Find research datasets worth reusing
Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.
768
datasets available to search
ShareScore release 0.9.0
Dataset results
768 results for “reinforcement”
Cyclic tensile-compressive tests on thin concrete boundary elements with a single layer of reinforcement prone to out-of-plane instability
<p>The growing need for residential housing in Latin American countries has led to the construction of reinforced concrete buildings with wall thicknesses as low as 8-10 cm. Such walls have typically only a single layer of vertical rebars and are therefore particularly susceptible to out-of-plane failure. To investigate the response of the corresponding wall boundary elements, twelve reinforced concrete columns with a single layer of vertical rebars were tested under tension-compression cycles. The objective of this test series was to gain insights into parameters triggering wall instability and out-of-plane failure. The experimental tests investigate the effect of thickness, reinforcement ratio, and eccentricity of the longitudinal rebars with respect to the element axis. This paper summarises the results of the test campaign. The specimen response is analysed at the global and local level, and the influence of the crack pattern on the out-of-plane response of the column and the conditions leading to out-of-plane failure are described. Furthermore, the differences between members with a single layer of vertical rebars to members with two layers are discussed. The influence on the response of the parameters analysed in the experimental campaign is addressed, showing that section with small thickness and large reinforcement content are more prone to out-of-plane failures. Finally, the predictions of existing models are compared to the new experimental data. The entire data set is publically available.</p>
Reaching the limit in autonomous racing: Optimal control versus reinforcement learning
<p>A central question in robotics is how to design a control system for an agile mobile robot. This paper studies this question systematically, focusing on a challenging setting: autonomous drone racing. We show that a neural network controller trained with reinforcement learning (RL) outperformed optimal control (OC) methods in this setting. We then investigated which fundamental factors have contributed to the success of RL or have limited OC. Our study indicates that the fundamental advantage of RL over OC is not that it optimizes its objective better but that it optimizes a better objective. OC decomposes the problem into planning and control with an explicit intermediate representation, such as a trajectory, that serves as an interface. This decomposition limits the range of behaviors that can be expressed by the controller, leading to inferior control performance when facing unmodeled effects. In contrast, RL can directly optimize a task-level objective and can leverage domain randomization to cope with model uncertainty, allowing the discovery of more robust control responses. Our findings allowed us to push an agile drone to its maximum performance, achieving a peak acceleration greater than 12 times the gravitational acceleration and a peak velocity of 108 kilometers per hour. Our policy achieved superhuman control within minutes of training on a standard workstation. This work presents a milestone in agile robotics and sheds light on the role of RL and OC in robot control.</p>
Tensile test results for 3D printed unidirectional composites reinforced with continuous fibres
Open the record for dataset details and reuse information.
Supplemental data to AALE 2024 publication "Adaptive manufacturing: dynamic resource allocation using multi-agent reinforcement learning"
<p>Release as supplementary material for our contribution at AALE 2024: "Adaptive manufacturing: dynamic resource allocation using multi-agent reinforcement learning"<br><br>The evaluation datasets stored in this collection are used to compare the performance of multi-agent reinforcement learning. In addition, the performance of other methods such as (meta-) heuristic algorithms or single agent reinforcement learning algorithms or novel methods of search space reduction can also be compared.</p>
A multiscale theory for spreading and migration of adhesion-reinforced mesenchymal cells
<p>We present a chemomechanical whole-cell theory for the spreading and migration dynamics of mesenchymal cells that can actively reinforce their adhesion to an underlying viscoelastic substrate as a function of its stiffness. Our multiscale model couples the adhesion reinforcement effect at the subcellular scale with the nonlinear mechanics of the nucleus-cytoskeletal network complex at the cellular scale to explain the concurrent monotonic area-stiffness and non-monotonic speed-stiffness relationships observed in experiments: We consider that large cell spreading on stiff substrates flattens the nucleus, increasing the viscous drag force on it. The resulting force balance dictates a reduction in the migration speed on stiff substrates. We also reproduce the experimental influence of the substrate viscosity on the cell spreading area and migration speed by elucidating how the viscosity may either maintain adhesion reinforcement or prevent it depending on the substrate stiffness. Additionally, our model captures the experimental directed migration behavior of the adhesion-reinforced cells along a stiffness gradient, known as durotaxis, as well as up or down a viscosity gradient (viscotaxis or anti-viscotaxis), the cell moving towards an optimal viscosity in either case. Overall, our theory explains the intertwined mechanics of the cell spreading, migration speed and direction in the presence of the molecular adhesion reinforcement mechanism. </p>
Keynote: Bringing Reinforcement learning Into Radio Light Network for Massive Connections
<blockquote><p>3GPP standardization has been progressing at an astonishingly rapid phase, where Release 15 and Release 16 have set the foundations of the 5G system, while Release 17 provides enhancements and optimizations to enable support for further use cases. In parallel to 5G standardization efforts, several initiatives worldwide endeavour to drive and support the evolution of smart networks and services. Among others Europe is establishing the <i>Joint Undertaking on Smart Networks and Services</i> in the frame of the Horizon Europe programme for research and innovation. Other initiatives are complementing the European initiative, such as <i>Secure 5G & Beyond Act</i> in the U.S., <i>roadmap towards 6G </i>in Japan, <i>MSIT 6G programme</i> in S. Korea, and <i>MIIT 6G programme</i> in China.</p></blockquote><blockquote><p>The workshop will provide an opportunity for reflection and discussion about requirements and architectural considerations for future generations of mobile systems. The focus will be on presenting version 4.0 of the Architecture white paper developed by the 5G PPP architecture working group. It will allow to move from 5G and beyond towards a fully-fledged 6G architecture.</p></blockquote>
