---
title: Three-Level Ramsey Interferometry
url: https://www.emergentmind.com/topics/three-level-ramsey-interferometry
type: topic
---

# Three-Level Ramsey Interferometry

Searching arXiv for the listed papers and closely related three-level Ramsey interferometry work.
Three-level Ramsey interferometry denotes a family of Ramsey-type protocols in which three states, branches, or readout ports participate in coherent splitting, phase accumulation, and recombination. In contrast to the standard two-level sequence, the additional state can serve as a shared readout channel for two signal-collecting paths, as one bare port of a tripod system whose adiabatic dynamics are confined to a two-dimensional dark-state subspace, or as part of a genuine three-state superposition addressed by generalized multilevel control. Across these realizations, the common structure remains Ramsey-like—preparation, free evolution, and readout—but the phase-to-signal map becomes multilevel, nonlinear, and in some cases explicitly geometric [2606.18443, 2309.10192, 1810.10367].

## 1. Principal architectures

The literature uses the three-level Ramsey concept in several distinct but related ways. In the 2026 three-level metrology proposal, the participating levels are one shared readout state \(|s\rangle\) and two signal-collecting states \(|1\rangle\) and \(|2\rangle\); the defining feature is projected recombination of two internal-path amplitudes onto one detection channel [2606.18443]. In the tripod realization with ultracold \(^{87}\mathrm{Sr}\), three bare ground states \(|1\rangle,|2\rangle,|3\rangle\) are coupled to a common excited state \(|e\rangle\), but the interferometric evolution is effectively restricted to two dark states, so the system behaves as a multiport Ramsey device rather than a three-arm interferometer in the naive sense [2309.10192]. In the \(^{87}\mathrm{Rb}\) Bose–Einstein-condensate treatment, the three-level structure is an equally spaced ladder \(\ket{+1},\ket{0},\ket{-1}\) used to derive closed-form Ramsey expressions before extension to the experimentally relevant five-level manifold [2308.11095]. In the single nuclear-spin qudit experiment, a three-state superposition inside a \(d=4\) quartit is interrogated by a Hadamard–Ramsey sequence, making the three-level sector itself the object of coherence measurement [1810.10367]. A closely related qutrit-specific generalization appears in Wigner–Majorana systems, where the \(D=3\) case already yields a distinct Ramsey resource with enhanced fringe density [2509.06290].

| Realization | Participating states | Defining feature |
|---|---|---|
| Shared-readout three-level Ramsey | \(|s\rangle, |1\rangle, |2\rangle\) | Two signal paths projected onto one readout channel |
| Geometric tripod Ramsey | \(|1\rangle, |2\rangle, |3\rangle\) plus \(|e\rangle\) | Two-state interference inside a dark-state manifold with three-port bare-state readout |
| Three-level ladder Ramsey | \(\ket{+1},\ket{0},\ket{-1}\) | Closed-form multilevel Ramsey algebra under RWA and equal-Rabi condition |
| Hadamard–Ramsey in a quartit | \(\left|-\frac{3}{2}\right\rangle,\left|-\frac{1}{2}\right\rangle,\left|+\frac{1}{2}\right\rangle\) | Coherence measurement of a genuine 3-state superposition |
| Qutrit Ramsey in WM systems | \(D=3\) WM manifold | Doubled oscillation density for the central state transition |

Taken together, these works show that the term is not used in a single narrow sense. Sometimes it denotes three participating bare levels; sometimes it denotes two effective interferometric arms embedded in a three-level platform; and sometimes it denotes a three-state superposition measured by generalized Ramsey gates. That multiplicity is a central conceptual feature of the subject rather than a terminological inconsistency.

