Skip to content

A universal set of geometric quantum gates through dressed-states assisted shortcuts to adiabaticity

FieldValue
JournalJournal of Physics B Atomic Molecular and Optical Physics
Published2026-02-02
AuthorsMaria Estefanía Rus, Alejandro Ferrón, Omar Osenda, Sergio Santiago S Gomez
DOI10.1088/1361-6455/ae407b

Numerical simulations report universal single-qubit geometric gates with fidelity above 99.9% under intermediate systematic errors. The dressed-state shortcut also generates CS and CNOT gates in an NV center coupled to a nearby ({}^{13}\mathrm{C}) nuclear spin. These results are predictions for a proposed device, not experimental measurements.

The model uses a two-level system with energy difference (\hbar\omega_e), driven by a microwave field in the rotating-wave approximation. Its controls are the Rabi frequency (\Omega_R(t)) and detuning (\Delta(t)=\omega_e-\omega(t)).

The NV center’s spin-1 ground state becomes an effective qubit when a sufficiently strong static field (B_z) isolates its two lowest Zeeman levels. Ordinary geometric gates require adiabatic evolution, which lengthens gate time and increases exposure to decoherence. The shortcut targets the intended endpoint map without retaining slow adiabatic dynamics.

An orange-slice trajectory on Bloch’s sphere supplies the geometric phase for the gate. The (U_z(\gamma_g)) construction starts in (|0\rangle) and follows path ABCDA. The (U_x(\gamma_g)) construction starts in ((|0\rangle+|1\rangle)/\sqrt{2}) and follows BCDAB.

Each of the four intervals covers one quarter of the orange-slice path, with constant azimuth (\phi). The resulting operator (U(\chi,\gamma_g)) gives a z rotation for (\chi=0) and an x rotation for (\chi=\pi/2). Thus (U_z(\pi/2)) is the S gate and (U_x(\pi/2)) is the NOT gate. These rotations form a universal single-qubit gate set.

SATD applies two unitary transformations, first to the adiabatic frame and then to the dressed-state frame. The dressed-state Hamiltonian becomes diagonal, so the dressed-state evolution is simple. Choosing the dressed-state basis leaves the correction function (g_z(t)) available for shaping the dynamics.

SATD was introduced by Baksic et al. in 2016 for speeding up adiabatic state transfer. Here, the dressed-state construction extends it beyond STIRAP-like systems. Standard transitionless driving is recovered when the dressed-state transformation is the identity.

The total dressed-state phase contains a geometric term and a dynamical term. The geometric term is (\gamma_g=\pi-(\phi_2-\phi_1)), while the dynamical term is (\gamma_d=I_{\phi_1}-I_{\phi_2}).

SATD matches the intended endpoints but does not require the shortcut trajectory to follow the adiabatic eigenstates throughout. That difference creates an open path and a residual dynamical phase.

The authors choose (g_{z,B}=-g_{z,A}=g_z), which symmetrizes the dressed-state energy about (t=T/2). This makes (I_{\phi_1}=I_{\phi_2}), cancels (\gamma_d), and leaves a purely geometric operation.

The (g_z) correction changes the pulse profile without directly changing the adiabatic state trajectory. It scales as (\dot{\theta}^2), so it vanishes in the adiabatic limit.

Shortcut pulses can require larger amplitudes than the original adiabatic controls. The paper defines (R) as the ratio between maximum original and corrected pulse amplitudes. Values of (R\leq1) mean that the corrected pulses preserve or reduce the relevant maximum amplitude.

The ratio scales with (x=\tau\Omega_0), so reducing (\tau) need not increase (R) if (\Omega_0) is retuned. The worked pulse example uses (\Omega_0/2\pi=3) MHz, (\eta=\Delta_0/\Omega_0=1), and (x=2).

The resulting scaling profile (P) leaves the maximum amplitude unchanged and causes no abrupt profile changes. The peak ratio (|g_z/\Omega|) decays as the inverse square of (x). Its minimum occurs near (\eta=2), while shorter evolution and (\eta\ll1) increase the correction.

The original microwave phase changes abruptly at (T/2). A smooth interpolation avoids a discontinuity in the corrected pulses. For (\eta\geq1), the appendix reports no significant fidelity loss from this modulation.

The paper links the transition width (\sigma) to microwave-control bandwidth. It states that typical Rabi amplitudes between 10 and 20 MHz allow (\sigma) between 5 and 10 ns.

Robustness to control errors and decoherence

Section titled “Robustness to control errors and decoherence”

Systematic errors enter as a qubit-frequency shift (\delta) and a driving deviation (\epsilon). The simulations use (\Omega_0/2\pi=3) MHz, (\tau=2/\Omega_0), and (\eta=0.5,1,2).