Test results from quasi-static impact tests over SMC materials reinforced with glass fibre (ramdom and textile)
<p>Test results from quasi-static impact tests over SMC materials with structural properties conceived for automotive applications, reinforced with glass fibre (samples with ramdom distribution of fibres and samples reinforced with GF textile)</p>
Results from DSC analysis over glass fibre reinforced SMC (50009092 grade by Menzolit supplier)
<p>Results from DSC analysis over glass fibre reinforced SMC automotive grade (50009092 grade provided by Menzolit supplier)</p>
Results from DSC analysis over glass fibre reinforced SMC (50007150 grade by Menzolit supplier)
<p>Results from DSC analysis over glass fibre reinforced SMC for automotive applications (50007150 grade provided by Menzolit supplier)</p>
Interaction Diagrams of Geopolymer FRC Slender Columns with Double-Layer Reinforcement_Dataset
<p>The peer-reviewed data descriptor for this dataset has now been published in MDPI Data - an open-access journal aiming at enhancing data transparency and reusability, and can be accessed here: <a href="https://www.mdpi.com/2306-5729/6/5/43">https://www.mdpi.com/2306-5729/6/5/43</a>. Please use the following citation when using the dataset: AlHamaydeh, Mohammad, and Amin, Fouad. "Data for Interaction Diagrams of Geopolymer FRC Slender Columns with Double-Layer GFRP and Steel Reinforcement," <em>Data</em>, Vol. 6, No. 5, 2021, pp. 43-49. DOI:10.3390/data6050043.</p> <p>This dataset provides the normalized axial load/bending moment capacities of circular concrete columns with and without fiber reinforcement. Different longitudinal and transverse reinforcement types are incorporated. Each data file represents a group of columns with specific reinforcement configurations. Within each file, the axial load/bending moment capacities are given for different reinforcement ratios, slenderness ratios, reinforcement strength levels, and loading eccentricities. The provided data could be used in the development of axial load bending moment interaction diagrams. Moreover, further processing of the data could be used for plotting the analytical strength curves of the analyzed columns.</p> <p>Version v1: Assumes identical compressive and tensile strengths for the GFRP reinforcement. This assumption is made based on the lateral support provided by the concrete surrounding the reinforcement. </p> <p>Version v2: Assumes a more realistic assumption of the compressive strength being 80% of the tensile strength for the reinforcement. Selected supporting references:<a name="_Hlk83211528"></a></p> <p>[1] ACI Committee 440. Guide for the Design and Construction of Structural Concrete Reinforced with FRP Bars, ACI 440.1R-15. vol. 88 Reappro. Farmington Hills, MI : American Concrete Institute: 2015.</p> <p>[2] Hadhood A, Mohamed HM, Benmokrane B. Flexural Stiffness of GFRP- and CFRP-RC Circular Members under Eccentric Loads Based on Experimental and Curvature Analysis. ACI Struct J 2018;115:1185–98. https://doi.org/10.14359/51702235.</p> <p> </p>
Experimental and genetic analysis of selfing reveals no reinforcement in Phlox cuspidata
<p>Reinforcement is the process through which prezygotic reproductive barriers evolve in sympatry due to selection against hybridization between co-occurring, closely related species. The role of self-fertilization in reinforcement and reproductive isolation is uncertain in part because its efficiency as a barrier against heterospecific mating can depend on the timing of autonomous selfing.</p> <p>To investigate whether increased autonomous selfing has evolved as a mechanism for reinforcement, we compared <em>Phlox cuspidata</em> populations across their native Texas range using both estimates of genetic diversity and experimental manipulation with morphological measurements. Specifically, we investigated patterns of variation in floral traits and timing of selfing between individuals from allopatric populations of <em>P. cuspidata</em> and from populations sympatric with the closely related species, <em>P. drummondii</em>.</p> <p>We infer intermediate rates of selfing across field-collected individuals with no significant difference between allopatric and sympatric populations. Among greenhouse grown plants, we find no differences in timing of selfing or other floral traits including anther dehiscence timing, anther-stigma distances, autonomous selfing rate and self-seed count between allopatric and sympatric populations. However, our statistical analyses indicate that <em>P. cuspidata </em>individuals sympatric with <em>P drummondii</em> seem to have generally larger flowers compared to allopatric individuals.</p> <p>Despite strong evidence of costly hybridization with <em>P. drummondii</em>, we find no evidence of trait divergence due to reinforcement in <em>P. cuspidata. </em>Although we document nearly complete autonomous self-seed set in the greenhouse, estimates of selfing rates from genetic data imply realized selfing is much lower in nature suggesting an opportunity for reinforcing selection to act on this trait. </p>
Data of "Self-consistency Reinforced minimal Gated Recurrent Unit for surrogate modeling of history-dependent non-linear problems: application to history-dependent homogenized response of heterogeneous materials"
<h1>Development of the Self-Consistency reinforced Minimum Recurrent Unit (SC-MRU)</h1> <p>This directory contains the data and algorithms generated in publication<sup><a href="#fn-1-5292">1</a></sup></p> <h2>Table of Contents</h2> <ol> <li><a href="#dependencies-and-prerequisites">Dependencies and Prerequisites</a></li> <li><a href="#structure-of-repository">Structure of Repository</a></li> <li><a href="#part-1-data-preparation">Part 1: Data preparation</a></li> <li><a href="#part-2-rnn-training">Part 2: RNN training</a></li> <li><a href="#part-3-multiscale-analysis">Part 3: Multiscale analysis</a></li> <li><a href="#part-4-reproduce-paper1-figures">Part 4: Reproduce paper[^1] figures</a></li> </ol> <h2>Dependencies and Prerequisites</h2> <p> </p> <ul> <li> <p>Python, pandas, matplotlib, texttabble and latextable are pre requisites for visualizing and navigating the data.</p> </li> <li> <p>For generating mesh and for vizualization, gmsh (<a href="http://www.gmsh.info" target="_blank" rel="nofollow noreferrer noopener">www.gmsh.info</a>) is required.</p> </li> <li> <p>For running simulations, cm3Libraries (<a href="http://www.ltas-cm3.ulg.ac.be/openSource.htm" target="_blank" rel="nofollow noreferrer noopener">http://www.ltas-cm3.ulg.ac.be/openSource.htm</a>) is required.</p> </li> </ul> <h3>Instructions using apt & pip3 package manager</h3> <p>Instructions for Debian/Ubuntu based workstations are as follows.