Efficient, direct compilation of SU(N) operations into SNAP & Displacement gates

Abstract

We present a function which connects the parameter of a previously published short sequence of selective number-dependent arbitrary phase (SNAP) and displacement gates acting on a qudit encoded into the Fock states of a superconducting cavity, $V_k(\alpha)=D(\alpha)R_\pi(k)D(-2\alpha)R_\pi(k)D(\alpha)$ to the angle of the Givens rotation $G(\theta)$ on levels $|k\rangle,|k+1\rangle$ that sequence approximates, namely $\alpha=\Phi(\theta) = \frac{\theta}{4\sqrt{k+1}}$. Previous publications left the determination of an appropriate $\alpha$ to numerical optimization at compile time. The map $\Phi$ gives us the ability to compile directly any $d$-dimensional unitary into a sequence of SNAP and displacement gates in $O(d^3)$ complex floating point operations with low constant prefactor, avoiding the need for numerical optimization. Numerical studies demonstrate that the infidelity of the generated gate sequence $V_k$ per Givens rotation $G$ scales as approximately $O(\theta^6)$. We find numerically that the error on compiled circuits can be made arbitrarily small by breaking each rotation into $m$ $\theta/m$ rotations, with the full $d\times d$ unitary infidelity scaling as approximately $O(m^{-4})$. This represents a significant reduction in the computational effort to compile qudit unitaries either to SNAP and displacement gates or to generate them via direct low-level pulse optimization via optimal control.