Under detuning error, (\eta=2) produces the gentlest fidelity decline. The NOT gate retains more fidelity than the S gate under the same detuning error. Under driving error, the S gate retains more fidelity than the NOT gate.

For driving error, differences among the tested (\eta) values stay on the order of (10^{-3}) at the largest deviations. With both errors active and (\eta=2), fidelity stays above 0.96 for deviations as large as 15%.

The systematic-error fidelity compares the ideal gate (U_0) with the numerical result (U_r) through (M=U_0^\dagger U_r). The contrasting responses follow from the gate mechanisms. The S gate depends on geometric phase accumulation, whereas the NOT gate is a population inversion.

The decoherence model uses a Lindblad equation with (\kappa_1=1/T_1) and (\kappa_2=1/T_\phi). Typical room-temperature NV values are (T_1) about 1 to 100 ms and (T_\phi) about 0.1 to 1 ms. Isotopically enriched diamond can increase both coherence times.

The simulation fixes (\kappa_1=5\times10^{-4},\mu\mathrm{s}^{-1}) and uses (\delta=\epsilon=0.05). The decoherence fidelity averages state overlaps over 1001 initial states spanning 0 to (2\pi).

Fidelity falls linearly as (\kappa_2) rises, with a steeper decline for the S gate. Dephasing attacks the relative phase used by the S gate, while relaxation affects populations on a longer timescale. Across the tested (\kappa_2) sweep, the authors report single-qubit fidelity above 99.4% under strong environmental decoherence.

Two-qubit gates with a nearby ({}^{13}\mathrm{C}) spin

Section titled “Two-qubit gates with a nearby ({}^{13}\mathrm{C}) spin”

The two-qubit model couples an NV electron spin to a nearby ({}^{13}\mathrm{C}) nuclear spin with (I=1/2). The electronic basis uses (m_s=0,-1), and the nuclear basis uses down and up projections. A microwave drive acts only on the electron spin.

The resulting Hamiltonian has two uncoupled (2\times2) blocks. The upper block reproduces the single-qubit gate, while the lower block supplies the control operation. When (A_{hf}\gg\Omega_R), the lower block produces no transitions. The controlled operation therefore has the form (U_{sq}\otimes I).

The simulations use (A_{hf}/2\pi=130) MHz, a coupling previously reported as experimentally feasible. CS and CNOT gates reach about 99.8% maximum fidelity when deviations remain below 5%.

The fixed hyperfine coupling lowers the ideal-case fidelity relative to single-qubit gates. Both two-qubit gates retain higher fidelity at larger control deviations. The CS gate infidelity also decreases as the hyperfine coupling increases.

The study is entirely numerical, so the reported fidelities do not establish experimental performance. It implements a closed orange-slice trajectory only. The authors state that the protocol can apply to other paths in principle. Open trajectories remain under investigation and are left for future work.

Short gate times raise the correction amplitude, while the microwave phase transition remains bandwidth limited. For NV hardware, the paper emphasizes electron-spin coherence, magnetic noise, static-field level selection, and microwave access. For the two-qubit case, the local ({}^{13}\mathrm{C}) environment must also provide the required hyperfine coupling.

The paper does not specify sample dimensions, surface finish, metallization, or diamond growth method. Those choices require separate material and device qualification.

6CCVD’s diamond materials catalog lists single-crystal and polycrystalline diamond grown by MPCVD. The paper needs a diamond host that supports an NV electron-spin qubit, microwave coupling, and, for the two-qubit extension, a nearby ({}^{13}\mathrm{C}) nuclear spin. The paper specifies no sample size, surface finish, metallization, or growth method, so the engineering support page addresses separate integration decisions.

View Original Abstract

Abstract Geometric quantum computation relies on the geometric phase that arises in adiabatic cyclic evolutions of non-degenerate quantum systems, enabling the design of robust quantum gates. However, the adiabatic condition requires long evolution times, making the system vulnerable to decoherence. In this work, we propose a scheme to realize fast and high-fidelity geometric quantum gates by applying the superadiabatic transitionless driving protocol within the dressed-state framework. We analyze the implementation of single-qubit gates in a two-level system driven by a microwave field, focusing in particular on the nitrogen-vacancy center in diamond. We show how the dynamical phase can be canceled to obtain purely geometric operations. The robustness of the gates is assessed under systematic errors and environmental decoherence, demonstrating high fidelity even in regimes with strong fluctuations. Finally, we extend the protocol to construct nontrivial two-qubit gates, highlighting its feasibility for scalable quantum information processing.