</p> <h3>python, pandas and dependencies</h3> <div> <pre><code> sudo apt install python3 python3-scipy libpython3-dev python3-numpy python3-pandas</code></pre> </div> <h3>matplotlib, texttabble and latextable</h3> <div> <pre><code> pip3 install matplotlib texttable latextable</code></pre> </div> <h3>Pytorch (only for run with cm3Libraries)</h3> <ul> <li>Without GPU</li> </ul> <div> <pre><code> pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu</code></pre> </div> <ul> <li>With GPU</li> </ul> <div> <pre><code> pip3 install torch torchvision torchaudio</code></pre> </div> <h3>Libtorch (for compiling the cells)</h3> <ul> <li>Without GPU: In a local directory (e.g. <code>~/local</code> with <code>export TORCHDIR=$HOME/local/libtorch</code>)</li> </ul> <div> <pre><code> wget https://download.pytorch.org/libtorch/cpu/libtorch-shared-with-deps-2.1.1%2Bcpu.zip unzip libtorch-shared-with-deps-2.1.1%2Bcpu.zip</code></pre> </div> <ul> <li>With GPU: In a local directory (e.g. <code>~/local</code> with <code>export TORCHDIR=$HOME/local/libtorch</code>)</li> </ul> <div> <pre><code> wget https://download.pytorch.org/libtorch/cu121/libtorch-shared-with-deps-2.1.1%2Bcu121.zip unzip libtorch-shared-with-deps-2.1.1+cu121.zip</code></pre> </div> <h2>Structure of Repository</h2> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/All_Path_Res">All_Path_Res</a>: results of the direct numerical simulations used as training and testing data, see details in <a href="#part-1-data-preparation">Part 1: Data preparation</a>.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/ConstRVE">ConstRVE</a>: script to run direct numerical finite element simulations, see details in <a href="#part-1-data-preparation">Part 1: Data preparation</a>.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale">MultiScale</a>: scripts to run and visualise the multiscale analyses, see details in <a href="#part-3-multiscale-analysis">Part 3: Multiscale analysis</a>.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU</a>: implementation of the RNN and scripts to train them, see details in <a href="#part-2-rnn-training">Part 2: RNN training</a>.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/TrainingData">TrainingData</a>: scripts to collect, normalise and truncate the RVEs direct simulation results as training and testing data, see details in <a href="#part-1-data-preparation">Part 1: Data preparation</a>. The director also contained the storred processed data used in <sup><a href="#fn-1">1</a></sup>.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/TrainingPaths">TrainingPaths</a>: scripts to generate the different loading paths for the direct numerical simulations used as training and testing data, see details in <a href="#part-1-data-preparation">Part 1: Data preparation</a>.</li> </ul> <h2>Part 1: Data preparation</h2> <h3>Generate the loading paths</h3> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingPaths/testGenerationData.py">TrainingPaths/testGenerationData.py</a> is used to generate random walk paths, with the options <ul> <li><code>Rmax = 0.11</code> # bound on the final Green Lagrange strain</li> <li><code>TimeStep = 1.</code> # in second</li> <li><code>EvalStep = [1e-4,5e-3]</code> #Bounds on the Green Lagrange increments</li> <li><code>Nmax = 2500</code> #maximum length of the sequence</li> <li><code>k = 4000</code> # number of path to generate</li> <li>The path are storred by default in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/ConstRVE/Paths">ConstRVE/Paths/</a>. The path has to be existing before launching the script. You can change the name in line 123 <code>saveDir = '../ConstRVE'+'/Paths/'</code>.</li> <li>Examples of generated paths can be found in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/ConstRVE/PathsExamples">ConstRVE/PathsExamples/</a></li> <li>The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/TrainingPaths">TrainingPaths</a> is</li> </ul> </li> </ul> <div> <pre><code>(mkdir ../ConstRVE/Paths) #if needed python3 testGenerationData.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingPaths/generationData_Cyclic.py">TrainingPaths/generationData_Cyclic.py</a> is used to generate random cylic paths, with the options <ul> <li><code>Rmax = [np.random.uniform(0.,0.04),np.random.uniform(0.,0.06),np.random.uniform(0.0,0.09),0.12]</code> # bound on the final Green Lagrange strain is random</li> <li><code>TimeStep = 1.</code> # in second</li> <li><code>EvalStep = [1e-4,5e-3]</code> #Bounds on the Green Lagrange increments</li> <li><code>Nmax = 2500</code> #maximum length of the sequence</li> <li><code>k = 2000</code> # number of path to generate</li> <li>The path are stored by default in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/ConstRVE/Paths">ConstRVE/Paths/</a>. You can change the name in line 123 <code>saveDir = '../ConstRVE'+'/Paths/'</code>.</li> <li>The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/TrainingPaths">TrainingPaths</a> is</li> </ul> </li> </ul> <div> <pre><code>(mkdir ../ConstRVE/Paths) #if needed python3 generationData_Cyclic.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingPaths/countPathLength.py">TrainingPaths/countPathLength.py</a> gives average, minimum and maximum lengths of the generated paths and the distribution of the <code>\Delta R</code>. By default the paths are read in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/ConstRVE/Paths">ConstRVE/Paths/</a> but the directory can be given as an argument. The file can be used to read <ul> <li>either the generated loading paths</li> </ul> </li> </ul> <div> <pre><code> python3 countPathLength.py '../ConstRVE/PathsExamples'</code></pre> </div> <ul> <li> <ul> <li>or the results of the <a href="#generate-the-rves-direct-simulation-results">simulations</a></li> </ul> </li> </ul> <div> <pre><code> python3 countPathLength.py '../All_Path_Res/Path_Res9'</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingPaths/graphData.py">TrainingPaths/graphData.py</a> generates illustrations from randomly picked paths in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/ConstRVE/Paths">ConstRVE/Paths/</a> and generate png figures.</li> </ul> <h3>Generate the RVEs direct simulation results</h3> <ul> <li>Uses the loading paths existing in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/ConstRVE/Paths">ConstRVE/Paths/</a>.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/ConstRVE/rve.geo">ConstRVE/rve.geo</a> is the RVE geometry file that can be read by gmsh (<a href="http://www.gmsh.info" target="_blank" rel="nofollow noreferrer noopener">www.gmsh.info</a>).</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/ConstRVE/rve.msh">ConstRVE/rve.msh</a> is the RVE mesh file that can be read by gmsh (<a href="http://www.gmsh.info" target="_blank" rel="nofollow noreferrer noopener">www.gmsh.info</a>).