Yurke-Type SU(1,1) Interferometer

• Yurke-type SU(1,1) interferometer is a nonlinear quantum device that replaces passive beam splitters with two-mode parametric amplifiers for enhanced phase measurement.
• It employs an SU(1,1) algebraic structure with cascaded two-mode squeezing operations and a phase shifter to surpass the shot-noise limit.
• Advanced designs integrate multimode operation, dispersion engineering, photon subtraction, and optimized detection protocols to achieve Heisenberg scaling.

A Yurke-type SU(1,1) interferometer is a nonlinear, quantum-enhanced interferometric configuration in which passive beam splitters are replaced by two-mode parametric amplifiers tunable via the SU(1,1) algebra. This design provides phase sensitivity that can surpass the shot-noise limit (SQL) and approach Heisenberg scaling, by leveraging quantum correlations between modes rather than classical intensity division. The canonical implementation consists of two cascaded parametric amplifiers (PAs) with a controllable phase shift inserted in one mode between them. Multimode implementations, dispersion engineering, seeding strategies, detection protocols, and photon subtraction at the output further optimize sensitivity and metrological performance.

1. SU(1,1) Algebraic Structure and Operational Principle

The essential operators of a Yurke-type SU(1,1) interferometer are the SU(1,1) generators: $K_+ = a_s^\dagger a_i^\dagger$, $K_- = a_s a_i$, and $$K_0 = \frac{1}{2}(a_s^\dagger a_s + a_i^\dagger a_i + 1)$$. They satisfy $[K_0, K_+] = +K_+$, $[K_0, K_-] = -K_-$, $[K_-, K_+] = 2K_0$. The core building block is the unitary two-mode squeezer $$U_P(g, \theta) = \exp[g e^{i\theta} K_+ - g e^{-i\theta} K_-]$$, where $g$ is the gain and $\theta$ the pump phase.

A Yurke-type SU(1,1) interferometer comprises two PAs (with gain parameters $g_1$, $g_2$), separated by a phase shifter $\phi$ applied to one of the modes (typically the idler). The total unitary sequence is $$U_\mathrm{tot}=U_P(g_2, 0) \cdot e^{i\phi a_i^\dagger a_i} \cdot U_P(g_1, 0)$$. In the plane-wave, single-frequency limit, this implements a two-mode squeezing and rotation that can surpass conventional shot-noise scaling (Ferreri et al., 2020).

2. Input–Output Relations and Phase Sensitivity

The Heisenberg-picture input–output relations for signal ($a_s$) and idler ($a_i$) through the interferometer, for equal gain in both PAs, are: $$\begin{aligned} a_s^{(\text{out})} &= a_s^{(0)} \cosh g_1 \cosh g_2 + a_i^{(0)\dagger} \sinh g_1 \cosh g_2 + e^{-i\phi} [ a_i^{(0)\dagger} \cosh g_1 \sinh g_2 + a_s^{(0)} \sinh g_1 \sinh g_2 ], \ a_i^{(\text{out})} &= \text{... (similarly) } \end{aligned}$$ The measured observable is typically the signal photon number $N_s = a_s^{(\text{out})\dagger} a_s^{(\text{out})}$, whose mean and variance as functions of $\phi$ are: $$\langle N_s(\phi) \rangle = 2 \sinh^2 g [ \cosh^2 g - \sinh^2 g \cos \phi ], \quad \text{Var}(N_s) = \sinh^2 g\, \cosh^2 g [1 + \cos \phi ].$$ Phase sensitivity is quantified as $$\Delta\phi = \Delta N_s / |d\langle N_s\rangle/d\phi|$$. For large $g$, $\Delta\phi \simeq 1 / N$ (Heisenberg scaling), where $N$ is the total probe photon number (Ferreri et al., 2020).

3. Single-Mode vs. Highly Multimode Interferometry

Realistic sources generate highly multimode biphoton states with joint spectral amplitude (JSA) $$F(\omega_s, \omega_i) = \sum_k \sqrt{\lambda_k} u_k(\omega_s) v_k(\omega_i)$$. Schmidt mode operators $A_k, B_k$ diagonalize the mode structure, with each $(A_k, B_k)$ pair squeezed independently by gain $\gamma_k = G\sqrt{\lambda_k}$. For $K$ Schmidt modes,

$$\langle N_s \rangle = \sum_k \sinh^2 \gamma_k, \quad \mathrm{Var}(N_s) = \tfrac{1}{4} \sum_k \sinh^2(2\gamma_k).$$

Single-mode operation ($\lambda_1 = 1$) achieves ideal Heisenberg scaling, while strong multimode occupation degrades phase sensitivity due to leakage “spilling out” from the dark fringe, especially at higher gains (Ferreri et al., 2020, Frascella et al., 2019).

