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.
3,339
datasets available to search
ShareScore release 0.9.0
Dataset results
3,339 results for “heterogeneities”
Spatial data for Chaplin-Kramer et al. "Spatial heterogeneity in forest carbon storage affects priorities for reforestation"
<p>Spatial data generated for the forest carbon edge effects analysis and optimization performed in Chaplin-Kramer et al. "Spatial heterogeneity in forest carbon storage affects priorities for reforestation". The files included are described below.</p><p>fc_stack_hansen_forest_cover2014_compressed_std_forest_edge_result.tif - Carbon (Mg/ha) predicted by the regression model for Hansen forest cover 2014 (this is the LULC map that produced the model with the best fit and was used in the model validation)</p><p>fig1_edge_effect_hansen_compressed_nr_md5_63366a.tif – Figure 1 in the manuscript</p><p>fig2_ipcc1_regression2_overlap3_md5_43e130.tif – Figure 2 in the manuscript</p><p>figS5_ipcc1_regression2_overlap3_avgov_md5_fe8b72.tif – Alternate styling of manuscript Fig. 2 (resampling with "average" instead of "near"), shown in Supplemental Fig. S5</p><p>ipcc_carbon_restoration_limited_md5_cd012d.tif - Carbon (Mg/ha) in restoration LULC calculated with the IPCC Tier 1 approach</p><p>ipcc_tier1_carbon_data.zip - Data needed to run the IPCC approach are contained within two files in the zip folder: the carbon table (carbon zones are rows and ESA LULC codes are columns) IPCC_carbon_table_md5_a91f7ade46871575861005764d85cfa7.csv; and the carbon zones map: carbon_zones_md5_aa16830f64d1ef66ebdf2552fb8a9c0d.gpkg</p><p>ipcc_carbon_esa_compressed_md5_5b4803.tif - Carbon (Mg/ha) in ESA 2014 LULC calculated with the IPCC Tier 1 approach (shown in Fig. S1)</p><p>PNV_jsmith_060420_md5_8dd464e0e23fefaaabe52e44aa296330.tif – Potential Natural Vegetation (PNV) created by Jeff Smith (unpublished) through a cross-walk of ESA to Dinerstein biomes</p><p>regression_carbon_esa_compressed_md5_c867a0.tif - Carbon (Mg/ha) in ESA 2014 LULC calculated with the regression approach (shown in Fig. S1)</p><p>regression_carbon_restoration_md5_1f5ca8.tif - Carbon (Mg/ha) in restoration LULC calculated with the regression approach</p><p>restoration_limited_md5_372bdfd9ffaf810b5f68ddeb4704f48f.tif – the LULC scenario of reforestation (based on PNV, but excluding agriculture and urban areas) </p>
ChromoPhyloGen: characterizing copy number alteration patterns in heterogeneous tumor cell populations at Single-Cell Resolution
<p>The human liver cancer cell line Huh7 was obtained from the American Type Culture Collection (ATCC). Huh7 cells were cultivated in Dulbecco's Modified Eagle Medium (DMEM, Gibco, C11995), supplemented with 1% penicillin/streptomycin (Gibco, 15140122), and 10% fetal bovine serum (FBS, Excell, FSP500). Huh7 cell line was maintained under a 95% O2 and 5% CO2 humidified atmosphere in an incubator at 37˚C. </p> <p>The scDNA-seq library was performed using the Chromium Single cell DNA Library & Gel Bead kit (10x Genomics, PN1000040) in combination with the Chromium instrument. The samples were processed on Chromium Single cell Chip C and D (10x Genomics, 1000022 and 1000042, respectively) according to the manufacturer's user guide and subsequently run on a thermocycler. The barcoded libraries were sequenced using the Novaseq 6000 300 cycle high-output flow cells.</p> <p>The scRNA-seq library was generated using the 10x Genomics Chromium Single Cell 3' & Gel Bead Kit v3 (10x Genomics, PN100075) in combination with the Chromium instrument. The samples were processed on Chromium Single cell Chip B (10x Genomics,1000154) according to the manufacturer's protocol and subsequently run on a thermocycler. The 3' gene expression libraries were sequenced using the Novaseq 6000 300 cycle high-output flow cells.</p> <p> </p>
Tree diversity across multiple scales and environmental heterogeneity promote ecosystem multifunctionality in a large temperate forest region
<p><strong>Aim</strong>: Biodiversity across different scales provides multidimensional insurance for ecosystem functioning. Although the positive effects of local scale (α-diversity) biodiversity on ecosystem multifunctionality are widely accepted, species turnover across communities (β-diversity) which is often an important driver of ecosystem functioning did not receive the same attention. This study broadens the understanding of how multiple attributes of biodiversity maintain ecosystem multifunctionality from local to regional scales, across diverse environmental gradients.</p> <p><strong>Location</strong>: North-eastern China.</p> <p><strong>Time period</strong>: 2017.</p> <p><strong>Major taxa studied</strong>: Woody plants.</p> <p><strong>Methods</strong>: We estimate ecosystem multifunctionality using both averaging and modified multiple thresholds (50%, 70% and 90%) approaches. Multiple dimensions of biodiversity across varying spatial scales were measured within the framework of Hill‒Chao numbers. Linear and nonlinear models were used to evaluate the optimal patterns of multifunctionality and biodiversity along the latitude. Using variance decomposition, structural equation modeling and linear mixed models, we explored how multiple attributes of tree diversity at varying spatial scales affect multifunctionality, and how these relationships are modulated by environmental drivers.