</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/ConstRVE/utilsFunc.py">ConstRVE/utilsFunc.py</a> contains python tools to be used.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/ConstRVE/Rve_withoutInternalVars.py">ConstRVE/Rve_withoutInternalVars.py</a> is used to run all the RVE simulations: <ul> <li>This requires cm3Libraries (<a href="http://www.ltas-cm3.ulg.ac.be/openSource.htm" target="_blank" rel="nofollow noreferrer noopener">http://www.ltas-cm3.ulg.ac.be/openSource.htm</a>).</li> <li>All the ouptus are stored in <code>All_Path_Res/Path_Res12</code>, you can change the name in line 71 <code>Path_Res = '../All_Path_Res/Path_Res12/'</code>. The results of RVE simulations are saved as the sequence (one configuration per line) of the Green-Lagrange strains and Second Piola-Kirchhoff stress (in column). One example can be found in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/All_Path_Res/Path_Res1/data_path1000.csv">All_Path_Res/Path_Res1/data_path1000.csv</a>.</li> <li>The command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/ConstRVE">ConstRVE</a> directory is</li> </ul> </li> </ul> <div> <pre><code> python3 Rve_withoutInternalVars.py</code></pre> </div> <h3>Collect, normalised and truncate the RVEs direct simulation results as training and testing data</h3> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/CheckNanData.py">TrainingData/CheckNanData.py</a> is used to check the integrity of the direct numerical simulations results <ul> <li>DNS results are read from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/All_Path_Res/Path_res1">All_Path_Res/Path_res1</a> to <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/All_Path_Res/Path_res11">All_Path_Res/Path_res11</a> subdirectories.</li> <li>The command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/TrainingData">TrainingData</a> directory is</li> </ul> </li> </ul> <div> <pre><code> python3 CheckNanData.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/CollectData.py">TrainingData/CollectData.py</a> is used to gather all the direct numerical simulations results <ul> <li>DNS results are read from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/All_Path_Res/Path_res1">All_Path_Res/Path_res1</a> to <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/All_Path_Res/Path_res11">All_Path_Res/Path_res11</a> subdirectories. It can be changed in line 31 <code>for ll in range(11):</code>.</li> <li>It saves the bounds and raw data in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data/Bounds_GS">TrainingData/Processed_Data/Bounds_GS</a> and <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data/Origin_GS">TrainingData/Processed_Data/Origin_GS</a>, respectively.</li> <li>The command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/TrainingData">TrainingData</a> directory is</li> </ul> </li> </ul> <div> <pre><code> python3 CollectData.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Normalization.py">TrainingData/Normalization.py</a> is used to normalise the gathered direct numerical simulations results <ul> <li>Bounds and raw data are read from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data/Bounds_GS">TrainingData/Processed_Data/Bounds_GS</a> and <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data/Origin_GS">TrainingData/Processed_Data/Origin_GS</a>, respectively.</li> <li>It saves the normalized training (75%) and testing (25%) data in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data/Normalized_GS_Train">TrainingData/Processed_Data/Normalized_GS_Train</a> and <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data/Normalized_GS_Test">TrainingData/Processed_Data/Normalized_GS_Test</a>, respectively.</li> <li>The command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/TrainingData">TrainingData</a> directory is</li> </ul> </li> </ul> <div> <pre><code> python3 Normalization.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/GaussCollectData.py">TrainingData/GaussCollectData.py</a> is an alternative using Gaussian normaliation and is used to gather all the direct numerical simulations results.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/GaussNormalization.py">TrainingData/GaussNormalization.py</a> is an alternative to normalise following a Gaussian the gathered direct numerical simulations results.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Data_Padding.py">TrainingData/Data_Padding.py</a> is used to pad and trim the normalised data <ul> <li>Normalized training (75%) and testing (25%) data in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data/Normalized_GS_Train">TrainingData/Processed_Data/Normalized_GS_Train</a> and <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data/Normalized_GS_Test">TrainingData/Processed_Data/Normalized_GS_Test</a>, respectively.</li> <li>The final length of the sequence (including zero padding and trimming) is given by <code>N = 200</code>.</li> <li>It saves the trimmed and padded normalised training and testing data in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data">TrainingData/Processed_Data/</a> as <code>'GS_Train'N</code> and <code>'GS_Test'N</code>, respectively.</li> <li>The command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/TrainingData">TrainingData</a> directory is</li> </ul> </li> </ul> <div> <pre><code> python3 Data_Padding.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Tool.py">TrainingData/Tool.py</a> is a list of function used to normalise dat.</li> </ul> <h2>Part 2: RNN training</h2> <h3>Available rnn</h3> <ul> <li>The different cells are in the following directories <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/MGRU/NNW_SMRU">SC_MRU/MGRU/NNW_SMRU</a>: Neural network with SMRU recurrent cell.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/MGRU/NNW_SCMRU_T">SC_MRU/MGRU/NNW_SCMRU_T</a>: Neural network with SC-MRU-T recurrent cell.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/MGRU/NNW_MGRU">SC_MRU/MGRU/NNW_MGRU</a>: Neural network with orginal MGRU recurrent cell.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/MGRU/NNW_MGRU_M">SC_MRU/MGRU/NNW_MGRU_M</a>: Neural network with modified MGRU recurrent cell.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/FNN_LeakyReLU">SC_MRU/FNN_LeakyReLU/NNW_<code>X</code>Fw</a>: Neural network with SC-MRU-I recurrent cell using <code>X</code> Feed Forward non-linear transition layers.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/Quadratic_NLT">SC_MRU/Quandratic_NLT/NNW_<code>X</code>Q, NNW_Q_Fw, NNW_Fw_Q</a>: Neural network with SC-MRU-I recurrent cell using <code>X</code> quadratic or mixed feed-forward-quadratic non-linear transition layers.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/ReferenceRNN">SC_MRU/ReferenceRNN/NNW_<code>X</code>Layers</a>: Neural network with SC-LMSC recurrent cell using <code>X</code> non-linear transition layers.