ICCAD Contest 2024 Problem A

I built an ABC-based automated optimization flow for logic-netlist optimization and designed an evaluation framework around the contest’s black-box cost estimator.

• Constructed a logic-synthesis action space for reinforcement learning and explored transformations including resub, rewrite, refactor, and balance.
• Designed and implemented Genlib Maker to convert the official JSON cell library into ABC-compatible .genlib and .lib inputs.
• Improved search efficiency through incremental iterative optimization.

Technologies: ABC, C++, Python, reinforcement learning, logic synthesis