This dataset contains the pretrained baseline models, namely FNO, U-Net, and PINN. These models are trained on different PDEs, such as 1D advection, 1D Burgers', 1D and 2D diffusion-reaction, 1D diffusion-sorption, 1D, 2D, and 3D compressible Navier-Stokes, 2D Darcy flow, and 2D shallow water equation. In addition the dataset contains the pre-trained model for the 1D Inverse problem for FNO and U-Net. These models are stored using the same structure as the dataset they trained on. All the files are saved in .pt files, the default file type for the PyTorch library. More detailed information are also provided in our Github repository (https://github.com/pdebench/PDEBench) and our submitting paper to NeurIPS 2022 Benchmark track.

The Research Data Management Plan (RDMP) of the priority program SPP 2170 is the formal document that should help to mangage the handling of data. Since enormous amounts of research data (Big Data) will be generated, the exchange and access to the data should be ensured. Every experiment in the laboratory, or every simulation generates huge amounts of unstructured data. To make these findable, accessible, interoperable, and reusable (FAIR), discipline-specific criteria must be defined in addition to the hardware and software that form the general platform. Therefore the RDMP of the DFG-funded priority program SPP2170 describes how this information could be processed in the future.

The benchmark dataset was generated through a comprehensive simulation study of the deep drawing process for DP600 sheet metal, incorporating variations in geometry, material properties, and process parameters. The simulations were based on the deep drawing of modified quadratic cups with a length of 210 mm and a drawing depth of 30 mm. Three distinct base geometries - Concave, Convex, and Rectangular - were derived from a rectangular reference shape, with key geometric parameters varied in two increments (minimum and maximum). For each geometry, material and process parameters such as the hardening factor (MAT), friction coefficient (FC), sheet thickness (SHTK), and binder force (BF) were systematically varied, resulting in 32,076 unique simulations. Each simulation included stress, strain, thickness distribution, and nodal displacement data for the deep drawing and subsequent springback analysis. The simulation data were stored in HDF5 format, with metadata linking each dataset to its corresponding geometry, material, and process parameters. This structured format ensures efficient retrieval and processing of simulation results, facilitating further analysis and benchmarking.

General A hardware prototype of a four-bar linkage was constructed to generate the presented data set. The data consists of desired input currents supplied to a servo motor and the measured resulting velocities. The mechanism is portrayed in the lab_mechanism_x.jpg images. Further details of the mechanism can be found in the section "Mechanism Setup". For each input trajectory in the input/ folder, the experiment was performed three times. The corresponding measurement files are in the output_empty/ and output_honey/ folders identified by the extension output_xx, where xx is either 00, 01, or 02. For the measurements in the output_honey/ folder, a non-symmetrical stirrer was mounted to the mechanism and was moved through regular supermarket forest honey introducing additional viscous damping into the system. This also allows to supply higher currents for relevant amounts of time to the motor because the maximal motor velocity will not be reached as soon. For the files in the output_empty/ folder, no stirrer was mounted on the mechanism. File Setup The input and output files are comma-separated text files. In the input files, the first line contains a column description (% Time [s], Prescribed Current [mA]) and the following lines indicate the input commands. In the input-command lines, the first value is a time marker in seconds, and the second value is a desired current that should be supplied to the motor from that time on until the time marker in the next line. The output files have a column description in the first line (% Time [s], Goal Current [mA], Present Current [mA], Present Voltage [V], Present Position [rad], Present Velocity [rad/s]) and the following lines are the measurements from the servo motor. It is important to note that the servo motor only has a granularity of 2.69 mA steps for the supplied currents. Hence, the goal current will be the closest multiple of 2.69 below the desired current in the input file. The present position is denoted in rad, where the null position is with the left link in a horizontal position (parallel to the ground link) pointing to the right. The motor then actuates this link in counter-clockwise direction when viewed from the top. Mechanism Setup The four-bar linkage consists of aluminum blocks connected by revolute joints. The joints of the three moving links are 10 mm apart from the edges of the aluminum blocks. The lengths of the moving links are the following (with joint distances denoted in brackets): left link / crank link: 50 mm (30 mm) top link / coupler link: 124 mm (104 mm) right link / rocker link: 80 mm (60 mm) The ground link can be freely adjusted between 45 mm and 120 mm, but was fixed to 95 mm in the conducted experiments. A stirrer can be mounted on the mechanism and can be moved through a liquid introducing viscous damping into the system. A Dynamixel XH430-W350-R servo motor actuates the left link. The servo motor has a built-in controller and can be supplied with a desired current signal to enforce a moment on the left link. The motor is controlled via a C++ program running under Ubuntu 20.04. The baud rate is set to the highest admissible value of 4.5 Mb/s and the USB latency is set to 1 ms. An accelerometer (Bosch BMA456) is mounted to the top of the mechanism but has not been used in the experiments. Python Notebook tutorial.ipynb This Python 3 notebook visualizes the trajectories to get an intuition about the presented data set. It exemplifies how to load and extract values from the input and output files. Afterwards, it plots the input trajectories together with corresponding velocity measurements.