</li> </ul> </li> </ul> <h3>Compile the neural network models</h3> <ul> <li>In the adequate directory, e.g. <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/MGRU/NNW_SMRU">SC_MRU/MGRU/NNW_SMRU</a> for the Neural network with SMRU recurrent cell.</li> </ul> <div> <pre><code> cd build rm -rf * cmake -DCMAKE_PREFIX_PATH=$TORCHDIR .. make</code></pre> </div> <ul> <li>This create the RNN_<code>CELL</code> model in the build directory</li> </ul> <h3>Train the neural network models</h3> <ul> <li>In the adequate directory, e.g. <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/MGRU/NNW_SMRU">SC_MRU/MGRU/NNW_SMRU</a> for the Neural network with SMRU recurrent cell.</li> <li>Requires to have <a href="#compile-the-neural-network-models">compiled</a> the RNN model.</li> <li>The file <code>Train.py</code> <ul> <li>Uses the RNN_<code>CELL</code> model compiled in the <code>build</code> directory.</li> <li>Uses the trimmed and padded normalised training and testing data in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/TrainingData/Processed_Data">TrainingData/Processed_Data/</a> as <code>'GS_Train'N</code> and <code>'GS_Test'N</code>, respectively.</li> <li>Can be modified to use the requested ratios of training and testing sequences of different lengths to prepate the mini-baches, e.g.: <ul> <li><code>ratio = [0.2,0.02,0.6,0.06,0.01,0.9]</code></li> <li><code>PathIn1 = ['../../../TrainingData/Processed_data/GS_Train200','../../../TrainingData/Processed_data/GS_Train2500','../../../TrainingData/Processed_data/GS_Test200','../../../TrainingData/Processed_data/GS_Test2500']</code></li> <li><code>PathIn2 = ['../../../TrainingData/Processed_data/GS_Test200','../../../TrainingData/Processed_data/GS_Test400','../../../TrainingData/Processed_data/GS_Test2500','../../../TrainingData/Processed_data/GS_Test400']</code></li> </ul> </li> <li>Saves the mini-batches in <ul> <li><code>PathOut = "TrainingData.pt"</code></li> </ul> </li> <li>Saves the model <code>module-checkpoint.pt</code> and <code>module-checkpoint-optimizer.pt</code>, and loss evolution <code>Loss.txt</code> in <ul> <li>Module/H<code>n</code> with <code>n</code> the number of hiden variables.</li> <li>Output directory can be changes in NNW_<code>CELL</code>.cpp of the cell name <code>CELL</code> (and recompiling).</li> <li>Warm start can be disabled in by commenting <code>torch::load(net, "./Module/H120/module-checkpoint_CM0.pt");</code> and <code>torch::load(optimizer, "./Module/H120/module-optimizer-checkpoint_CM0.pt");</code> in NNW_<code>CELL</code>.cpp of the cell name <code>CELL</code> (and recompiling).</li> <li>Is executed with</li> </ul> </li> </ul> </li> </ul> <div> <pre><code> python3 Train.py</code></pre> </div> <h3>Convert trained c++ models for pyTorch (in view of multiscale simulations)</h3> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/CheckLoss/Save_RU.py">SC_MRU/CheckLoss/RU.py</a>: functions used to read and use the rnn models.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/CheckLoss/Save_RNN_script.py">SC_MRU/CheckLoss/Save_RNN_script.py</a>: <ul> <li>Converts c++ trained models to pyTorch models</li> <li>Reads the trained models in SC_MRU/CELL_<code>Kind</code>/NNW_<code>CELL</code>/Module/H<code>N</code>/module.pt</li> <li>Save the converted models to <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a> and to <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/Model">MultiScale/Model/</a></li> <li><code>InpType = cell[0]</code> #choose the recurrent cell typeAmong the different available cells <code>cell=["SC_MRU_T","SC_MRU_I","SMRU"]</code>.</li> <li>The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a> is</li> </ul> </li> </ul> <div> <pre><code>python3 Save_RNN_script.py</code></pre> </div> <h3>Vizualize loss evolution and testing results</h3> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/CheckLoss/Data_prepare.py">SC_MRU/CheckLoss/Data_prepare.py</a> contains fucntion used for testing.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/CheckLoss/LossVS_insertN.py">SC_MRU/CheckLoss/LossVS_insertN.py</a> tests the effect of testing data augmentation: <ul> <li>Requires to have <a href="#compile-the-neural-network-models">compiled</a> the RNN models.</li> <li>Cell type and augmentation kind can be modified: <ul> <li><code>repeat = 10</code> #number of tests</li> <li><code>dataType ='random'</code> #'even' or 'random'</li> <li><code>InpType = cell[0]</code> #choose the recurrent cell typeAmong the different available cells ```cell=["SC_MRU_T","SC_MRU_I","SMRU"]``</li> <li><code>reevaluate= False</code> # <code>False</code> to use the saved loss values and <code>True</code> to evaluate the loss values</li> <li><code>fastshifting=False</code> # <code>False</code> to vizualize before fast shifting (Fig. 11) and <code>True</code> after (Fig. 12). When <code>reevaluate== True</code>, this has no effect: the training with fast-shifting or not has to be done manually, see <a href="#train-the-neural-network-models">details</a>.</li> </ul> </li> <li>Generates new Loss_<code>CELL</code>.txt and TrainingData.pt files in case <code>reevaluate== True</code>, read the existing one if <code>reevaluate== False</code>.</li> <li>Generates Figs. 11 and 12 from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a></li> </ul> </li> </ul> <div> <pre><code> python3 LossVS_insertN.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/CheckLoss/Plot_GS.py">SC_MRU/CheckLoss/Plot_GS.py</a> test the different NNW on RVE testing paths: <ul> <li>Requires to have <a href="#compile-the-neural-network-models">compiled</a> the RNN models.</li> <li>Cell type and augmentation kind can be modified: <ul> <li><code>InpType = cell[0]</code> #choose the recurrent cell typeAmong the different available cells ```cell=["SC_MRU_T","SC_MRU_I","SMRU"]``</li> </ul> </li> <li>Generate Figs. 13, 14 and 15 from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a></li> <li>Command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a> is</li> </ul> </li> </ul> <div> <pre><code> python3 Plot_GS.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/CheckLoss/Plot_GS_step.py">SC_MRU/CheckLoss/Plot_GS_step.py</a> tests the different pyTorch NNWs on the RVE testing paths: <ul> <li>Requires to have <a href="#convert-trained-c-models-for-pytorch-in-view-of-multiscale-simulations">converted</a> the RNN models.</li> <li>Cell type and augmentation kind can be modified: <ul> <li><code>InpType = cell[0]</code> #choose the recurrent cell typeAmong the different available cells ```cell=["SC_MRU_T","SC_MRU_I","SMRU"]``</li> </ul> </li> <li>Command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a> is</li> </ul> </li> </ul> <div> <pre><code> python3 Plot_GS_step.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/PlotLoss_HiddenV.py">SC_MRU/PlotLoss_HiddenV.py</a> shows the loss evolution for the different number of hidden variables --<a href="#train-the-neural-network-models">after having trained</a> or using saved loss files-- for the SMRU cell: <ul> <li>Generates Fig. 8.