QSP-based pulse characterization

Exact pulse implementations for quantum systems are critical for tasks in analog quantum computing. However, calibrating composite pulse sequences exactly is challenging, oftentimes requiring iterative approaches which fail under realistic noise models. Using quantum signal processing, we produce an analytic form for composite pulse sequences, finding they can be expressed as Laurent polynomials. Via Fourier-based polynomial interpolation, we may learn these polynomials and the underlying rotations, enabling us to characterize pulse sequences. We test this learning scheme numerically and under open quantum dynamics.