Find research datasets worth reusing

opencc-by-4.0Mar 2022View details →

Diverse Topologies for Evaluation of Geometric Similarity Metrics

A collection of 7 datasets with each set containing 3D shapes with varying topological complexity. The datasets can be used to compare different metrics of geometric dissimilarity. Two of the datasets have topologically complex shapes that resemble designs obtained from topology optimization, a widely used design optimization method for engineering structures. We used this dataset for a related journal article with the following abstract: "In the early stages of engineering design, multitudes of feasible designs can be generated using structural optimization methods by varying the design requirements or user preferences for different performance objectives. Data mining such potentially large datasets is a challenging task. An unsupervised data-centric approach for exploring designs is to find clusters of similar designs and recommend only the cluster representatives for review. Design similarity can be defined not only on a purely functional level but also based on geometric properties, such as size, shape, and topology. While metrics such as chamfer distance measure the geometrical differences intuitively, it is more useful for design exploration to use metrics based on geometric features, which are extracted from high-dimensional 3D geometric data using dimensionality reduction techniques. If the Euclidean distance in the geometric features is meaningful, the features can be combined with performance attributes resulting in an aggregate feature vector that can potentially be useful in design exploration based on both geometry and performance. We propose a novel approach to evaluate such derived metrics by measuring their similarity with the metrics commonly used in 3D object classification. Furthermore, we measure clustering accuracy, which is a state-of-the-art unsupervised approach to evaluate metrics. For this purpose, we use a labeled, synthetic dataset with topologically complex designs. From our results, we conclude that Pointcloud Autoencoder is promising in encoding geometric features and developing a comprehensive design exploration method." For each dataset, shapes/designs are saved as surface mesh files (extension: stl) and point cloud files (extension: ply) in the folders "stls" and "plys" respectively. A brief description of the 7 different datasets is in the following table. For each dataset, the designs are named using numbers starting from 0, e.g., “0.stl, 1.stl, …, 19.stl” in the folder for the surface mesh files. Some of the datasets are labeled, i.e., each design belongs to a class. In a labeled dataset, all classes have the same number of designs, and the designs are named in the order of their class. For example, a labeled dataset with 4 designs and 2 classes contains files whose names start with {0, 1, 2, 3} where the designs {0, 1} belong to class 1, and {2, 3} belong to class 2. <table> <thead> <tr> <th scope="col">Dataset name</th> <th scope="col">Directory name</th> <th scope="col">Number of designs</th> <th scope="col">Number of classes</th> </tr> </thead> <tbody> <tr> <td>Beam-rotation</td> <td>"rotate_beam"</td> <td>20</td> <td>None</td> </tr> <tr> <td>Beam-elongation</td> <td>"elongate_beam"</td> <td>20</td> <td>None</td> </tr> <tr> <td>Beam-translation</td> <td>"move_beam"</td> <td>20</td> <td>None</td> </tr> <tr> <td>Three cube trusses</td> <td>"three_cube_truss"</td> <td>150</td> <td>6</td> </tr> <tr> <td>Single cube trusses</td> <td>"single_cube_truss"</td> <td>275</td> <td>11</td> </tr> <tr> <td>Random topologies</td> <td>"three_cube_truss_random"</td> <td>1000</td> <td>50</td> </tr> <tr> <td>Topologically optimized designs</td> <td>"cube_opt_shapes"</td> <td>1500</td> <td>None</td> </tr> </tbody> </table>

opencc-by-4.0May 2022View details →

Source data belonged to "Geometric flow control in lateral flow assays: Macroscopic single-phase modeling"

This record contains all the necessary data to obtain the results of the study "Geometric flow control in lateral flow assays: Macroscopic single-phase modeling" (<a href="https://doi.org/10.1063/5.0093316">https://doi.org/10.1063/5.0093316</a>).