## 2. Interference geometry beyond the qubit

The clearest departure from ordinary Ramsey interferometry is the shared-readout geometry. Instead of preparing a superposition of two states and reading out one relative phase, the three-level protocol prepares amplitudes in \(|s\rangle\), \(|1\rangle\), and \(|2\rangle\), allows the two signal branches to accumulate different dynamical phases at frequencies \(\omega_1\) and \(\omega_2\), and then projects both branches back onto the same state \(|s\rangle\). The central interference object is
\[
A_1 e^{i\phi/2}+A_2 e^{-i\phi/2}
=
\cos\!\left(\frac{\phi}{2}\right)+i\epsilon\sin\!\left(\frac{\phi}{2}\right),
\]
with \(A_1+A_2=1\), \(\epsilon=A_1-A_2\), and
\[
\phi=\Delta\omega\, t=(\omega_1-\omega_2)t.
\]
The shared-state population is therefore
\[
P_s(t)=\frac12\!\left[1-V_s(\phi)\cos\!\big(\bar{\omega}t+\Phi(\phi)\big)\right],\qquad
\bar{\omega}=\frac{\omega_1+\omega_2}{2},
\]
where
\[
V_s(\phi)=\sqrt{\cos^2\!\left(\frac{\phi}{2}\right)+\epsilon^2\sin^2\!\left(\frac{\phi}{2}\right)},
\]
and
\[
\Phi(\phi)=\arctan\!\left[\frac{\epsilon\sin(\phi/2)}{\cos(\phi/2)}\right].
\]
The same result can also be written as a projection of two ordinary Ramsey oscillations,
\[
P_s(t)=\frac12\!\left[A_1(1-\cos\omega_1 t)+A_2(1-\cos\omega_2 t)\right].
\]
The crucial difference from the qubit case is therefore not merely the presence of an extra level, but the coherent projection of two signal-sensitive internal paths onto one readout state [2606.18443].

A related but structurally different departure appears in the tripod interferometer. There, the useful evolution occurs in the two-dimensional dark-state subspace even though the bare-state system has three ground-state ports. The first geometric \(\pi/2\) pulse transfers population initially in \(|3\rangle\) into a coherent superposition of \(|3\rangle\) and \(|1\rangle\), while the second geometric \(-\pi/2\) pulse closes the interferometer. Because of the pulse ordering, the “missing” population can appear in \(|2\rangle\), which makes the readout intrinsically multiport [2309.10192].

In qutrit Ramsey interferometry with Wigner–Majorana symmetry, the multilevel structure is again explicit, but the relevant interfering channel can reduce to an effective two-state problem. The qutrit central-state probability exhibits two times more oscillations over the same detuning range than the qubit benchmark \(\cos^2(\Delta\tau/2)\), and the paper attributes this to the structured multilevel evolution rather than to a replacement of the Ramsey sequence itself [2509.06290].

## 3. Dynamical, geometric, and projected phases

Three-level Ramsey interferometry is distinguished not only by state multiplicity but also by the diversity of phase mechanisms that can be converted into fringes. In the shared-readout protocol, the measured phase \(\Phi(\phi)\) is a noncyclic geometric phase defined by the Pancharatnam/Samuel–Bhandari geodesic closure of the projected path in state space. As the relative rotation of the two internal-path amplitudes passes through \(\phi=\pi\), the geodesic closure switches branches. Because the arctangent branch is chosen continuously along the interferometric path, the readout phase shows a sharp transition even though the signal phase \(\phi\) varies smoothly. The phase response is therefore geometric and nonlinear rather than simply proportional to interrogation time [2606.18443].

The tripod realization provides a second geometric mechanism. The effective Hamiltonian in the dark-state subspace is
\[
\hat{H}=\frac{\hat{p}^2\otimes\mathds{1}}{2m} -\frac{\hat{A}\cdot\hat{p}}{m} +\hat{Q}+\hat{w},
\]
with
\[
A_{\mu\nu}=i\hbar\langle D_\mu|\boldsymbol{\nabla}D_\nu\rangle,\qquad
Q_{\mu\nu}=\frac{\hbar^2}{2m}\langle \boldsymbol{\nabla}D_\mu|\boldsymbol{\nabla}D_\nu\rangle,\qquad
w_{\mu\nu}=-i\hbar\langle D_\mu|\partial_t D_\nu\rangle.
\]
The geometric beam-splitter term is
\[
\hat{w}=\hbar\cos\vartheta\frac{\partial\varphi}{\partial t}\hat{\sigma}_y.
\]
During free evolution, \(\partial_t\varphi\to 0\) so that \(\hat{w}\to 0\), but the scalar term remains. In the relevant asymptotic limit,
\[
\lim_{\varphi\to \pi/2,\ \vartheta\to \pi/2}\hat{Q}
=
-\frac{p_r^2}{m}
\begin{pmatrix}
2 & 0\\
0 & 0
\end{pmatrix},
\]
which gives the relative phase
\[
\Delta \phi = \frac{|\hat{Q}_{11}-\hat{Q}_{22}|\,T}{\hbar}
=
\frac{2p_r^2}{m\hbar}\,T.
\]
The notable point is that the interferometric phase does not vanish when the light is turned off; it persists because the geometric scalar term encodes a kinetic-energy offset between the dark-state branches [2309.10192].