</li> <li>Command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU/</a> is</li> </ul> </li> </ul> <div> <pre><code> python3 PlotLoss_HiddenV.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/PlotLoss_3MRU.py">SC_MRU/PlotLoss_3MRU.py</a> shows the loss evolution for the different recurrent units --<a href="#train-the-neural-network-models">after having trained</a> or using saved loss files-- for the 120 hidden variables: <ul> <li>Generates Fig. 9.</li> <li>Command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU/</a> is</li> </ul> </li> </ul> <div> <pre><code> python3 PlotLoss_3MRU.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/PlotLoss_RefNL_H120.py">SC_MRU/PlotLoss_RefNL_H120.py</a> shows the loss evolution for the different SC-LMSC and SC-MRU-I cells --<a href="#train-the-neural-network-models">after having trained</a> or using saved loss files-- for 120 hidden variables: <ul> <li>Generates Fig. 10.</li> <li>Command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU/</a> is</li> </ul> </li> </ul> <div> <pre><code> python3 PlotLoss_RefNL_H120.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/PlotLoss_MGRU.py">SC_MRU/PlotLoss_MGRU.py</a> shows the loss evolution for the original and modified MGRU --<a href="#train-the-neural-network-models">after having trained</a> or using saved loss files-- and for the 120 hidden variables: <ul> <li>Generates Fig. A.18.</li> <li>Command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU/</a> is</li> </ul> </li> </ul> <div> <pre><code> python3 PlotLoss_MGRU.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/PlotLoss_NL.py">SC_MRU/PlotLoss_NL.py</a> shows the loss evolution for the different non-liner transition layers (quadratic and Leaky ReLU) of the SC-MRU-I cell --<a href="#train-the-neural-network-models">after having trained</a> or using saved loss files-- and for the 120 hidden variables: <ul> <li>Generates Fig. B.19.</li> <li><code>Case=1</code> # 0 for quadratic transition blocks and 1 for Leaky ReLU transition blocks</li> <li>Command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU/</a> is</li> </ul> </li> </ul> <div> <pre><code> python3 PlotLoss_NL.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/SC_MRU/PlotLoss_DiffNL_H120.py">SC_MRU/PlotLoss_DiffNL_H120.py</a> shows the loss evolution for the different non-liner transition layers (hybrid) of the SC-MRU-I cell --<a href="#train-the-neural-network-models">after having trained</a> or using saved loss files-- and for the 120 hidden variables: <ul> <li>Generates Fig. B.20.</li> <li>Command to be run from <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU/</a> is</li> </ul> </li> </ul> <div> <pre><code> python3 PlotLoss_DiffNL_H120.py</code></pre> </div> <h2>Part 3: Multiscale analysis</h2> <h3>Trained surrogate models</h3> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/Model">MultiScale/Model</a>: contains the different rnn <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/Model/Bounds_GS">MultiScale/Model/Bounds_GS</a>: Bounds.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/Model/DInpFullModel.pt">MultiScale/Model/DInpFullModel.pt</a>: trained rnn with <code>SC-MRU-T</code> recurrent cell.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/Model/IncrementModel.pt">MultiScale/Model/IncrementModel.pt</a>: trained rnn with <code>SC-MRU-I</code> recurrent cell.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/Model/FullModel.pt">MultiScale/Model/FullModel.pt</a>: trained rnn with <code>SMRU</code> recurrent cell.</li> </ul> </li> </ul> <h3>Run mutiscale simulations using the surrogates</h3> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/2D_MultiScale/model.geo">MultiScale/2D_MultiScale/model.geo</a>: geometry of the macro-scale model that can be read by gmsh (<a href="http://www.gmsh.info" target="_blank" rel="nofollow noreferrer noopener">www.gmsh.info</a>).</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/2D_MultiScale/model.msh">MultiScale/2D_MultiScale/model.msh</a>: mesh of the macro-scale model that can be read by gmsh (<a href="http://www.gmsh.info" target="_blank" rel="nofollow noreferrer noopener">www.gmsh.info</a>).</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/2D_MultiScale/model.py">MultiScale/2D_MultiScale/model.py</a>: is used to run all the multiscale simulation: <ul> <li>This requires cm3Libraries (<a href="http://www.ltas-cm3.ulg.ac.be/openSource.htm" target="_blank" rel="nofollow noreferrer noopener">http://www.ltas-cm3.ulg.ac.be/openSource.htm</a>).</li> <li>Uses the bounds and trained models in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/Model">MultiScale/Model</a>.</li> <li><code>InpType = cell[0]</code> #choose the recurrent cell typeAmong the different available cells <code>cell=["SC_MRU_T","SC_MRU_I","SMRU"]</code>.</li> <li><code>factorStep= 100</code> # is the number of steps x 20 during the reloading stage between points B and C.</li> <li>The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/2D_MultiScale">MultiScale/2D_MultiScale/</a> is</li> </ul> </li> </ul> <div> <pre><code>python3 model.py</code></pre> </div> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/FE2">MultiScale/FE2</a>: <ul> <li>Contains the reference displacement-force results of the FE2 simulation.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/FE2/Distributions">MultiScale/FE2/Distributions</a>: contains the macro-scale displacement and stress fields. They can be vizualized with gmsh (<a href="http://www.gmsh.info" target="_blank" rel="nofollow noreferrer noopener">www.gmsh.info</a>) using the mesh file <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/FE2/Distributions/model.msh">model.msh</a>.</li> </ul> </li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale">MultiScale/S<code>CELL</code>_120_<code>STEPS</code></a>: <ul> <li>Contain the reference displacement-force results of the rnn-based multiscale simulations for the different recurrent cells <code>CELL</code> and steps number <code>STEPS</code> during the reloading stage between points B and C.</li> <li>For the cases <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale">MultiScale/S<code>CELL</code>_120_20</a>, the macro-scale displacement and stress fields are also available and can be vizualized with gmsh (<a href="http://www.gmsh.info" target="_blank" rel="nofollow noreferrer noopener">www.gmsh.info</a>) using the mesh file <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/FE2/Distributions/model.msh">model.msh</a>.</li> </ul> </li> </ul> <h3>Vizualize mutiscale simulations results</h3> <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/2D_MultiScale/plot_force.py">MultiScale/2D_MultiScale/plot_force.py</a>: is used to plot the multiscale simulations results: <ul> <li><code>InpType = cell[0]</code> #choose the recurrent cell typeAmong the different available cells <code>cell=["SC_MRU_T","SC_MRU_I","SMRU"]</code>.