opencc-by-4.0Apr 2022View details →

Dataset for Direct Geometric Probe of Singularities in Band Structure

Included here is the processed data illustrated in the figures of both the main text, and the supplemental material. Below is a description of each file's contents.   Figure2Dcode.m contains the MATLAB code that generates Figure 2D of the main text. It takes the band populations inferred from five iterations of measurements, and calculates the means and standard errors for data taken at each theta as defined in the main text.   Figure2Ddata.csv contains the data illustrated in Figure 2D of the main text. The data provided are normalized band populations, such that the value 1 corresponds to the entire atom number in the sample. The rows provide the band index; the first row of data corresponds to the n=1 band, the second row corresponds to the n=2 band, etc. The columns provide the measured turning angle in units of radians; the first column corresponds to a turning angle of zero, and the angle is incremented by pi/12 radians for each column that follows. Row 5 is the error for the n=1 population, row 6 is the error for the n=2 population, and row 7 is the error on the sum of the population in bands with index not equal to 1 or 2.   Figure3Bcode.ipynb contains the jupyter notebook that generates Figure 3B of the main text. It takes the band populations inferred from four iterations of measurements, and calculates the means and standard errors for data taken for each intermediate point along K - M - K'. For this plot, the x-axis is chosen to be the intermediate quasi-momenta, and different colors are used to differentiate between different acceleration times.   Figure3Bdata.csv contains the data illustrated in Figure 3B of the main text. The five columns correspond to the five different trajectory evolution times (0.5, 0.9, 1.3, 1.7, 2.1 milliseconds) shown in the Figure 3B. The first nine rows correspond to the nine trajectory midpoint positions in the Brillouin zone, as showed in Figure 3A; the first row corresponds to a midpoint at K. The next nine rows are the errors on the measurements.   Figure4Ccode.m contains the MATLAB code that generates Figure 4C of the main text. It takes the band populations inferred from twelve iterations of measurements, each at a different theta as defined in the main text, and calculate the means and standard errors for data taken at each theta.   Figure4Cdata.csv contains the data illustrated in Figure 4C of the main text. The data provided are normalized band populations, such that the value 1 corresponds to the entire atom number in the sample. The rows provide the band index; the first row of data corresponds to the n=1 band, the second row corresponds to the n=2 band, etc. The columns provide the measured turning angle in units of radians; the first column corresponds to a turning angle of zero, and the angle is incremented by pi/6 radians for each column that follows. Row 11 is the error for the n=3 population, row 12 is the error for the n=4 population, and row 13 is the error on the sum of the population in bands with index not equal to 3 or 4.   FigureS3Bcode.m contains the MATLAB code that generates Figure S3B of the main text. It takes the band populations inferred from seven iterations of measurements, and calculates the means and standard errors for data taken for each hold time at quasi-momentum Q as defined in the main text. The result is then fitted to a sine with exponentially decaying envelope. push_ramp.py (in Full Hamiltonian simulation.zip) starts with an initial state and evolves it according to the discretized schrödinger equation along the path in q-space. The Hamiltonian is calculated in basic_fcts.py. The final state is projected on the eigenstates at the final q to extract the band population. Different time intervals are used to obtain all the data. A decay to account for coherence loss is added.   FigureS3data.csv contains the data illustrated in Figure 3 of the supplementary material. The first row is the data values, and the second row are the error bars.   FigureS4Bcode.m contains the MATLAB code that generates Figure S4B of the main text. It takes the band populations inferred from four iterations of measurements, and calculates the means and standard errors for data taken for each intermediate point along K - M - K'. For this plot, the x-axis is chosen to be acceleration time, and different colors are used to differentiate between different intermediate points. push_ramp.py (in Full Hamiltonian simulation.zip) starts with an initial state and evolves it according to the discretized schrödinger equation along the paths in q-space. The Hamiltonian is calculated in basic_fcts.py. The final state is projected on the eigenstates at the final q to extract the band population. Different time intervals are used to obtain all the data.   FigureS4data.csv contains the data illustrated in Figure 4 of the supplementary material. The first five rows are the normalized ground band population for five different trajectory mid points on the K - M - K' line of the Brillouin zone.; the first row is for a midpoint at K, and the fifth row is for a midpoint at M. The columns give the trajectory traversal times; the first column corresponds to a traversal time of 0.1 ms and each column corresponds to a new traversal time incremented by 0.2 ms. Rows 6-10 are the error bars for the measurements.   FigureS5Bcode.zip contains the codes that generate Figure S5B of the main text. For each subplots in Fig.S5B, the corresponding MATLAB code in the zip file takes the band populations inferred from three iterations of measurements, and calculate the means and standard errors for data taken at each acceleration time.   FigureS5Bdata.csv contains the data illustrated in Figure 5B of the supplementary material. Rows 1-20 correspond to subpanel (iii) in the Figure S5 of the supplementary material.  Rows 21-40 correspond to subpanel (ii) in the Figure S5 of the supplementary material.  Rows 31-60 correspond to subpanel (i) in the Figure S5 of the supplementary material. Rows 1-10 correspond to the band index and give the normalized band population; row 1 corresponds to band index n=1 and row 10 corresponds to band index n=10. Rows 21-30 correspond to the band index and give the normalized band population; row 21 corresponds to band index n=1 and row 30 corresponds to band index n=10. Rows 41-50 correspond to the band index and give the normalized band population; row 41 corresponds to band index n=1 and row 50 corresponds to band index n=10. Rows 11-20 (31-40) [51-60] give the error in the band populations for measurements in panel iii (ii) [i].   FigureS5Cdata.csv contains the data illustrated in Figure 5C of the supplementary material. The first (second) column is the vertical (horizontal) axis. The fourth (third) column is the error in the points on the vertical (horizontal) axis.   push_ramp.py (in Full Hamiltonian simulation.zip) starts with an initial state and evolves it according to the discretized schrödinger equation along the path in q-space. The Hamiltonian is calculated in basic_fcts.py. In figure S6A, at each point in time shown the state is projected onto the instantaneous eigenbasis and the different band populations are extracted. In figure S6B and figure S6C, the whole experiment sequences corresponding to figure 2 and figure 4 in the main text are simulated, and the final population obtained is plotted, with the measurement results copied for reference.   FigureS7code.nb contains the mathematica notebook that generate Figure S7 of the main text. This code uses the two-band model described in the supplemental material to perform simulation.   Image_fitting.zip contains the MATLAB code and functions that were used to analyze the band mapping images. multiboxFit_v7_1.m is the main code that uses other MATLAB functions in the zip file. Overall, it takes absorption images as input, finds the position of each peak (BoxGenerator_v1_0.m), fit for the population in each peak in the images (createFit2D.m), assign the correct band number given the final quasi-momentum in the sequence (BoxesBandsThing_v2.m), and finally plot the inferred band populations, along with a visualization of the original images overlain with a Brillouin zone (PlotBZ_v2.m). The result of fits are saved in a separate file that are accessed by other analysis codes. Figure S2 and Figure S5C are also generated with this code.   Additional codes Q_path_BZ.py, group_velo.py & diffr_img.py are included in Full Hamiltonian simulation.zip to ensure the correct functionality of the codes included.

