Datasets for "Automatic Design of Mechanical Metamaterial Actuators"

<p><strong>Publication</strong></p> <p>arXiv preprint:&nbsp;<a href="https://arxiv.org/abs/2002.03032">https://arxiv.org/abs/2002.03032</a></p> <p>journal version: [add link to journal]</p> <p>&nbsp;</p> <p><strong>Code</strong></p> <p><a href="https://github.com/ComplexityBiosystems/metamech">https://github.com/ComplexityBiosystems/metamech</a></p> <p><a href="https://github.com/complexitybiosystems/metamech_datasets">https://github.com/complexitybiosystems/metamech_datasets</a></p> <p>&nbsp;</p> <p><strong>Datasets</strong></p> <p>Figure 2:</p> <ul> <li><em>antiparallel_{human,machine}_design.stl</em>: STL files for human design and machine design configurations performing antiparallel movement.</li> </ul> <p>Figure 3:&nbsp;</p> <ul> <li><em>crane_scaling_dataframe.csv: </em>raw efficiency values and execution times used to generate&nbsp;Figure 3</li> </ul> <p>Figure 4:</p> <ul> <li><em>eta_random.dat</em>,&nbsp;<em>eta_triang.dat, plot_distrib.gnupot</em>: raw values of efficiency and gnuplot script used to generate Figure 4 panel d</li> <li><em><em>eta_vs_displ_random.dat</em>,&nbsp;<em>eta_vs_displ_triang.dat</em></em><em>, eta_vs_displ.gnuplot</em>: raw values of efficiency&nbsp;and gnuplot script&nbsp; used to generate Figure 4 panel c</li> <li><em>structures.tar.bz2</em>: all the structures relative to Figure 4 panel d, in a format suitable for visualization with ovito software</li> </ul> <p>Figure 5:&nbsp;&nbsp;data files for human and machine generated pairs of pliers in a format suitable for visualization with ovito software.&nbsp;&nbsp;&quot;*.agr&quot; and &quot;*.data&quot; files can be opened with any text editor to directly visualize the raw data.&nbsp;&nbsp;</p> <ul> <li><em>*_rest.data:</em>&nbsp;rest configurations which are represented in grey in Fig. 5a and Fig. 5b.&nbsp;</li> <li><em>*_F0.01.data</em>:&nbsp;displaced configurations with input force Fext = 0.01 and output spring stiffness kext = 0.01, also reported in Fig. 5a and Fig. 5b.</li> <li><em>*.agr</em>: dataset for Figure 5c and 5d in xmgrace format.</li> </ul> <p>Figure 6:</p> <ul> <li><em>Fig6/Fig6b/*.csv</em>: raw values of efficiency used to generate Figure 6 panel b</li> <li><em>Fig6/</em><em>Fig6c-training-data/*/*.png:&nbsp;</em>a total of&nbsp;1.163.733 PNG images, of size 192 x 168 pixels, as described in the methods section of the manuscript, organized in directories corresponding to different runs.</li> <li><em>Fig6/Fig6c-training-data/metadata.csv: </em>maps each&nbsp;image path to its associated efficiency value</li> <li><em>Fig6/Fig6c-efficiency-data/eta_{CNN,DEM}.txt</em>: raw values of efficiency used to generate&nbsp;Figure 6 panel c.</li> </ul> <p>Figure 7:</p> <ul> <li><em>Fig7/transfer_learning_regression_values.csv: full raw data, including regression coefficients, p-values, slopes and intercepts corresponding to Figure 7 panels i and j</em></li> </ul> <p><em>Figure S3:</em></p> <ul> <li><em>orthogonal_{human,machine}_design.stl: STL files for human design and machine design configurations performing orthogonal movement.</em></li> </ul> <p>&nbsp;</p>