</li> <li>The multiscale simulations results to be plotted are saved in the directories <code>CELL_120_Step</code>, where <code>Step</code> # is the number of steps during the reloading stage between points B and C.</li> <li>The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/2D_MultiScale">MultiScale/2D_MultiScale/</a> is</li> </ul> </li> </ul> <div> <pre><code>python3 plot_force.py</code></pre> </div> <ul> <li>To vizualize the macro-scale displacement and stress fields distributions: <ul> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/FE2/Distributions">MultiScale/FE2/Distributions</a>: contains the macro-scale displacement and stress fields. They can be vizualized with gmsh (<a href="http://www.gmsh.info" target="_blank" rel="nofollow noreferrer noopener">www.gmsh.info</a>) using the mesh file <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/FE2/Distributions/model.msh">model.msh</a>.</li> <li><a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale">MultiScale/S<code>CELL</code>_120_20</a>, the macro-scale displacement and stress fields are also available and can be vizualized with gmsh (<a href="http://www.gmsh.info" target="_blank" rel="nofollow noreferrer noopener">www.gmsh.info</a>) using the mesh file <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/blob/main/MultiScale/FE2/Distributions/model.msh">model.msh</a>.</li> </ul> </li> </ul> <h2>Part 4: Reproduce paper<sup><a href="#fn-1">1</a></sup> figures</h2> <ul> <li>Fig. 7: The command to be run from the diretory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/TrainingPaths">TrainingPaths</a> is</li> </ul> <div> <pre><code> python3 countPathLength.py '../ConstRVE/PathsExamples'</code></pre> </div> <ul> <li>Fig. 9: The command to be run from the diretory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU/</a> is</li> </ul> <div> <pre><code> python3 PlotLoss_HiddenV.py</code></pre> </div> <ul> <li>Fig. 10: The command to be run from the diretory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU/</a> is</li> </ul> <div> <pre><code> python3 PlotLoss_3MRU.py</code></pre> </div> <ul> <li>Fig. 11: The command to be run from the diretory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU">SC_MRU/</a> is</li> </ul> <div> <pre><code> python3 PlotLoss_DiffNL_H120.py</code></pre> </div> <ul> <li>Figs. 12 and 13: The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a> is, see <a href="#vizualize-loss-evolution-and-testing-results">details</a></li> </ul> <div> <pre><code> python3 LossVS_insertN.py</code></pre> </div> <ul> <li>Figs. 14, 15 and 16: The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a></li> </ul> <div> <pre><code> python3 Plot_GS.py</code></pre> </div> <ul> <li>Figs. 17(b)(c)(d): The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/2D_MultiScale">MultiScale/2D_MultiScale/</a> is, see <a href="#vizualize-mutiscale-simulations-results">details</a>:</li> </ul> <div> <pre><code>python3 plot_force.py</code></pre> </div> <ul> <li>Figs. 18-23: Need gmsh to vizualize the results stored in <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale/FE2/Distributions">MultiScale/FE2/Distributions</a> and <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/MultiScale">MultiScale/S<code>CELL</code>_120_20</a>, see <a href="#vizualize-mutiscale-simulations-results">details</a></li> <li>Fig. A.24: The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a> is, see <a href="#vizualize-loss-evolution-and-testing-results">details</a></li> </ul> <div> <pre><code> python3 PlotLoss_MGRU.py.py</code></pre> </div> <ul> <li>Fig. B.25: The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a> is, see <a href="#vizualize-loss-evolution-and-testing-results">details</a></li> </ul> <div> <pre><code> python3 PlotLoss_NL.py</code></pre> </div> <ul> <li>Fig. B.26: The command to be run from the directory <a href="../didearot/didearotPublic/publicationsData/2024_scmru/-/tree/main/SC_MRU/CheckLoss">SC_MRU/CheckLoss/</a> is, see <a href="#vizualize-loss-evolution-and-testing-results">details</a></li> </ul> <div> <pre><code> python3 PlotLoss_DiffNL_H120.py</code></pre> </div> <h2>Disclaimer</h2> <p>This project has received funding from the European Union’s Horizon Europe Framework Programme under grant agreement No. 101056682 for the project “DIgital DEsign strategies to certify and mAnufacture Robust cOmposite sTructures (DIDEAROT)”. The contents of this publication are the sole responsibility of ULiege and do not necessarily reflect the opinion of the European Union. Neither the European Union nor the granting authority can be held responsible for them.</p> <ol> <li> <p>The work is described in:<br>"<em>Wu, L. and Noels, L. (2024).</em> <strong>Self-consistency Reinforced minimal Gated Recurrent Unit for surrogate modeling of history-dependent non-linear problems: application to history-dependent homogenized response of heterogeneous materials</strong> 424: 116881, <a href="https://doi.org/10.1016/j.cma.2024.116881" target="_blank" rel="nofollow noreferrer noopener">doi: 10.1016/j.cma.2024.116881</a>" which can be downloaded. We would be grateful if you could cite this publication in case you use the files. <a href="#fnref-1-5292">↩</a> <a href="#fnref-1-2">↩<sup>2</sup></a> <a href="#fnref-1-3">↩<sup>3</sup></a></p> </li> </ol>
TensorBoard runs for reinforcement learning on automated conjectured bounds on Laplacian spectral radius of graphs
<p>The zip archive contains separate folders with TensorBoard event files for the runs of our reinforcement learning implementation on the conjectured upper bounds for the Laplacian spectral radius, which are listed in Appendix B of the forthcoming paper: S. Al-Yakoob, M. Ghebleh, A. Kanso, D. Stevanović, Reinforcement learning for graph theory, I. Reimplementation of Wagner’s approach, Art Discrete Appl. Math. (2024).</p> <p>To view the contained graphs and the evolution of rewards, unzip the archive in a folder of your choice, and run in the terminal the command "tensorboard --logdir runs" from the parent folder of the unzipped "runs" folder.</p>
GIS web application for sustainable dike reinforcement solutions using nearby floodplains
<p>This is the dataset and instructions used for the GIS web application developed for the Dutch RAAKpubliek project Rivierwerken (Project nr.: RAAK.PUB09.018).</p>
Reinforcing RRI project reviews
<p>Collection of reviews of 203 Responsible Research and Innovation (RRI) related projects. Short reviews on structured templates including objectives, key results, Reinforcing relevant approches and tools, and link to the project web-site for further information. Files named after reviewing organization and reviewed project (acronyms).</p>
Recorded temperature data of a reinforced concrete, 1.2-m long, beam-slab model and associated meteorological data.
