---
language:
- en
pretty_name: Discrete Elastic Rods Simulation Dataset
tags:
- physics
- simulation
- synthetic-data
- graph-neural-networks
- gnn
- geometric-deep-learning
- physics-informed-machine-learning
- regression
- computational-physics
- deformable-objects
- discrete-elastic-rods
- dynamics
- physical-simulation
- computer-graphics
- trajectory-prediction
task_categories:
- graph-ml
- time-series-forecasting
task_ids:
- multivariate-time-series-forecasting
size_categories:
- 1M<n<10M
license: apache-2.0
---

# Discrete Elastic Rods Simulation Dataset

## Dataset Description

This dataset contains synthetic data generated from Discrete Elastic Rods (DER) simulations, a physical model used to represent deformable slender structures such as hair strands, ropes, cables, and elastic fibers.

The dataset was generated frame-by-frame during physical simulations and stores geometric, kinematic, and dynamic properties for each rod vertex.

The primary objective of this dataset is to support machine learning research involving:

- Graph Neural Networks (GNNs)
- Physics-informed learning
- Dynamics prediction
- Physical regression
- Deformable object simulation

Each sample corresponds to a vertex at a specific simulation frame.

---

## Dataset Structure

The dataset is divided into three splits:

| Split | Samples |
|---|---|
| Train | 4,265,580 |
| Validation | 376,200 |
| Test | 460,920 |

---

## Features

| Feature | Description |
|---|---|
| `frame` | Simulation frame index |
| `strand` | Rod/strand identifier |
| `vertex_id` | Vertex identifier |
| `pos_x/y/z` | Vertex position |
| `vel_x/y/z` | Vertex velocity |
| `force_x/y/z` | Applied forces |
| `curvature` | Local curvature |
| `torsion` | Local torsion |
| `prev_segment_direction` | Previous segment direction vector |
| `next_segment_direction` | Next segment direction vector |
| `prev_segment_length` | Previous segment length |
| `next_segment_length` | Next segment length |
| `boundary` | Boundary condition information |

---

## Data Generation

The dataset was generated using a Discrete Elastic Rods simulation environment with randomized physical parameters and dynamic interactions.

The simulations include temporal evolution of elastic rods under physical constraints and external/internal forces.

---

## Intended Use

This dataset is intended for:

- Training Graph Neural Networks
- Physics regression tasks
- Simulation approximation
- Temporal dynamics prediction
- Deformable object learning

---

## Limitations

- The dataset is fully synthetic.
- Results may not perfectly generalize to real-world rod dynamics.
- Physical behavior depends on the simulation assumptions and parameter ranges.

---

## License

- apache-2.0

---

## Citation

```bibtex
@dataset{der_simulation_dataset,
  title={Discrete Elastic Rods Simulation Dataset},
  author={Samuel Ferreira Santos},
  year={2026}
}
```