4. Integrated, Spectrally Multimode Designs and Dispersion Engineering

Recent architectures have focused on integrated waveguide platforms (e.g., KTP) with polarization converters and CW pump to realize compact, multimode Yurke-type SU(1,1) interferometers (Ferreri et al., 2020, Ferreri et al., 2022). Key features include:

• Use of a mid-line polarization converter to compensate group-velocity mismatch, swapping o/e polarizations after the first PDC stage.
• The output JSA takes the form $F(\omega_s, \omega_i) \propto \alpha(\omega_s+\omega_i) \sinc[\Delta\beta L/2] \cos(\phi/2) e^{i...}$.
• Near-perfect destructive interference occurs at the “dark fringe” when $\phi_{dark} = \pi$ ($\cos(\phi/2)=0$).

Spectral filtering around the pump central frequency and dispersion suppression via careful polarization converter placement maximize phase sensitivity and visibility, enabling sub-SQL operation at photon numbers up to $10^4$ and visibilities $\gtrsim98\%$ (Ferreri et al., 2020, Ferreri et al., 2022).

5. Detection Protocols and Seeding Strategies

Detection modalities include direct photon-counting and homodyne detection:

• Photon-counting: Optimal for moderate gain and vacuum seeding; phase estimation uses the classical error-propagation bound.
• Homodyne detection: Local oscillator matched to dominant Schmidt mode yields quadrature measurement $$H = |\beta_\mathrm{lo}| [ e^{i\theta} A_1^{(\mathrm{out})} + e^{-i\theta} A_1^{(\mathrm{out}) \dagger} ]$$. This can outperform counting at low gains (Ferreri et al., 2020, Anderson et al., 2016).
• Seeding: Vacuum input already beats SQL. Bright seeding or single-mode seeding disrupts signal–idler balance or multimode correlations. Distribution of weak seeds across Schmidt modes preserves two-mode quantum advantage.

Filtering of the output further restricts phase estimation to best-compensated modes, eliminating deleterious side modes (Ferreri et al., 2020).

6. Enhanced Phase Sensitivity via Output Photon Subtraction

Photon subtraction at the output port of the interferometer (operator $M_m = a^m$) is an advanced protocol that enhances phase sensitivity and quantum Fisher information (QFI). For input $|0\rangle_a \otimes |\alpha\rangle_b$, subtraction of $m$ photons induces a non-Gaussian transformation that tightens both the SQL and HL, especially in super-Poissonian regimes:

• Phase sensitivity $\Delta\phi$ and QFI $F_Q$ improve monotonically with $m$ (number of subtracted photons).
• Internal photon losses (within the interferometer) have a more significant adverse effect than external losses. Photon subtraction partially compensates internal loss, maintaining sub-SQL operation up to $T \approx 0.6$.
• For $m \ge 2$, the scheme can approach HL over a wide range of losses.
• The protocol is robust to input (vacuum vs. coherent state), and the refinement is mathematically explicit in error propagation and QFI expressions (Jiang et al., 2024).

7. Comparative Analysis and Future Directions

The Yurke-type SU(1,1) interferometer has been implemented across photonic (integrated and fiber-based (Ferreri et al., 2020, Ferreri et al., 2022)), atomic (Linnemann et al., 2017, Krešić et al., 2023), and optomechanical platforms (Meng et al., 26 Sep 2025). It exceeds the SQL for phase sensitivity by leveraging two-mode quantum correlations and active nonlinear operations.

Key technical challenges remain:

• Achieving single-Schmidt-mode dominance in multimode platforms to preserve Heisenberg scaling.
• Suppressing internal loss, which is more detrimental than detection inefficiency.
• Implementing time-reversal protocols (matched gain, dark fringe operation) for robust performance against technical noise and instabilities.

The future of Yurke-type SU(1,1) interferometry lies in integrated photonics, robust multimode dispersion engineering, advanced quantum-enabled detection, and hybrid matter-light platforms, with immediate applications in quantum metrology, imaging, and simulation.