Other three-level formulations retain a more conventional dynamical phase. In the \(^{87}\mathrm{Rb}\) ladder model, free evolution with detuning \(\Delta\) is represented by
\[
\hat{U}_{\Omega_R=0}^{\mathrm{Free}}=
\begin{bmatrix}
e^{-iT\Delta} & 0 & 0\\
0 & 1 & 0\\
0 & 0 & e^{iT\Delta}
\end{bmatrix},
\]
so the outer states acquire opposite phases while the middle state remains unchanged. The resulting Ramsey signal,
\[
\frac{\langle \hat{F}_{Z} \rangle}{\hbar}=\frac{1}{9}\left(1+8\cos(\Delta T)\right),
\]
is therefore a multilevel analogue of the ordinary detuning fringe [2308.11095]. In the Tb\(^{3+}\) quartit, geometric phases are also measured explicitly, with \(\Phi_G=\Omega\) in one manifold and \(\Phi_G=2\Omega\) in another, and the fringes become \(\cos^2(\theta/2)\) or \(\cos^2(2\theta/2)\) accordingly [1810.10367].

## 4. Pulse sequences and analytical formulations

Despite their structural differences, three-level Ramsey protocols are usually expressed by a compact split–evolve–recombine algebra. The \(^{87}\mathrm{Rb}\) three-level derivation writes the sequence as
\[
\ket{\Psi_{\mathrm{sys}}(t)}=
\hat{U}^{\mathrm{Split}}\,
\hat{U}^{\mathrm{Free}}\,
\hat{U}^{\mathrm{Split}}\,
\ket{\Psi(0)},
\]
with the three-level interaction Hamiltonian under the RWA and equal-Rabi condition given by
\[
\hat{H}_I = \hbar
\begin{bmatrix}
\Delta & \frac{1}{\sqrt{2}}\Omega_R e^{-i\phi} & 0\\
\frac{1}{\sqrt{2}}\Omega_R e^{i\phi} & 0 & \frac{1}{\sqrt{2}}\Omega_R e^{-i\phi}\\
0 & \frac{1}{\sqrt{2}}\Omega_R e^{i\phi} & -\Delta
\end{bmatrix}.
\]
Its eigenvalues are
\[
\lambda_1 = 0,\qquad \lambda_2=-\hbar\Omega_G,\qquad \lambda_3=\hbar\Omega_G,
\]
with \(\Omega_G=\sqrt{\Delta^2+\Omega_R^2}\). The corresponding analytic propagator yields closed-form Rabi and Ramsey populations, and for the resonant equal-splitting pulse the first solution is
\[
t_{\mathrm{Split}}=\frac{\arccos(1/3)}{\Omega_R}.
\]
This three-level model serves as the derivational testbed for the later five-level treatment [2308.11095].

In the quartit experiment, generalized Ramsey control is expressed by
\[
U_{\mathrm{Had}}\, W_z\, U_{\mathrm{Had}},
\]
where the first Hadamard gate creates a 3-state superposition of
\[
\left|-\frac{3}{2}\right\rangle,\quad
\left|-\frac{1}{2}\right\rangle,\quad
\left|+\frac{1}{2}\right\rangle,
\]
and the second Hadamard gate maps the accumulated phases back into population. The final state for the chosen initialization is written as
\[
U_{\mathrm{Had}}\, W_z\, U_{\mathrm{Had}}
\begin{pmatrix}
0\\
1\\
0
\end{pmatrix}
=
\frac{1}{3}
\begin{pmatrix}
e^{i\omega T}+2\\
e^{i\omega T}-1\\
e^{i\omega T}-1
\end{pmatrix},
\]
which makes the oscillation of the \(\left|-\frac{1}{2}\right\rangle\) population the key coherence observable [1810.10367].

