Tier 2: Inverse Multi-Objective Synthesis

Tier 2 incorporates genetic optimization and adjoint gradients to synthesize optimal semiconductor geometries.

Features

• NSGA-II Non-Dominated Sorting: Multi-objective genetic algorithm optimizing power, delay, and thermal dissipation simultaneously.
• Transistor Genome Encoding: Gate-All-Around (GAA) Nanosheet, Complementary FET (CFET), and FinFET geometric representations.
• Fast Adjoint Gradient Refinement: Hybrid search combining broad evolutionary exploration with rapid local gradient ascent.
• Pareto Optimal Frontier Exploration: Interactive trade-off curves exportable directly to compact SPICE models.