Public API: torchdyn package

Subpackages

• torchdyn.core
  • Module contents
• torchdyn.datasets
  • Module contents
• torchdyn.models
  • Submodules
  • torchdyn.models.cnf module
  • torchdyn.models.energy module
• torchdyn.nn
  • Module contents
• torchdyn.numerics
  • torchdyn.numerics.odeint module
  • torchdyn.numerics.sensitivity module
  • torchdyn.numerics.solvers module
  • torchdyn.numerics.hypersolvers module
  • torchdyn.numerics.interpolators module
  • torchdyn.numerics.root module
  • torchdyn.numerics.systems module
  • torchdyn.numerics.utils module

Submodules

torchdyn.utils module

General plotting utilities. These are used in tutorials and are designed for narrow uses.

torchdyn.utils.plot_2D_depth_trajectory(s_span, trajectory, yn, n_samples=128)

Plots 2-dimensional trajectories of points.

torchdyn.utils.plot_2D_space_depth(s_span, trajectory, yn, n_lines)

Plots 2D trajectories in a 3D space (2 dimensions of the system + time).

torchdyn.utils.plot_2D_state_space(trajectory, yn, n_samples=128)

Plots state-space trajectories.

torchdyn.utils.plot_2d_boundary(model, X, y, mesh, num_classes=2, figsize=(8, 4), alpha=0.8)

Plots decision boundary of a 2-dimensional task

torchdyn.utils.plot_2d_flows(trajectory, num_flows=2, figsize=(8, 4), alpha=0.8)

Plots datasets flows learned by a neural differential equation.

torchdyn.utils.plot_3D_dataset(X, yn)

Plots set of points in 3D.

torchdyn.utils.plot_static_vector_field(model, trajectory, t=0.0, N=50, device='cuda')

Plots vector field and trajectories on it.

torchdyn.utils.plot_traj_vf_1D(model, s_span, traj, device, x_span, n_grid, n_levels=30, contour_alpha=0.7, cmap='winter', traj_color='orange', traj_alpha=0.1)

Plots 1D datasets flows.

Module contents