Fast quantum state preparation via AI-based Graph Decimation

We present a fast and general method for preparing stabilizer states with minimum quantum resource consumption. Starting from a graph-state representation of the stabilizer state, our graph decimation algorithm produces a unitary preparation circuit by iteratively reducing the size of the graph using two-qubit Clifford gates. Our approach is powered by AI-based techniques that minimize both gate count and circuit depth. We find that our approach outperforms the state-of-the-art when applied to the preparation of codewords of quantum error correcting codes. We also comment on the application of our approach to other problems such as Clifford synthesis, Hamiltonian simulation, and the construction of arbitrary encoding circuits.