</p> <p><strong>Results</strong>: Our results show that multifunctionality decreased with increasing latitude, mirroring the pattern of tree diversity along latitudinal gradients. Phylogenetic β-diversity and species α-diversity emerged as crucial diversity indices for sustaining multifunctionality in these temperate forests. Soil and climatic conditions had either direct effects on multifunctionality, or indirect ones mediated by tree diversity. Environmental heterogeneity played a pivotal role in maintaining high levels of multifunctionality, exerting influence both directly and indirectly via phylogenetic β-diversity.</p> <p><strong>Main conclusions</strong>: This study underscores the positive effects of biodiversity on multifunctionality across multiple dimensions. Based on our findings, we conclude that any design of a forested landscape that is aimed at maximizing multifunctionality should consider maintaining high local diversity as well as forest community heterogeneity at varying scales.</p>
Data sets for phylogenomic analyses in: Ant backbone phylogeny resolved by modelling compositional heterogeneity among sites in genomic data
<p>Ants are the most ubiquitous and ecologically dominant arthropods on Earth, and understanding their phylogeny is crucial for deciphering their character evolution, species diversification, and biogeography. Although recent genomic data have shown promise in clarifying intrafamilial relationships across the tree of ants, inconsistencies between molecular datasets have also emerged. Here I re-examine the most comprehensive published Sanger-sequencing and genome-scale datasets of ants using model comparison methods that model among-site compositional heterogeneity to understand the sources of conflict in phylogenetic studies. My results under the best-fitting model, selected on the basis of Bayesian cross-validation and posterior predictive model checking, identify contentious nodes in ant phylogeny whose resolution is <a>modelling-dependent. </a>I show that the Bayesian infinite mixture CAT model outperforms empirical finite mixture models (C20, C40 and C60) and that, under the best-fitting CAT-GTR+G4 model, the enigmatic <a><em>Martialis</em> </a><em>heureka</em> is sister to all ants except Leptanillinae, rejecting the more popular hypothesis supported under worse-fitting models, that place it as sister to Leptanillinae. These analyses resolve a lasting controversy in ant phylogeny and highlight the significance of model comparison and adequate modelling of among-site compositional heterogeneity in reconstructing the deep phylogeny of insects.</p>
High quality figures of "An Unstructured Mesh Generation Tool for Efficient High-Resolution Representation of Spatial Heterogeneity in Land Surface Models"
<p>High quality figures of "An Unstructured Mesh Generation Tool for Efficient High-Resolution Representation of Spatial Heterogeneity in Land Surface Models"</p>
Data from: Heterogeneity promotes resilience in restored prairie: implications for the 'environmental heterogeneity hypothesis'
<p>Enhancing resilience in formerly degraded ecosystems is an important goal of restoration ecology. However, evidence for the recovery of resilience and its underlying mechanisms requires long-term experiments and comparison to reference ecosystems. We used data from an experimental prairie restoration that featured long-term soil heterogeneity manipulations and data from comparable remnant (reference) prairie to (1) quantify the recovery of ecosystem functioning (i.e., productivity) relative to remnant prairie, (2) compare resilience of restored and remnant prairies to a natural drought, and (3) test whether soil heterogeneity enhances resilience of restored prairie. We compared sensitivity and legacy effects between prairie types (remnant and restored) and among four prairie sites that included two remnant prairie sites and prairie restored under homogeneous and heterogeneous soil conditions. We measured sensitivity and resilience as the proportional change in aboveground net primary productivity (ANPP) during and following drought (sensitivity and legacy effects, respectively) relative to average ANPP based on four pre-drought years (2014-2017). In non-drought years, total ANPP was similar between remnant and restored prairie, but remnant prairie had higher grass productivity and lower forb productivity compared to restored prairie. These ANPP patterns generally persisted during drought. Sensitivity of total ANPP to drought was similar between restored and remnant prairie, but grasses in restored prairie were more sensitive to drought. Post-drought legacy effects were more positive in restored prairie, and we attributed this to the more positive and less variable legacy response of forb ANPP in restored prairie, especially in the heterogeneous soil treatment. Our results suggest that productivity recovers in restored prairie and exhibits similar sensitivity to drought as remnant prairie. Furthermore, imparting heterogeneity promotes forb productivity and enhances prairie resilience to drought.</p>
