Compile-Time Simplification of Classically Controlled Operations in Dynamic Circuits

This research paper presentation by Innocenzo Fulginiti covers compile-time simplification of classically controlled operations in dynamic quantum circuits. Dynamic circuits extend static circuits with mid-circuit measurements and classically controlled operations, enabling classical feed-forward in which measurement outcomes stored in a classical register determine subsequent quantum operations. While this makes circuits more expressive and enables adaptive quantum algorithms, measurements are noisier than unitary gates, and feed-forward introduces runtime latency from communication between the quantum processor and the classical controller, reducing execution reliability. The proposed optimization moves runtime control decisions to compile time via a two-phase pipeline. In phase one, the compiler symbolically simulates the dynamic circuit, tracking quantum and classical states; where sufficient information exists, dynamic components are removed or replaced with probabilistic constructs, yielding an intermediate probabilistic circuit that represents feed-forward behavior through probabilistic gates and controls. In phase two, these constructs are sampled per execution to produce concrete executable circuit instances that contain no classical feed-forward while preserving the original dynamic-circuit semantics. Technical contributions include extending the probabilistic circuit model with probabilistic controls, defining classical constant propagation as a static analysis over the classical register and probabilistic bits, and implementing the combined optimization pass. Evaluation on randomly generated dynamic circuits shows a reduction in the number of classically controlled operations, implying reduced reliance on classical feed-forward at runtime.