CircuitBuilder: From Polynomials to Circuits via Reinforcement Learning

ArXi:2603.17075v1 Announce Type: new Motivated by auto-proof generation and Valiant's VP vs. VNP conjecture, we study the problem of discovering efficient arithmetic circuits to compute polynomials, using addition and multiplication gates. We formulate this problem as a single-player game, where an RL agent attempts to build the circuit within a fixed number of operations. We implement an AlphaZero-style