← Back

Four-Phase Step Breakdown

PD-NCA operates in 4 phases: (1) Processing - agents propose Δs via CNN, (2) Competition - pairwise strengths from attack·defense, (3) Normalization - softmax(τ) produces weights, (4) State Update - weighted Δs applied and clipped.

Phase 1: Processing
(Agent Proposals)

Phase 2: Competition
(Strength Calculation)

Phase 3: Normalization
(Softmax Weights)

Phase 4: Update
(Final State)

Current Phase: Ready

Full Steps: 0

Current Phase: -

Click "Single Step" to advance one phase at a time, or "Play" for continuous simulation.