Simulation code for "Qubit encodings in the p-orbital-valley spectrum for enhanced coherence and tunable two-qubit interaction"

Abstract

Python notebooks reproducing the numerical results and figures of the manuscript "Qubit encodings in the p-orbital-valley spectrum for enhanced coherence and tunable two-qubit interaction" by John H. Caporaletti and J. P. Kestner (University of Maryland Baltimore County). The manuscript proposes the pOv qubit, an encoding in the two-level subspace spanned by the lowest p-orbital state in the excited valley and the excited p-orbital in the ground valley of an anisotropic silicon quantum dot, coupled by disorder-induced valley-orbit coupling (VOC). This archive contains three self-contained notebooks: VOC_std_calc.ipynb — evaluates the VOC statistics of Appendix VI A using the discretized effective-mass theory of alloy disorder: the vertical envelope function is obtained from a finite-difference solve of the heterostructure confinement potential, and the standard deviations of the disorder-induced generator coefficients, together with the mean valley splitting and mean VOC magnitude, follow from the lattice sum over independent cells. The correlation matrix of the VOC generators is also reported. fig6_Qeff_hist.ipynb — reproduces Fig. 6. For each of $10^4$ alloy-disorder realizations it finds the pair of dot orientations that switches off the two-qubit coupling Omega_zz while maximizing the effective two-qubit quality factor Q_eff, and histograms the result. fig7_inf_scatter_plot.ipynb — reproduces Fig. 7. It synthesizes the R_ZZ($\pi$/2) entangling gate for each disorder realization and evaluates its infidelity against a Monte-Carlo ensemble of quasistatic two-level-fluctuator dipole configurations, comparing a single-segment analytic pulse with a three-segment composite pulse obtained from a two-stage optimization (L-BFGS-B on the deterministic infidelity with analytic gradients, then SLSQP minimizing total gate time subject to a deterministic-infidelity budget). All notebooks use the device parameters of Table I, run top to bottom with no input files, and require only NumPy, SciPy, and Matplotlib (developed under Python 3.12). All random draws are seeded, so the published figures reproduce exactly.