type | [NoSymmetry, RotationalSymmetry, ReflectionalSymmetry] | Symmetry imposed to the star-shaped set governing the growth process |
| symmetry_degree | integer | Degree of the symmetry of the star-shaped set (if any). For instance, a value of three and having RotationalSymmetry implies three-fold rotational symmetry |
| random_seed | integer | Random seed for the growth process |
| rsa_max_dist_rel_pixel | float | If RSA point process is selected, this defines the maximum distance between two points of the point process |
| num_sites_per_pixel | float | If a Poisson point process is selected, this defines the Poisson point process intensity |
| radial_spans | array of floats separated by comma | List of known values of the star-shaped set S (radial spans) |
| max_growth_length_radial_spans | array of floats separated by comma | List of known values of the star-shaped set S* (radial spans) |
The quantities image_size, rsa_max_dist_rel_pixel, and num_sites_per_pixel are relative to the size of a pixel, and a pixel has dimension 1x1.
For other parameters, please refer to the exemplar files in the folder named code/parameters/, and also have a look at the file code/parameters.h
A single script named code/all.sh is provided that generates most of the results and figures of the article. This script must be executed from the folder code/ with
./all.sh
and the resulting data will populate the folder results/
This work was partly supported by ANR MuFFin (ANR-17-CE10-0002).
The 2D homogenization function from code/homogenize_2d.m was taken from the article
How to determine composite material properties using numerical homogenization
Andreassen, E., & Andreasen, C. S.
Computational Materials Science, 83, 488-495. (2014)
The binary file in the folder code/data/ is a large set of points (RSA point process) precomputed with PDSample