These datasets correspond to the training/validation/testing data for the graph neural network (GNN)-based simulator (GNS) used in Choi and Kumar (2023) "Graph Neural Network-based Surrogate Model for Granular Flows" (https://arxiv.org/abs/2305.05218). It deals with three different types of granular flow problems: "BarrierInteraction", "ColumnCollapseSimple", and "ColumnCollapseFrictional". The details of these granular flow problems are explained in Choi and Kumar (2023). The datasets also include the trained GNS models that can be used to simulate the granular flow problems.
GNS is a generalizable, efficient, and accurate machine learning (ML)-based surrogate simulator that uses GNNs. GNS is a viable surrogate for numerical methods such as Material Point Method, Smooth Particle Hydrodynamics and Computational Fluid dynamics and can be extended to simulate natural hazards. GNS can handle complex boundary conditions and multi-material interactions. GNS exploits distributed data parallelism to achieve fast multi-GPU training. The source code is available at https://github.com/geoelements/gns. Detailed instructions on how to use this data are published along with this version of the dataset and also located as a `README.md` file in the GitHub repository.
Simulation | Training, validation, testing data, and trained model
Cite This Data:
Choi, Y., K. Kumar (2023). "Training, validation, testing data, and trained model", in Dataset for Graph Neural Network-based Surrogate Model for Granular Flows. DesignSafe-CI. https://doi.org/10.17603/ds2-4nqz-s548
The dataset includes the training/validation/testing data for the graph neural network (GNN)-based simulator (GNS) used in Choi and Kumar (2023) "Graph Neural Network-based Surrogate Model for Granular Flows" cited in the Related Work. Graph neural network (GNN)-based simulator (GNS) is a generalizable, efficient, and accurate machine learning (ML)-based surrogate simulator that uses Graph Neural Networks (GNNs) originally introduced in the paper "Learning to Simulate Complex Physics with Graph Networks" by DeepMind (see the Related work). GNS is a viable surrogate for numerical methods such as Material Point Method, Smooth Particle Hydrodynamics and Computational Fluid dynamics and can be extended to simulate natural hazards. GNS can handle complex boundary conditions and multi-material interactions. We improved GNS to be able to exploit distributed data parallelism to achieve fast multi-GPU training. This improved version of GNS is published in GitHub "Graph Network Simulator (GNS) and MeshNet" (see the Related Work). We investigate the performance of GNS in learning to simulate complex granular flow problems: "BarrierInteraction", "ColumnCollapseSimple", and "ColumnCollapseFrictional". The details of these granular flow problems are explained in Choi and Kumar (2023). The GNS is trained on the `train.npz`, and validated and tested on the `validation.npz` and `test.npz` uploaded in this dataset. Detailed instructions on how to train and test the GNS are explained in `README.md` file in the GitHub repository https://github.com/geoelements/gns in the Related Work. The trained model is also included in this dataset. The result shows that GNS can accurately simulate the granular flow dynamics faster than high-fidelity simulators, and is generalizable to other configurations not seen during the training. More information about the result can be found in Choi and Kumar (2023).
Report | Instruction
Description:
This is an instruction about how we generated the simulation data for `BarrierInteraction`, `ColumnCollapseSimple`, and `ColumnCollapseFrictional`, and how to make `train.npz`.
File Name
Simulation Model | BarrierInteraction
Description:
This category includes the training/validation/testing data for "BarrierInteraction" along with the trained model to simulate it. The details of the simulation and results are described in Choi and Kumar (2023).
It includes the inputs to run MPM simulations that are used to make `train.npz` for the corresponding simulation model ("BarrierInteraction", "ColumnCollapseFrictional", or "ColumnCollpaseSimple")
This category includes the training/testing or validation data in `.npz` format the corresponding simulation model ("BarrierInteraction", "ColumnCollapseFrictional", or "ColumnCollpaseSimple") compatible with the open-source GNS code: Kumar, K., & Vantassel, J. (2023). GNS: A generalizable Graph Neural Network-based simulator for particulate and fluid modeling. Journal of Open Source Software, 8(88), 5025. https://doi.org/10.21105/joss.05025.
In addition, it includes the configuration data (such as information about boundaries and hyperparameters) for the corresponding simulation model ("BarrierInteraction", "ColumnCollapseFrictional", or "ColumnCollpaseSimple") in `.json` format to train, test, and validate GNS. It is named `metadata.json`
This category includes the trained GNS models for the corresponding simulation models (i.e., "BarrierInteraction", "ColumnCollapseSimple", or "ColumnCollapseFrictional") along with its training state.
This category includes the training/validation/testing data for "ColumnCollapseSimple" along with the trained model to simulate it. The details of the simulation and results are described in Choi and Kumar (2023).
It includes the inputs to run MPM simulations that are used to make `train.npz` for the corresponding simulation model ("BarrierInteraction", "ColumnCollapseFrictional", or "ColumnCollpaseSimple")
This category includes the training/testing or validation data in `.npz` format the corresponding simulation model ("BarrierInteraction", "ColumnCollapseFrictional", or "ColumnCollpaseSimple") compatible with the open-source GNS code: Kumar, K., & Vantassel, J. (2023). GNS: A generalizable Graph Neural Network-based simulator for particulate and fluid modeling. Journal of Open Source Software, 8(88), 5025. https://doi.org/10.21105/joss.05025.
In addition, it includes the configuration data (such as information about boundaries and hyperparameters) for the corresponding simulation model ("BarrierInteraction", "ColumnCollapseFrictional", or "ColumnCollpaseSimple") in `.json` format to train, test, and validate GNS. It is named `metadata.json`
This category includes the trained GNS models for the corresponding simulation models (i.e., "BarrierInteraction", "ColumnCollapseSimple", or "ColumnCollapseFrictional") along with its training state.
This category includes the training/validation/testing data for "ColumnCollapseFrictional" along with the trained model to simulate it. The details of the simulation and results are described in Choi and Kumar (2023).
It includes the inputs to run MPM simulations that are used to make `train.npz` for the corresponding simulation model ("BarrierInteraction", "ColumnCollapseFrictional", or "ColumnCollpaseSimple")
This category includes the training/testing or validation data in `.npz` format the corresponding simulation model ("BarrierInteraction", "ColumnCollapseFrictional", or "ColumnCollpaseSimple") compatible with the open-source GNS code: Kumar, K., & Vantassel, J. (2023). GNS: A generalizable Graph Neural Network-based simulator for particulate and fluid modeling. Journal of Open Source Software, 8(88), 5025. https://doi.org/10.21105/joss.05025.
In addition, it includes the configuration data (such as information about boundaries and hyperparameters) for the corresponding simulation model ("BarrierInteraction", "ColumnCollapseFrictional", or "ColumnCollpaseSimple") in `.json` format to train, test, and validate GNS. It is named `metadata.json`
This category includes the trained GNS models for the corresponding simulation models (i.e., "BarrierInteraction", "ColumnCollapseSimple", or "ColumnCollapseFrictional") along with its training state.