Despite its importance for assessing the effectiveness of communicating information visually, fine-grained recallability of information visualisations has not been studied quantitatively so far. We propose a question-answering paradigm to study visualisation recallability and present VisRecall -- a novel dataset consisting of 200 information visualisations that are annotated with crowd-sourced human (N = 305) recallability scores obtained from 1,000 questions from five question types, which are related to titles, filtering information, finding extrema, retrieving values, and understanding visualisations. It aims to make fundamental contributions towards a new generation of methods to assist designers in optimising information visualisations. This dataset contains stimuli and collected participant data of VisRecall. The structure of the dataset is described in the README-File. Further, if you are interested in related codes of the publication, you can find a copy of the code repository (see Metadata for Research Software) within this dataset.

These measurements are taken in the subsurface at the pilot site next to the weather station of the University of Stuttgart and used to calibrate and validate our pde-based model. The subsurface has been instrumented with 64 temperature sensors, 8 soil moisture sensors. There are four locations, having different soil and soil cover layers. Soil moisture is measured at 60 cm and 100 cm depth, Temperature at 30, 60, 75, 100 cm. At drinking water pipe location, there are two sensors. Column description is to be found in a readme.txt file

The Satellite Altimetry-based Extension of global-scale in situ river discharge Measurements (SAEM) dataset provides a comprehensive solution for addressing gaps in river discharge measurements by leveraging satellite altimetry. This dataset offers enhanced coverage for river discharge estimations by utilizing data from multiple satellite missions and integrating it with existing river gauge networks. It supports sustainable development and helps address complex water-related challenges exacerbated by climate change. The first version of SAEM includes (1) height-based discharge estimates for 8,730 river gauges, covering approximately 88% of the total gauged discharge volume globally. These estimates demonstrate a median Kling-Gupta Efficiency (KGE) of 0.48, surpassing the performance of current global datasets. (2) Catalog of Virtual Stations (VSs): a catalog of VSs defined by specific criteria, including each station’s coordinates, associated satellite altimetry missions, distance to discharge gauges, and quality flags. (3) Altimetric Water Level Time Series: time series data of water levels from VSs that provide high-quality discharge estimates. The water level data are sourced from both existing Level-3 datasets and newly generated data within this study, including contributions from Hydroweb.Next, DAHITI, GRRATS, and HydroSat. Non-parametric quantile mapping functions: for VSs, which model the transformation of water level time series into discharge data using a Nonparametric Stochastic Quantile Mapping Function approach.

Instructions for the first steps with DaRUS

Understanding the link between visual attention and user’s needs when visually exploring information visualisations is under-explored due to a lack of large and diverse datasets to facilitate these analyses. To fill this gap, we introduce SalChartQA - a novel crowd-sourced dataset that uses the BubbleView interface as a proxy for human gaze and a question-answering (QA) paradigm to induce different information needs in users. SalChartQA contains 74,340 answers to 6,000 questions on 3,000 visualisations. Informed by our analyses demonstrating the tight correlation between the question and visual saliency, we propose the first computational method to predict question-driven saliency on information visualisations. Our method outperforms state-of-the-art saliency models, improving several metrics, such as the correlation coefficient and the Kullback-Leibler divergence. These results show the importance of information needs for shaping attention behaviour and paving the way for new applications, such as task-driven optimisation of visualisations or explainable AI in chart question-answering. The files of this dataset are documented in README.md.

Solving the energy balance at the atmosphere-subsurface interface drives heat input (in the summer) into the subsurface. We use this subsequently to calculate heat transport and water flow into the subsurface and then to calculate temperature s around drinking-water supply pipes. This data is from the weather station of the University of Stuttgart. We are providing the measured Boundary Conditions, needed to compute the interface boundary conditions: long wave radiation incoming short wave radiation incoming air temperature in 2 m above ground wind velocity in 2 m above ground relative humidity in 2 m above ground precipitation intensity Data is given tabulated, a readme-file explains the column names.

The Dumux source code is provided in a docker container, which compiles and produces an executable, which models heat transport and water flow from the atmosphere to the subsurface. The needed boundary and initial conditions for four locations, modeled in our paper "Vadose Zone Journal Submission VZJ-2023-06-0046-OA" are provided and can be calculated.

Videos showing water molecules at a sodium chloride (NaCl) solid surface for different water content. The force field for the water is TIP4P/epsilon (https://doi.org/10.1021/jp410865y), and the force field for the ions is from Loche et al. (https://doi.org/10.1021/acs.jpcb.1c05303). The trajectories have been generated using the GROMACS simulation package, and the videos have been created using VMD.