Data from: Born in heterogenous landscapes: birth timing, body mass and growth of roe deer (Capreolus capreolus) fawns in contrasting habitats
<p>Although the widespread effects of global change impact almost all ecosystems, we lack a detailed understanding of how wildlife that thrive in human-dominated environments are able to adjust their life history to modifications in land use of their natural habitat. In particular, spatial variation in environmental conditions is predicted to influence development during the crucial early life phase, with marked impacts on individual performance and population dynamics for long-lived species. Large herbivores such as roe deer (<em>Capreolus capreolus</em>), a synanthropic species, have increased substantially in number and distribution over the last half century across Europe. Roe deer have been particularly successful, gradually colonizing agricultural landscapes to cope with a global warming-driven phenological mismatch in their natural forest habitat. However, to date, little is known about how habitat heterogeneity impacts their demographic performance in this heavily human-impacted environment. Specifically, we predicted that fawns born in predominantly cultivated local habitats would achieve faster early development due to the food subsidies obtained by their mothers from agricultural crops. Contrary to our expectations, fawns in semi-natural forest habitats were around 10% heavier at birth than those born in more mixed (by 0.163 ± 0.058 kg) and open (by 0.169 ± 0.006 kg) agricultural habitats. However, all fawns subsequently grew at a similar average rate (0.148 ± 0.058 kg/day), irrespective of their habitat. This habitat-dependent variation in birth mass appeared to be driven by reproductive phenology, as i) early-born fawns were heavier than late-born fawns, and ii) mothers living in the forest gave birth around 10 days earlier than those living in the mixed and open habitats. As natural habitats become increasingly scarce and fragmented due to the activities of humans, the prospects for many wild populations will depend on their ability to subsist in the heavily modified habitats of anthropogenic landscapes.</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>
The impact of market integration on the digital divide: An analysis based on the heterogeneous effect of innovation resource synergy
<p>Resource market reforms have provided opportunities for reducing the digital divide (DD). In this study, we analyzed the theoretical mechanism of the impact of market integration (MI) on DD and the intermediary role of innovation resource (IR) synergy. Data from 271 cities in China were collected, and the DEA–Malmquist index and Thiel index were utilized to quantify DD among these cities during 2011–2019. The intermediary effect and SDM models were adopted for an empirical analysis. The results showed that (1) MI had a negative effect on DD in addition to significant spatial spillover effects. MI widened DD in the local city but narrowed the same in the adjacent cities. (2) IR synergy exerted positive intermediary effects, which were heterogeneous. Through human resource synergy, MI narrowed DD, through knowledge synergy, MI widened DD, and through the R&D capital synergy, MI did not significantly influence DD. (3) There was no significant spatial spillover effect of the intermediary role of IR synergy.</p>
Soil properties shape the heterogeneity of denitrification and N2O emissions across large-scale flooded paddy soils
<p>The dataset's contents include the relative abundance of different denitrifier types at the genus level across eight different flooded paddy soils. This dataset categorizes denitrifier types in its first column labeled 'Type.' The second column, 'Genus,' associates each type with various species at the genus level. Subsequent columns present the relative abundance of these species in diverse sampling locations. Researchers can leverage this dataset to derive more patterns regarding the differences within specific dominant denitrifier species. No ethical considerations are pertinent to the utilization of this dataset.</p>
A Large Fault Partially Reactivated During Two Contiguous Seismic Sequences in Central Italy: The Role of Geometrical and Frictional Heterogeneities
<p>Moment tensor catalog for events with M > 3.0, that occurred between January 2009 and April 2021, in Campotosto area, Italy. Moments tensor were calculated by applying the Time Domain Moment Tensor technique, originally proposed by Dreger and Helmberger (1993) and Pasyanos et al. (1996) and successively implemented at INGV by Scognamiglio et al. (2009).</p> <p>The catalog includes:</p> <p>Location of events: time, depth, lat and lon</p> <p>The moment magnitude: Mw</p> <p>The double-couple value: DC</p> <p>The Variance Reduction value: VR</p> <p>The six moment tensor components: Mxx, Mxy, Mxz, Myy, Myz, Mzz</p> <p>The orientation of nodal planes: strike1, dip1, rake1, strike2, dip2, rake2</p> <p> </p>
Minimal dataset for the manuscript "Better together against genetic heterogeneity: a sex-combined joint main and interaction analysis of 290 quantitative traits in the UK Biobank".