In the qutrit WM setting, the standard qubit sequence
\[
\mathbf{U_2}=R_2(T)F_2(\tau)R_2(T)
\]
is generalized by replacing the qubit Hamiltonian with the WM Hamiltonian
\[
\mathbf{H}_D = \sum_{d=1}^{D} \mathbf{H}_{dd}\,|d\rangle\langle d|
+ \sum_{d=1}^{D-1}\!\Big(\mathbf{H}_{d,d+1}\,|d\rangle\langle d{+}1|
+ \mathbf{H}_{d+1,d}\,|d{+}1\rangle\langle d| \Big),
\]
where
\[
\mathbf{H}_{dd} = \left(d-\frac{D+1}{2}\right)\Delta,
\qquad
\mathbf{H}_{d+1,d}=\mathbf{H}_{d,d+1}^{\ast}=\tfrac{1}{2}\sqrt{d(D-d)}\,\Omega.
\]
For \(D=3\), the qutrit Hamiltonian has equal nearest-neighbor couplings \(\Omega/\sqrt{2}\), and the authors emphasize that “a single driving field, as in standard qubit Ramsey spectroscopy, suffices” [2509.06290].

The tripod protocol departs from resonant pulse-area control and instead uses a first geometric \(\pi/2\) pulse, free evolution for time \(T\), and a second geometric \(-\pi/2\) pulse built from three Gaussian laser pulses in STIRAP-like order. The second sequence is the temporal mirror image of the first. This adiabatic construction replaces standard resonant \(\pi/2\) pulses by geometric state transfer in a degenerate dark-state manifold [2309.10192].

## 5. Sensitivity, visibility, and multilevel metrology

The principal metrological claim of three-level Ramsey interferometry is not that every multilevel protocol automatically improves precision, but that multilevel phase mappings create regimes unavailable to the qubit Ramsey geometry. In the shared-readout three-level proposal, differentiating the projected phase gives
\[
\frac{\partial \Phi}{\partial \phi}
=
\frac{\epsilon}{2\left[\cos^2(\phi/2)+\epsilon^2\sin^2(\phi/2)\right]}
=
\frac{\epsilon}{2V_s^2(\phi)},
\]
and the normalized gain relative to the ordinary Ramsey half-phase slope is
\[
G_N(\phi)\equiv 2\left|\frac{\partial\Phi}{\partial\phi}\right|
=\frac{|\epsilon|}{V_s^2(\phi)}.
\]
At the transition,
\[
V_{s,\min}=|\epsilon|=|A_1-A_2|,
\qquad
G_N^{\max}=G_N(\pi)=\frac{1}{|\epsilon|}.
\]
Hence nearly balanced internal paths yield a steep readout phase near \(\phi=\pi\), but only at the cost of reduced visibility. The representative example \(|A_1-A_2|=0.05\) gives a minimum visibility of \(0.05\) and a peak normalized gain of \(20\), making the gain–contrast tradeoff explicit [2606.18443].

The same work incorporates projection noise and additive classical phase noise through
\[
\Delta\Phi_{\rm PN}=\frac{1}{v\sqrt{N}},
\qquad
\Delta\Phi=\sqrt{\left(\Delta\Phi_{\rm PN}\right)^2+\xi_{\rm CLA}^2},
\]
leading to
\[
\Delta\phi = \frac{\Delta\Phi}{G_N}
=
\frac{1}{G_N}\sqrt{\frac{1}{v^2N}+\xi_{\rm CLA}^2}.
\]
This captures the central tradeoff: geometric gain suppresses the inferred signal-phase noise, whereas reduced visibility amplifies the projection-noise term by \(1/v\). At the shot-noise limit, the gain and visibility penalty mostly compensate, so the geometric response alone does not automatically improve single-shot sensitivity; sub-shot-noise performance would still require nonclassical states such as spin-squeezed or entangled inputs. The advantage arises in the technical-noise-limited regime, where the positive SNR enhancement region survives after both penalties are included. The same paper gives a clock-style stability analysis,
\[
\delta y_1 = \frac{1}{2\pi f_0T_mG_N} \sqrt{\frac{1}{v^2N}+\xi_{\rm CLA}^2},
\qquad
\sigma_y(\tau)=\delta y_1\sqrt{\frac{T_c}{\tau}},
\]
and introduces a phase-offset shortcut in which \(\phi_{\rm off}=0.98\pi\) places the interferometer near the high-slope window in advance. The supplement compares \(T_m=0.98\,\mathrm{s}\) with \(T_m^{\rm short}=0.098\,\mathrm{s}\) while retaining the same dead time and reports improved projected Allan deviation through more frequent sampling [2606.18443].

