metadata

size_categories:
  - n<1K

Fluid cube dataset

The Fluid Cube dataset contains 100 fluid dynamics simulations of a fluid block flowing inside a unit cube domain. For each simulation, the fluid block is set with different initial shape, position, velocity, and fluid viscosity.

For more information on how the dataset was generated look a this blog post. The dataset is the same as the one in the blog post but wrapped in HuggingFace's datasets library.

Versions

1000_simulations: The exact same dataset as presented in this blog post.
10_simulations: A subset of the previous the 1000_simulations dataset. Use for quick testing.

Usage

To use the dataset just use:

dataset = datasets.load_dataset('inductiva/fluid_cube', version='10_simulations', split='train')

The dataset has several columns:

['block_position', 'block_dimensions', 'fluid_volume', 'block_velocity',
 'block_velocity_magnitude', 'kinematic_viscosity', 'density', 'tank_dimensions',
 'time_max', 'time_step', 'particle_radius', 'number_of_fluid_particles', 'simulation_time_steps']

The most important of which is the simulation_time_steps which is a list of length equal to the number of time steps in the simulation. Each element on the list is an array with shape (num_particles, 6) that, on each row i, has first the velocity in the x, y and z axis and secondly the position of the particle i in the x, y and z axis.

Dataset columns

block_position: The Initial position of the block;
block_dimensions: The dimensions of the block on each axis;
fluid_volume: Volume of the fluid;
block_velocity: The initial velocity of the block;
block_velocity_magnitude: Initial velocity magnitude;
kinematic_viscosity: Viscosity of the fluid;
density: Fluid density;
tank_dimensions: The dimensions of the tank where the fluid is contained;
time_max: Time, in seconds, of the simulation;
time_step: Elapsed time between each time steps in the simulation;
particle_radius: Radius of the particles;
number_of_fluid_particles: Number of particles;