<p>Dataset "lin2024-sex_combined_interaction-association_signifincant_in_one_or_more_tests-summary.txt" is a minimal dataset to reproduce the figures and tables in the manuscript "Better together against genetic heterogeneity: a sex-combined joint main and interaction analysis of 290 quantitative traits in the UK Biobank". </p> <p><br>To generate this dataset, see "https://github.com/BoxiLin/t2meta" Steps 0, 1.</p> <p>This dataset is the input for Steps 2, 3, 4, 5 to generate Figures 1-3 and Table 2-3.</p> <p> </p> <p>##### Column information ########################</p> <p>The following columns are annotations on each variant in the GWAS, calculated across the analysis subset of 361,194 samples by the Neale lab:</p> <p>code: Phenotype identifier in the form of "[UKB Data field]_raw"<br>variant: Unique variant identifier in the form "chr:pos:ref:alt", where "ref" is aligned to the forward strand.<br>chr: Chromosome of the variant.<br>pos: Position of the variant in GRCh37 coordinates.<br>rsid: rs ID<br>ref: Reference allele on the forward strand.<br>alt: Alternate allele (not necessarily minor allele).<br>p_hwe: Hardy-Weinberg p-value.<br>info: Imputation INFO score as provided by UK Biobank.</p> <p> </p> <p>The following columns are sex-stratified test statistics calculated by the Neale lab:</p> <p>minor_allele.x: Minor allele (AF < 0.5) in the female GWAS <br>minor_AF.x: Minor allele frequency in the female GWAS <br>beta.x: Estimated effect size of alt allele in the female GWAS <br>se.x: Estimated standard error of beta in the female GWAS<br>tstat.x: t-statistic of beta estimate (= beta/se) in the female GWAS <br>pval.x: p-value of beta significance test in the female GWAS </p> <p>minor_allele.y: Minor allele (AF < 0.5) in the male GWAS <br>minor_AF.y: Minor allele frequency in the male GWAS <br>beta.y: Estimated effect size of alt allele in the male GWAS <br>se.y: Estimated standard error of beta in the male GWAS <br>tstat.y: t-statistic of beta estimate (= beta/se) in the male GWAS <br>pval.y: p-value of beta significance test in the male GWAS </p> <p> </p> <p><br>The following columns are sex-combined test statistics calculated in our analysis:</p> <p>T.I: test statsitic for interaction effect-only <br>p.T.I: p-value of the interaction effect-only test <br>TSG.L: test statsitic for inverse variance weighted meta-analysis<br>p.TSG.L: p-value of the inverse variance weighted meta-analysis<br>TSG.Q: test statsitic for the omnibus meta-analysis<br>p.TSG.Q: p-value for the omnibus meta-analysis</p>
Data from: Landscape heterogeneity drives population structure in four western bumble bee species
<p>Bumble bees are critical pollinators in wild, agricultural, and urban ecosystems—providing the necessary ecological services for food and crop production. In western North America, mountain ranges have high bumble bee species richness. However, as climate change increases temperatures and restricts montane populations to higher elevational spaces, their ability to disperse and maintain genetic diversity decreases. This genetic isolation could lead to the extirpation of local pollinator communities and an overall loss of pollinators. We analyzed the genetic diversity of four broadly sympatric species of bumble bees across the Rocky and Cascade Mountains of western North America to assess habitat isolation's impact on population genetic structure. We expected species restricted to higher elevation habitats to display higher population structure and less genetic diversity across the landscape. We sampled approximately 150 bees per species from seven to eight sites across each species' range. We genotyped bees with an average of 10 loci and used FST and Bayesian Structure analysis to quantify population differentiation. Using isolation by distance and isolation by resistance analyses, species with both narrow and broad habitat suitability requirements showed evidence of habitat suitability restricting gene flow. Although each species showed varying degrees of genetic structure and gene flow, knowing how habitat heterogeneity drives genetic structure and isolation can help guide conservation efforts and determine focal regions for bumble bee conservation in the face of climate change.</p>