opencc-by-4.0Jul 2022View details →

Geometric Control on Seismic Rupture and Earthquake Sequence along the Yingxiu-Beichuan Fault with Implications for the 2008 Wenchuan Earthquak

A 65000 years seismic sequence is numerical simulated using TriBIE on the unplanar fault plane with variation normal stress. The code is now available  in an open-source Git-hub project, <a href="https://github.com/daisy20170101/TriBIE/tree/normal_stress_variation">https://github.com/daisy20170101/TriBIE/tree/normal_stress_variation</a>. The modeling will output the bianary format files of  normal stress, fault slip velocity, shear stress, slip during the interseismic loading and coseismic rupture stage, respectively. Since it is impossible to output the data at every time step, especially for the large-scale  fault model. Thus, during the interseismic loading, we set a constant time interval to output data and the t-inter-***.dat file will record every time, when the data is outputted. During coseimic rupture, t-cos-**.dat file records time of outputing data.  So, the size of t-inter-**.dat and t-cos-**.dat file is the number of outputting steps.The fault plane is discretized into 3,1440 elements and the simulation is carried out by parallel computing on 6 servers with 120 CPUs . Each CPU will dispose data of 262 elements.

opencc-by-4.0Sep 2021View details →

Classification of hierarchical text using geometric deep learning: the case of clinical trials corpus

We consider the hierarchical representation of documents as graphs and use geometric deep learning to classify them into different categories. While graph neural networks can efficiently handle the variable structure of hierarchical documents using the permutation invariant message passing operations, we show that we can gain extra performance improvements using our proposed selective graph pooling operation that arises from the fact that some parts of the hierarchy are invariable across different documents. We applied our model to classify clinical trial (CT) protocols into completed and terminated categories. We use bag-of-words based as well as pre-trained transformer-based embeddings to featurize the graph nodes, achieving f1-scores $\simeq 0.85$ on a publicly available large scale CT registry of around 360K protocols. We further demonstrate how the selective pooling can add insights into the CT termination status prediction.

opencc-by-4.0Aug 2022View details →

Blank predetermination in the Iberian Acheulean: fact or fiction? Insight from the cleaver on flake assemblage from Casal do Azemel site (Leiria, Portugal) by a Geometric Morphometric approach

