Workflow from Scientific Research

The autoregressive policy network used to generate metamaterials (graphs) with target functional responses. At each step k , the SRV graph, G k , and the target curve, y , are input into the model through a graph and response encoder, respectively. The policy network predicts the start node u and end node v to form a new edge, and the stop token S . When the generation stops (step K ), the final SRV graph, G K , is transformed into the unit cell graph, G K . During RL training, G K is input into the forward model to predict the functional response, guiding the policy optimization.

DOI