This is the source code for paper A Dynamical Graph Prior for Relational Inference.
python>=3.3.7
torch>=1.9.0
torch-cluster>=1.5.9
torch-geometric>=2.0.0
torch-scatter>=2.0.8
torch-sparse>=0.6.11
tqdm
This code was tested on macOS and Linux.
python train_DYGR.py --suffix MM_ER50_exp0
-
Michaelis–Menten kinetics, a model for gene regulation circuits.
-
Rössler oscillators on graphs, which generate chaotic dynamics.
-
Diffusion, a continuous-time linear dynamics.
-
Spring model that describes particles connected by springs and interacts via Hooke’s law;
-
Kuramoto model that describes phase-coupled oscillators placed on a graph.
-
Friedkin-Johnsen dynamics, a classical model for describing opinion formation, polarization and filter bubble in social networks;
-
The coupled map network, a discrete-time model with chaotic behavior.
-
Netsim, a simulated fMRI data.