Efficient Pulse-Level Recompilation of Quantum Circuits: Jenga-Krotov Algorithm and Beyond

Sunny Xin Wang^1*

¹Department of Physics, City University of Hong Kong, Hong Kong, Hong Kong

* Presenter:Sunny Xin Wang, email:x.wang@cityu.edu.hk

As quantum devices move toward practical utility, pulse-level recompilation offers a powerful route to bridge abstract circuit models and hardware-native operations. In this talk, we present three recent advances that collectively enhance gate efficiency, robustness, and learnability. We begin with the Jenga-Krotov (JK) algorithm, a hybrid gradient-based approach tailored for exchange-only qubits, which synthesizes compact, high-fidelity multi-qubit gates with significant reductions in circuit depth and noise sensitivity [1]. Next, we present the inverse Physics-Informed Neural Network framework (iPINN-HL), which reformulates Hamiltonian learning as a PDE-constrained inverse problem [2]. By embedding Schrödinger dynamics into the learning loop, iPINN-HL achieves high-accuracy characterization under realistic sampling constraints. Together, these methods provide a coherent and flexible approach to pulse-level compilation, with implications for quantum control, calibration, and scalable architecture design.

Keywords: Quantum Control, Quantum Circuit Recompilation, Hamiltonian Learning, Physics-Informed Neural Networks, Jenga-Krotov Algorithm