<div><strong>The data contain internal and surface temperature measurements of a 1.2 m-long, reinforced concrete girder-slab element exposed to the environmental conditions of the city of Bucaramanga, Colombia (Coordinates: 7°08'35", 73°07'18"). In addition, meteorological variables of the environment near the element were recorded. The data was measured from July 7, 2023, to November 6, 2023.</strong></div> <div><strong>The data were obtained by:</strong></div> <ul> <li> <div><strong>Internal Temperature (file name: InternalTemperature): 40 thermocouples were embedded in the element (Fig. 1) to collect data every 30 minutes from 6:30 AM on July 10, 2023 to 11:30 PM on November 6, 2023. Temperatures are in degrees Celsius (°C).</strong></div> </li> <li> <div><strong>Surface Temperature (file name: SurfaceTemperatureCameraEast and SurfaceTemperatureCameraWest): A FLIR E6 XT thermal camera was used to record Surface temperature of 10 points on the EAST (Fig. 2) and WEST (Fig. 3) faces of the element from 6:30 AM on June 29, 2023 to 4 PM on October 31, 2023. Twenty daily measurements were recorded every 30 minutes. Temperatures are in degrees Celsius (°C).</strong></div> </li> <li> <div><strong>Surface Temperature (file name: SurfaceTemperatureDroneEast): A DJI MAVIC 2 ENTERPRISE ADVANCED (EU) drone equipped with a thermal imaging camera measured the surface temperature of 10 points on the EAST side of the element (Fig. 4) every 30 minutes, from 6:30 AM on September 9 to 4 PM on October 31. Temperatures are in degrees Celsius (°C).</strong></div> </li> <li><strong>Meteorological variables (file name: WeatherStationData): A DAVIS VANTAGE PRO-2 WLRS weather station recorded data every 30 minutes from 12 AM on July 10, 2023, to 12:30 PM on November 6, 2023. The data included the average, maximum, and minimum ambient temperature (°C); relative humidity (%); speed (m/s) and wind direction (cardinal direction); atmospheric pressure (mb); precipitation (mm) and average and maximum solar radiation (W/m<sup>2</sup>).</strong></li> </ul>
Dataset for "Reinforcement Learning reveals fundamental limits on the mixing of active particles"
<p>Open access data set for manuscript "Reinforcement Learning reveals fundamental limits on the mixing of active particles" currently (2021) in preparation.</p>
The AI Economist: Taxation policy design via two-level deep reinforcement learning
<p>This dataset contains all raw experimental data for the paper "The AI Economist: Taxation Policy Design via Two-level Deep Multi-Agent Reinforcement Learning". </p> <p>The accompanying simulation, reinforcement learning, and data visualization code can be found at https://github.com/salesforce/ai-economist.</p> <p>For the one-step economy experiments, we provide:</p> <ul> <li> <p>training histories,</p> </li> <li> <p>configuration files (these experiments do not use phases), and</p> </li> <li> <p>final agent and planner models.</p> </li> </ul> <p>For the Gather-Trade-Build scenario, the data covers 6 spatial layouts: two Open-Quadrant (with 4 and 10 agents), and four Split-World maps with different configurations of the high-skilled and low-skilled agents. It also covers 4 tax policies (the AI Economist, Saez, free-market, and US federal). In addition, the AI Economist has been optimized for two social welfare functions: the product of equality and productivity, and inverse-income weighted utility. The Saez tax policy also uses estimated elasticities. </p> <p>Each experiment was repeated with different random seeds: 10 seeds for the Open-Quadrant scenarios, and 5 seeds for the Split-World scenarios. For each individual experiment, we provide: </p> <ul> <li> <p>Training histories (e.g. equality and productivity throughout training)</p> </li> <li> <p>the phase 1 and phase 2 configuration files, </p> </li> <li> <p>40 episode dense logs (the final 10 simulation logs across 4 environment replicas),</p> </li> <li> <p>phase 1 final agent models, and</p> </li> <li> <p>phase 2 final agent and planner models.</p> </li> </ul> <p>Finally, we include all data and results used to calibrate the Saez elasticity estimates and to estimate elasticity directly from a sweep over flat-rate tax policies:</p> <ul> <li> <p>training histories,</p> </li> <li> <p>the phase 1 and phase 2 configuration files, </p> </li> <li> <p>phase 1 final agent models, and</p> </li> <li> <p>phase 2 final agent and planner models.</p> </li> </ul>
Multifunctional corrosion control system as a sustainable approach for reinforced concrete elements
<p>Corrosion inhibitors can be utilized to decrease the corrosion kinetics and therefore increase the durability of reinforced concrete structures. Recently, a green synthesized organic compound, 1-benzyl-4-phenyl-1H-1,2,3-triazole (BPT), was shown to be a successful green organic corrosion inhibitor for mild steel. Studies suggested that the BPT adsorbs chemically onto the steel and acts as a mixed inhibitor, suppressing both the anodic and cathodic corrosion kinetics of steel. In addition, microcapsules have shown to be an efficient way for a controlled inhibitor release in reinforced concrete structures. On the other hand, geopolymers (GPs) comprised of a long range of covalently bonded alumino-silicates, with amorphous network structure are generally considered as a suitable substitute for OPC for many structural applications due to their high strength and durability. The use of recycled waste materials or natural abundant materials for the production of GPs has attracted world-wide attention as it presents an environment-friendly aspect that may shed light on for replacing traditional OPC by its sustainability. One of the advantages of GPs is the significant reduction to CO2 emission due to the energy consumption, the geopolymers utilize materials such as fly-ash, which is a byproduct of coal combustion, or natural precursor materials (clays, basalt rocks, etc.) and their derives (metakaolin), which does not produce net CO2 emission. Recent studies have shown that GPs based cements can hinder the corrosion of reinforcement steel in concrete structures when compared to OPC, mostly because of lower chloride ingress (due to barrier protective capabilities) and the highly alkaline pH nature of geopolymer cements.</p>
Deep Reinforcement Learning-based Project Prioritization for Rapid Post-Disaster Recovery of Transportation Infrastructure Systems
<p>Among various natural hazards that threaten transportation infrastructure, flooding represents a major hazard in Region 6's states to roadways as it challenges their design, operation, efficiency, and safety. The catastrophic flooding disaster event generally leads to massive obstruction of traffic, direct damage to highway/bridge structures/pavement, and indirect damages to economic activities and regional communities that may cause loss of many lives. After disasters strike, reconstruction and maintenance of an enormous number of damaged transportation infrastructure systems require each DOT to take extremely expensive and long-term processes. In addition, planning and organizing post-disaster reconstruction and maintenance projects of transportation infrastructures are extremely challenging for each DOT because they entail a massive number and the broad areas of the projects with various considerable factors and multi-objective issues including social, economic, political, and technical factors. Yet, amazingly, a comprehensive, integrated, data-driven approach for organizing and prioritizing post-disaster transportation reconstruction projects remains elusive. In addition, DOTs in Region 6 still need to improve the current practice and systems to robustly identify and accurately predict the detailed factors and their impacts affecting post-disaster transportation recovery. The main objective of this proposed research is to develop a deep reinforcement learning-based project prioritization system for rapid post-disaster reconstruction and recovery of damaged transportation infrastructure systems. This project also aims to provide a means to facilitate the systematic optimization and prioritization of the post-disaster reconstruction and maintenance plan of transportation infrastructure by focusing on social, economic, and technical aspects. The outcomes from this project would help engineers and decision-makers in Region 6's State DOTs optimize and sequence transportation recovery processes at a regional network level with necessary recovery factors and evaluating its long-term impacts after disasters.</p>
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.
Allen Brain Atlas
Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.
Annotated Behaviour and Observability Dataset (ABODe)
ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.
DANDI Archive for NWB datasets
DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.
International Brain Laboratory public data
The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.
OpenNeuro
OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.