Programmable quantum decision trees

Recent efforts have extended decision trees into the quantum domain. However, because root and internal nodes branch into multiple successors, such approaches require duplicating node quantum states, violating the no-cloning theorem. Existing implementations therefore rely on measurements at each node, interrupting coherent quantum evolution and obscuring correlations between data and attributes. Here, we encode the entire decision tree in a tight-binding Hamiltonian and employ non-equilibrium Green's functions to compute transmission from the root to the leaf nodes. This framework enables dynamic control of the influence of individual nodes, yielding a programmable quantum decision-tree architecture.