The DaRUS repository entails and supplements the simulation data, MATLAB model and graphics for the publication: "Mechanistic Modeling of In Vivo Translation in Escherichia coli Reliably Identifies Well-Adapted and Optimized RNA Sequences"

We propose Unified Model of Saliency and Scanpaths (UMSS) - a model that learns to predict multi-duration saliency and scanpaths (i.e. sequences of eye fixations) on information visualisations. Although scanpaths provide rich information about the importance of different visualisation elements during the visual exploration process, prior work has been limited to predicting aggregated attention statistics, such as visual saliency. We present in-depth analyses of gaze behaviour for different information visualisation elements (e.g. Title, Label, Data) on the popular MASSVIS dataset. We show that while, overall, gaze patterns are surprisingly consistent across visualisations and viewers, there are also structural differences in gaze dynamics for different elements. Informed by our analyses, UMSS first predicts multi-duration element-level saliency maps, then probabilistically samples scanpaths from them. Extensive experiments on MASSVIS show that our method consistently outperforms state-of-the-art methods with respect to several, widely used scanpath and saliency evaluation metrics. Our method achieves a relative improvement in sequence score of 11.5 % for scanpath prediction, and a relative improvement in Pearson correlation coefficient of up to 23.6 % for saliency prediction. These results are auspicious and point towards richer user models and simulations of visual attention on visualisations without the need for any eye tracking equipment. This dataset contains saliency maps and scanpaths for UMSS and baseline methods. The structure of the dataset is described in the README-File.

Data and scripts for replicating results and the investigation presented in the paper. This includes the dft parameters for generating training data, all training and data selection scripts for the neural networks, scripts for running and analysing the production simulations with the trained potentials.

The dataset offers sequences that are optimized for nonlinearity estimation. The sequences are compliant to IEEE 802.11 standards and are given as binary phase shift keying (BPSK) modulated orthogonal frequency division multiplexing (OFDM) symbol in frequency domain. The sequences are given for various numbers of total subcarriers and positions of occupied subcarriers that match to the training fields defined in the IEEE 802.11 standards. All sequences are normalized to unit power and comply to a maximum peak-to-average-power (PAPR) constraint of 13.05 dB. For each sequence format, one CSV file is given. The columns hold optimized sequences for nonlinearity estimation with orthonormal Laguerre polynomials for each estimation order from P=3 up to P=10.

We report an experiment (N=18) where participants where engaged in a dual task setting in a Social VR (Virtual Reality) scenario. We present a physiologically-adaptive system that optimizes the virtual environment based on physiological arousal, i.e., electrodermal activity. We investigated the usability of the adaptive system in a simulated social virtual reality scenario. Participants completed an n-back task (primary) and a visual detection (secondary) task. Here, we adapted the visual complexity of the secondary task in the form of the number of not-playable characters of the secondary task to accomplish the primary task. We show that an adaptive virtual reality can improve users’ comfort by adapting to physiological arousal the task complexity. Specifically we make available physiological (Electrodermal Activity - EDA, Electroencephalography - EEG; Electrocardiography - ECG) , behavioral and questionnaires data and lastly, the analysis code.

This dataset contains stimuli and collected participant data of VisRecall++. The structure of the dataset is described in the README-File. Further, if you are interested in related codes of the publication, you can find a copy of the code repository (see Metadata for Research Software) within this dataset.

General The one-legged hopper depicted in hooper.png is energetically conservative. The state of the system is given by the horizontal position (x), the vertical position (y), the leg rotation angle (alpha), and their respective velocities. The data required to comprehensively describe a periodic motion (gait) consists of these 6 states, the period (T), and the energy level (E). Since the energy remains constant throughout any motion, E is considered a (bifurcation) parameter. File Setup The files are comma-separated text files encoding matrices of different sizes. The time-series data of a single periodic motion is represented as a 6-by-103 matrix, where the first column and the second column store the energy level (E) and period (T), respectively. The remaining columns (6-by-101) contain the time-series data of the 6 states at 101 equidistant time-steps ranging from time 0 to period time T. While the two bifurcation points stored in the files BP1.txt and BP2.txt include only a single periodic trajectory (a matrix of size 6-by-103), the families/branches of periodic motions (M0.txt - M6.txt) contain multiple trajectories. These families are represented by a matrix of size 6xnPM-by-103, where nPM is the number of periodic motions. The Matlab script main.m and Python script main.py demonstrate how to import and modify the data from the text files and generates the bifurcation diagram shown in Fig. 3 of the corresponding journal paper. Matlab Live Script In addition to main.m and main.py, the Matlab live script main_LiveScript.mlx offers an interactive exploration of the data by allowing the user to select a periodic motion from the bifurcation diagram and display the corresponding time-series data.