Graph Neural Network Message Passing

Every Graph Neural Network layer runs the same core operation: each node collects and aggregates the feature vectors of its neighbors, then updates its own vector from that aggregate. This simulator renders a real graph in 3D, gives every node a 3-dimensional feature vector displayed as its color, and lets you step through real message-passing rounds — sum, mean, or the symmetric-normalized rule from Kipf & Welling's GCN — while animated pulses travel along the edges to show information actually flowing between neighbors. Live readouts track the hop count, Dirichlet energy and embedding variance, making the well-known over-smoothing effect — where deep stacks of GNN layers wash every node toward the same indistinguishable vector — directly visible and measurable rather than just described.