Data for: Viral receptor-binding protein evolves new function through mutations that cause trimer instability and functional heterogeneity
<p>When proteins evolve new activity, a concomitant decrease in stability is often observed because the mutations that confer new activity can destabilize the native fold. In the conventional model of protein evolution, reduced stability is considered a purely deleterious cost of molecular innovation because unstable proteins are prone to aggregation and are sensitive to environmental stressors. However, recent work has revealed that non-native, often unstable protein conformations play an important role in mediating evolutionary transitions, raising the question of whether instability can itself potentiate the evolution of new activity. We explored this question in a bacteriophage receptor binding protein (RBP) during host-range evolution. We studied the properties of the RBP of bacteriophage before and after host-range evolution and demonstrated that the evolved protein is relatively unstable and may exist in multiple conformations with unique receptor preferences. Through a combination of structural modeling and in vitro oligomeric state analysis, we found that the instability arises from mutations that interfere with trimer formation. This study raises the intriguing possibility that protein instability might play a previously unrecognized role in mediating host-range expansions in viruses.</p>
Landscape heterogeneity increases bird functional diversity within Neotropical vineyards
<p>Conversion of lands to agroecosystems has resulted in a decline in bird biodiversity. Analyzing functional diversity is a central tool for detecting changes in the ecological functions performed by birds in these landscapes. This paper aims to investigate the responses of bird taxonomic and functional diversity to landscape heterogeneity and native forest cover in Neotropical vineyards. We sampled 19 vineyard landscapes in southeastern Brazil. These landscapes covered a gradient of forest cover and heterogeneity resulting from various land uses. To assess bird diversity, we considered both taxonomic diversity and functional diversity (i.e., functional richness, evenness, and divergence). To examine the potential interactions between landscapes and bird assemblages, we employed generalized linear models (GLM). Taxonomic diversity showed no correlation with any landscape metrics. On the other hand, variation in the three metrics of functional diversity was related to landscape heterogeneity. However, in heterogeneous landscapes, these communities can be structured by limiting similarity processes. We highlight the impact of landscape homogenization on the ecological functions performed by birds in vineyards while finding no significant effect on species diversity. These findings can provide valuable support for the formulation of public policies aimed at striking a balance between agricultural production and biodiversity conservation.</p>
Model Datafiles for Observed seasonal changes in Martian hydrogen chloride consistent with heterogeneous chemistry on atmospheric dust and ice
<p>Datafiles produced by the 1-D photochemistry model used to create the publication "Observed seasonal changes in Martian hydrogen chloride consistent with heterogeneous chemistry on atmospheric dust and ice" - Taysum et al. 2024.</p> <p> </p> <p>matching_orbits_v2.txt : Lists the orbital parameters and water vapour / aerosol file names corresponding to each of the 77 ACS MIR HCl observations that we study.</p> <p> </p> <p><strong>MCD_GlobaMaps.zip</strong> : .nc files containing model output where the model is ran at Longitude 0 degrees, across Ls 0 --> 360 in intervals of 30 degrees, and latitudes 60 S to 60 N in intervals of 15 degrees.</p> <ul> <li>LowTau : model runs with the chlorine heterogeneous chemistry scheme active. <ul> <li>MY34/ : Runs driven with MY34 climatology from the MCDv6.1</li> <li>StandardClim/ : Runs driven with Standard Climatology from the MCDv6.