A different metrological metric appears in qutrit Ramsey interferometry, where the figure of merit is the resolution–contrast index
\[
\mathbf{RCI}_D=\mathbf{Re}^D\,\mathbf{Co}^D.
\]
The qubit baseline is
\[
R=3.584,\qquad C\approx0.999,\qquad \mathrm{RCI}=3.582,
\]
whereas the qutrit yields
\[
R=7.165,\qquad C\approx0.998,\qquad \mathrm{RCI}=7.151.
\]
The authors therefore conclude that qutrits “have doubled oscillation density for the central state transition” and that “three state systems (qutrits) achieve a twofold resolution increase compared to qubits without contrast degradation, emerging as optimal for the qudit approach” [2509.06290].

The quartit Hadamard–Ramsey experiment addresses a different performance quantity, namely multilevel coherence. The damping envelope of the three-state Ramsey oscillation yields a coherence time of about \(T_2\sim 90\,\mu\mathrm{s}\), and the oscillations remain visible even for waiting times around \(25\,\mu\mathrm{s}\) [1810.10367].

## 6. Conceptual boundaries, related analogies, and broader significance

A recurring misconception is that three-level Ramsey interferometry must mean three independent interferometric arms. The surveyed literature does not support that restriction. In the tripod scheme, the interferometer is “fundamentally a two-level interferometer in the dark-state subspace” even though three bare ground states are required for preparation and readout [2309.10192]. In the shared-readout architecture, the essential object is the interference of two signal-carrying internal pathways projected onto one common state \(|s\rangle\), not a symmetric three-arm network [2606.18443]. In the Tb\(^{3+}\) quartit, by contrast, the three-level sector is a genuine three-state superposition interrogated by two Hadamard gates [1810.10367]. The phrase therefore designates a class of multilevel Ramsey structures rather than one canonical topology.

Broader Ramsey analogies reinforce this point. Temporal-mode-selective optical Ramsey interferometry replaces atomic \(\pi/2\) pulses with two coherent, low-efficiency frequency-conversion stages acting on optical temporal modes. The first stage partially transfers amplitude from one frequency band to another, the two bands accumulate a relative phase \(\Delta\Phi_p+\Delta\Phi_s-\Delta\Phi_r\), and the second stage recombines them. The paper describes a “three-level” flavor because the pump, signal, and register bands participate coherently, even though the interferometric readout is conversion between signal and register bands [1710.06736]. In a different direction, sequences of alternating-sign electric-field pulses in the QED vacuum form a time-domain Ramsey interferometer whose “slits” are temporally separated nonadiabatic pair-production events. There the interference appears in the momentum-dependent pair spectrum, and the central peak scales as \(N^2\) for \(N\) pulses, showing that Ramsey logic extends beyond ordinary internal-state spectroscopy to repeated coherent production events [1109.3489].

These analogies do not collapse the differences between platforms, but they clarify the general mechanism: repeated coherent splitting, phase accumulation between branches, and recombination into a measurable fringe. In three-level Ramsey interferometry proper, the third level changes how those steps are encoded. It can create projected internal-path interference, enable dark-state geometric control, support multilevel analytic pulse algebra, or furnish a three-state coherence manifold. This suggests that the main significance of the subject lies less in adding one more level to the qubit protocol than in altering the phase-to-readout map itself.

Source: https://www.emergentmind.com/topics/three-level-ramsey-interferometry