The increase of data available for the study of the Middle Pleistocene in the Iberian Peninsula has favoured the understanding of the technological trends of the regional Acheulean techno-complex. This has features of Large Flake Acheulean -LFA- with a significant presence of cleavers on flake, a specific tool type that is of great cultural and technological value. Moreover, these tools are privileged to discuss the importance of predetermination in Acheulean assemblages. Following this reason, besides the traditional techno-typological approach, we perform 2D Geometric Morphometric Analysis (GMA) to explore this topic on the cleaver on flake assemblage from Casal do Azemel (Leiria, Portugal), a paradigmatic Iberian Acheulean site with a large number of cleavers on flake (more than 100 pieces), one of the largest collections of this type of tools in Western Europe. The results obtained note a strong degree of shape homogeneity and suggest that the different technological solutions underlying the definition of the distal cutting edge, or the intensity of secondary reshaping, do not produce major differences in the overall shape of the tool. Therefore, this homogeneity is a consequence of the existence of a specific pattern of support selection and/or is the outcome of the already predetermined nature of the chosen blank. These observations allowed to discuss the significance of blank predetermination in the Acheulean, highlighting the existence of highly structured technological and cognitive prerequisites.

opencc-by-4.0Sep 2022View details →

Ultraliser: a framework for creating multiscale, high-fidelity and geometrically realistic 3D models for in silico neuroscience

Supplementary Data  <ol> <li>Supplementary Data 1 contains the input (non-watertight) surface meshes of the block (shown in Figure 2a) reconstructed within the context of the EPFL-KAUST collaboration, and the corresponding output (watertight) meshes generated by Ultraliser.</li> <li>Supplementary Data 2 contains a set of 20 non-watertight meshes that were randomly selected from the block shown in Supplementary Figure S54 and another set of the their watertight counterparts.</li> <li>Supplementary Data 3 contains a set of 25 neuronal morphologies with different morphological types and their corresponding watertight meshes.</li> <li>Supplementary Data 4 contains a set of 25 synthetic astroglial morphologies 15 and their corresponding watertight meshes.</li> <li>Supplementary Data 5 contains the vascular morphology (shown in Supplementary Fig. S83) and a corresponding multi-partitioned watertight mesh.</li> <li>Supplementary Data 6 contains the datasets used for the comparative analysis shown in Supplementary Section 13. Neuronal, astrocytic and vascular morphologies are stored in SWC, H5 and VMV file formats respectively. The file structures of the SWC and VMV formats are publicly available online. The H5 files of the complete astrocyte cells can be made available from corresponding authors upon request. All the surface meshes are stored in Wavefront OBJ files. Additional STL meshes are generated to be used for TetGen to create corresponding tetrahedral meshes. All the input and generated data files are publicly available on Zenodo (10.5281/zenodo.7105941).</li> </ol> Data Sources  <ol> <li>Cellular and subcellular NGV meshes segmented from the volume shown in Figure 2 are provided by the collaborating co-authors affiliated with KAUST.</li> <li>Neuronal meshes shown in Figure 3, Supplementary Figures S55 - S75 and Supplementary Figures S85 are publicly available from the MICrONS program.</li> <li>Neuronal morphologies shown in Figure 4, Supplementary Figures S80 - S81 and Supplementary Figure S86 are publicly available from NeuroMorpho.Org.</li> <li>Astrocytic morphologies (Figure 5 and Supplementary Figure S82) are provided by Eleftherios Zisis.</li> <li>Vascular morphologies (rat’s cerebral microvasculature) shown in Figure 6 and Supplementary Figures S83 - S84 are courtesy of Bruno Weber, University of Zürich (UZH).</li> <li>The vascular morphology of the arterial arborizations shown in Supplementary Figure S88 is available from the Brain Vasculature (BraVa) database (cng.gmu.edu/brava).</li> </ol>

dryad40/100

PyTorch geometric datasets for morphVQ models

The methods of geometric morphometrics are commonly used to quantify morphology in a broad range of biological sciences. The application of these methods to large datasets is constrained by manual landmark placement limiting the number of landmarks and introducing observer bias. To move the field forward, we need to automate morphological phenotyping in ways that capture comprehensive representations of morphological variation with minimal observer bias. Here, we present Morphological Variation Quantifier (morphVQ), a shape analysis pipeline for quantifying, analyzing, and exploring shape variation in the functional domain. morphVQ uses descriptor learning to estimate the functional correspondence between whole triangular meshes in lieu of landmark configurations. With functional maps between pairs of specimens in a dataset, we can analyze and explore shape variation. morphVQ uses Consistent ZoomOut refinement to improve these functional maps and produce a new representation of shape variation and area-based and conformal (angular) latent shape space differences (LSSDs). We compare this new representation of shape variation to shape variables obtained via manual digitization and auto3DGM, an existing approach to automated morphological phenotyping. We find that LSSDs compare favorably to modern 3DGM and auto3DGM while being more computationally efficient. By characterizing whole surfaces, our method incorporates more morphological detail in shape analysis. We can classify known biological groupings, such as Genus affiliation with comparable accuracy. The shape spaces produced by our method are similar to those produced by modern 3DGM and to auto3DGM, and distinctiveness functions derived from LSSDs show us how shape variation differs between groups. morphVQ can capture shape in an automated fashion while avoiding the limitations of manually digitized landmarks and thus represents a novel and computationally efficient addition to the geometric morphometrics toolkit.