</li> </ul> </li> <li>No_Chlorine : model runs with no chlorine tracers in the model chemistry. <ul> <li>MY34/ : Runs driven with MY34 climatology from the MCDv6.1</li> <li>StandardClim/ : Runs driven with Standard Climatology from the MCDv6.1</li> <li> </li> </ul> </li> </ul> <p><strong>ACS_Comparisons.zip </strong>: .nc files containing the model output for specific ACS MIR HCl observations in Mars Year 34.</p> <ul> <li>LowTau : model runs with the chlorine heterogeneous chemistry scheme active. <ul> <li>MY34/ : Runs driven with MY34 climatology from the MCDv6.1</li> <li>StandardClim/ : Runs driven with Standard Climatology from the MCDv6.1</li> <li>TGO/ : Runs driven with MY34 climatology, and TGO measured water ice, dust, and H2O vapor profiles, driven at the position of the ACS-TIRVIM aerosol retrievals corresponding to the approximately colocated ACS MIR HCl observation</li> </ul> </li> <li>No_Chlorine : model runs with no chlorine tracers in the model chemistry. <ul> <li>MY34/ : Runs driven with MY34 climatology from the MCDv6.1</li> <li>StandardClim/ : Runs driven with Standard Climatology from the MCDv6.1</li> <li>TGO/ : Runs driven with MY34 climatology, driven at the position of the ACS-TIRVIM aerosol retrievals corresponding to the approximately colocated ACS MIR HCl observation</li> </ul> </li> </ul>
Data from the paper "Porphyrin central metal ion driven self-assembling in heterogeneous ZnTPP – CoTPP films grown on Fe(001)-p(1 × 1)O"
Open the record for dataset details and reuse information.
Not all Heterogeneity is Equal: Length Scale of Frictional Property Variation as a control on Subduction Megathrust Sliding Behavior
<p>The folder SSS_Results.zip contains the following files:</p> <p>1) A .txt for each simulation that contains (1st column) simulated timestep in seconds; (2nd column) maximum slip velocity at each timestep in m/s; and (3rd column) maximum slip velocity taken from the center of the velocity-strengthening blocks in m/s at each time step.</p> <p>2) A .mat file for each simulation that contains the same output as the .txt file, and in addition a MATLAB structure 'p' that serves as an input file to run each simulation using RSFaultZ, found at <a href="https://github.com/rmskarbek/RSFaultZ">https://github.com/rmskarbek/RSFaultZ</a></p> <p>3) PartialStabilityData.mat - results from linear stability analysis.</p> <p>4) ReducedArchiveData.m, SSS_Fig3.m, SSS_Fig4.m - Three MATLAB scripts that will automatically generate figures 3 and 4 in the related paper. </p> <p>NOTE: SSS_Fig3.m uses a perceptually uniform color map that requires an additional MATLAB package: <a href="https://www.mathworks.com/matlabcentral/fileexchange/68546-crameri-perceptually-uniform-scientific-colormaps">https://www.mathworks.com/matlabcentral/fileexchange/68546-crameri-perceptually-uniform-scientific-colormaps</a></p> <p>For any additional information please do not hesitate to contact the corresponding author Rob Skarbek at rskarbek@psi.edu.</p> <p>Additional MATLAB scripts that will reproduce the simulations themselves can be found at <a href="https://github.com/rmskarbek/RSFaultZ/tree/main/examples/blocks">https://github.com/rmskarbek/RSFaultZ/tree/main/examples/blocks</a></p>
Role of individual and population heterogeneity in shaping dynamics of multi-pathogen shedding in an island endemic bat
<p>Dataset for the manuscript "<strong>Role of individual and population heterogeneity in shaping dynamics of multi-pathogen shedding in an island endemic bat</strong>", including information of the bat samples and infection data.</p>
Spike-MD Data for "CryoBench: Diverse and challenging datasets for the heterogeneity problem in cryo-EM"
<p>A synthetic cryo-EM dataset with heterogeneity from a molecular dynamics simulation and ground truth atomic models, poses, labels, mask, and consensus volume:</p> <ul> <li>Spike-MD: 100k particle images (256x256, 1.5A/pix) of SARS-CoV-2 spike protein with conformations sampled from a molecular dynamics simulation from Wieczór et al. (2023)</li> <li>sampled_pdbs.xtc: atomic models stitched into a trajectory</li> <li>seed_structure.pdb: initial seed structure for .xtc trajectory</li> </ul>
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.