opencc-zeroSep 2022View details →

Figure 14 in Integrative approach of morphology and geometric morphometrics to species delimiation in Torrenticolidae (Acari: Hydrachnidiae)

Figure 14. Monatractides macrocorpis Gu & Guo, 2019, female. A. Dorsal view. B. Ventral view. C. Palp, lateral view. D. Infracapitulum and chelicera. Scale bars = 100 μm.

Figure 11 in Integrative approach of morphology and geometric morphometrics to species delimiation in Torrenticolidae (Acari: Hydrachnidiae)

Figure 11. Torrenticola brevisuturae Gu & Guo, sp. nov., female. A. Dorsal view. B. Ventral view. C. Palp, lateral view. D. Infracapitulum and chelicera. Scale bars = 100 μm.

Figure 10 in Integrative approach of morphology and geometric morphometrics to species delimiation in Torrenticolidae (Acari: Hydrachnidiae)

Figure 10. Torrenticola brevisuturae Gu & Guo, sp. nov., male. A. Leg-I. B. Leg-II. C. Leg-III. D. Ejaculatory complex. E. Leg-IV- 1–4. F. Leg-IV-5–6. Scale bars = 100 μm.

Figure 1. Landmark definition. A. Dorsoglandularia 1–4. B in Integrative approach of morphology and geometric morphometrics to species delimiation in Torrenticolidae (Acari: Hydrachnidiae)

Figure 1. Landmark definition. A. Dorsoglandularia 1–4. B. Infracapitulum (Torrenticola). C. Infracapitulum (Monatractides).

Figure 13 in Integrative approach of morphology and geometric morphometrics to species delimiation in Torrenticolidae (Acari: Hydrachnidiae)

Figure 13. Monatractides macrocorpis Gu & Guo, 2019, male. A. Dorsal view. B. Ventral view. C. Palp, lateral view. D. Ejaculatory complex. E. Infracapitulum and chelicera. Scale bars = 100 μm.

Figure 9 in Integrative approach of morphology and geometric morphometrics to species delimiation in Torrenticolidae (Acari: Hydrachnidiae)

Figure 9. Torrenticola brevisuturae Gu & Guo, sp. nov., male. A. Dorsal view. B. Ventral view. C. Palp, lateral view. D. Infracapitulum and chelicera. Scale bars = 100 μm.

Figure 12 in Integrative approach of morphology and geometric morphometrics to species delimiation in Torrenticolidae (Acari: Hydrachnidiae)

Figure 12. Torrenticola brevisuturae Gu & Guo, sp. nov., female. A. Leg-I. B. Leg-II. C. Leg-III-1–4. D. Leg-III-5–6. E. Leg-IV-1–4. F. Leg-IV-5–6. Scale bars = 100 μm.

Figure 6 in Integrative approach of morphology and geometric morphometrics to species delimiation in Torrenticolidae (Acari: Hydrachnidiae)

Figure 6. UPGMA tree based on the matrix of the Procrustes distances between species (the anterior margins of coxae).

Figure 4 in Classical and geometric morphometric methods reveal differences between specimens of Varroa destructor (Mesostigmata: Varroidae) from seven provinces of Iran

Figure 4. Dendrogram plotted by on UPGMA method based on morphometric measurement. The vertical line is the

Figure 3 in Classical and geometric morphometric methods reveal differences between specimens of Varroa destructor (Mesostigmata: Varroidae) from seven provinces of Iran

Figure 3. Distribution of morphometric characters in PCA analysis. This graph is based on the average size of the characters, is drawn.

Figure 2 in Classical and geometric morphometric methods reveal differences between specimens of Varroa destructor (Mesostigmata: Varroidae) from seven provinces of Iran

Figure 2. Distribution of six landmarks on the ventral surface of varroa mite for geometric measurement.

Figure 5 in Classical and geometric morphometric methods reveal differences between specimens of Varroa destructor (Mesostigmata: Varroidae) from seven provinces of Iran

Figure 5. Distribution of varroa mite based on a landmark in PCA analysis. Weight matrices data are used for this analysis. Circles show the closer groups.