Operational theory for photonic circuits: generalizing linear optics beyond quantum theory

AI summary

Overview

Research area: Quantum foundations and quantum optics — specifically generalised probabilistic theories (GPTs) applied to photonic circuits, interference, and nonclassicality.

Technical level: Advanced. The paper assumes familiarity with phase-space methods (quadratures, Wigner functions), linear-optical transfer matrices, and the GPT framework. The framing and conclusions are accessible, but the formalism is not introductory.

Scope (one sentence): The paper builds a phase-space GPT framework for linear-optical circuits (beam splitters, phase shifters, interferometers) and uses it to test whether the Hong-Ou-Mandel effect and Mach-Zehnder interference are uniquely quantum, comparing quantum theory against classical optics and a "sawtooth" theory.

What This Paper Is About

Quantum optics is routinely used to claim "genuinely quantum" behaviour, but linear-optical setups are described by the same transfer matrices in classical and quantum optics — the theories differ only in which states and measurement spectral measures are admissible. The authors exploit the analogy between optical quadrature operators and the phase-space coordinates of a mechanical harmonic oscillator to define beam splitters and phase shifters inside generalised probabilistic theories, then ask whether textbook effects survive outside quantum theory. The goal is to identify which photonic phenomena are truly quantum-specific and which are merely features of a broad class of operational theories.

Key Contributions

1. A phase-space GPT framework for photonic circuits. Beam splitters and phase shifters are defined as maps on bipartite phase-space pseudo-probability distributions, using the same unitary transfer matrix that appears in classical and quantum optics (Eqs. 1–3, 13, 17), with the oscillator scales $x_0=\sqrt{\hbar/m\omega}$ and $p_0=\sqrt{\hbar m\omega}$.

2. A structural lemma linking beam splitters and phase shifts. Lemma 1 proves that a phase shift applied to both input ports (or both output ports) commutes with the beam-splitter map $\Phi_U$, which is used to argue that coincidence statistics are insensitive to global phase but sensitive to the relative phase $\theta_{\mathrm{rel}}$.

3. A cross-theory comparison of two canonical interferometric setups. The Hong-Ou-Mandel and Mach-Zehnder setups are evaluated in quantum theory, classical theory, and sawtooth theory, with a coincidence probability defined via an energy threshold $E_{\mathrm{th}}$ and a visibility $V$ obtained by averaging over the relative phase (Eqs. 19, 21).

4. A proposed interferometric contextuality test. A Mach-Zehnder sequence $\Psi_\varphi$ combined with two measurements — the coincidence measurement and a "Hadamard" measurement (coincidence preceded by an interferometer with $\varphi=\pi/4$) — is used to generate joint numerical ranges of measurement statistics, aimed at contextuality inequalities derived from the Clauser-Horne-Shimony-Holt inequality.

Main Findings

• The Hong-Ou-Mandel dip is not quantum-specific. In quantum theory with input $|11\rangle$, the output state is $(|20\rangle-|02\rangle)/\sqrt{2}$ and the coincidence probability is exactly zero, giving $V = 100%$. But the authors show the same vanishing of coincidences arises for classical light and for a modified version of quantised light (sawtooth theory).

• Quantum theory with coherent states gives reduced visibility. For input coherent states with $|\alpha_1^{\mathrm{ini}}| = |\alpha_2^{\mathrm{ini}}| = 1$ (mean energy equal to a single photon), the initial coincidence probability is only $P_{\mathrm{co}} \approx 40%$, and the phase-averaged visibility is $V \approx 80%$ (Fig. 4).

• Classical theory can approach 100% visibility. Using delta-distributed classical states at energy $E_{\mathrm{cl}} = 3h_0/2$ (radius $r_{\mathrm{cl}} = \sqrt{3}$), the visibility goes to $V \rightarrow 100%$ as $E_{\mathrm{th}} \rightarrow E_{\mathrm{cl}}$. Figure 7 shows coincidence curves for $E_{\mathrm{th}}/E_{\mathrm{cl}} = 1/5$, $1/2$, and $9/10$; at $E_{\mathrm{th}}/E_{\mathrm{cl}} = 1-\varepsilon$ with small $\varepsilon>0$, coincidence is near zero for almost all $\theta_{\mathrm{rel}}$ because the output points are antipodal.

• Sawtooth theory caps visibility at $2/\pi$. With energy spectrum $E_n = h_0 n/2$ and threshold $E_{\mathrm{th}} = h_0/4$, using delta-distributed states at radius 1, the coincidence probability is $P_{\mathrm{co}} = 1-|\cos(\theta_{\mathrm{rel}})|$, reaching a maximum visibility of $V = 2/\pi \approx 64%$. The authors note this maximum holds only for the specific states considered, since sawtooth theory can accommodate negative phase-space states that might allow higher visibilities.

• Mach-Zehnder interferometry shows no qualitative separation. Quantum theory with $|11\rangle$ gives the signal $S(\varphi) = \cos(\varphi)^2$ (Eq. 23). Classical theory produces a measurement signal (Fig. 8, computed at $E_{\mathrm{th}}/E_{\mathrm{cl}} = 9/10$ for $\theta_{\mathrm{rel}} = \pi/10$, $\pi/2$, and phase-averaged) that is "only quantitatively but not qualitatively different." Sawtooth theory gives $S(\varphi) = 1-|\sin(\theta_{\mathrm{rel}})\sin(\varphi)|$, again showing an interference pattern (Fig. 10).

• Correlation sets differ by theory, with classical optics the most general. The introduction states that the achievable correlations in the interferometric setup (Fig. 11) form characteristic sets for each theory, with classical optics yielding the most general set that includes the correlations of the other theories. The detailed numerical results of this section are not available in the provided content, which is truncated mid-sentence.

Methodology in Plain English

The authors start from a simple observation: in classical optics, a beam splitter multiplies electric-field amplitudes by a unitary matrix, and in quantum optics the same matrix multiplies mode creation operators. Rewriting both in terms of position- and momentum-like quadratures reveals an identical transformation rule, so the optical element itself does not distinguish the theories.

The theories are distinguished instead by three ingredients: which phase-space distributions count as states, which functions count as observables, and which "spectral measures" convert observables into outcome probabilities. Classical theory uses point-like (delta) states and sharp energy spectral measures; quantum theory uses Wigner functions that may go negative and discrete energy levels. The authors take a third option, sawtooth theory, from prior work, which has evenly spaced energies $E_n = h_0 n/2$.

For each theory they prepare two input modes with a controlled relative phase, push them through a balanced beam splitter $U = \frac{1}{\sqrt{2}}\begin{pmatrix}1&1\1&-1\end{pmatrix}$, and compute the probability that both output modes exceed an energy threshold — the coincidence probability. Averaging this over the relative phase gives a visibility, making the three theories directly comparable on the same footing. The Mach-Zehnder case adds a phase shift between two beam splitters and plots the coincidence signal against that phase.

Why This Matters

Impact on research. The results sharpen what counts as evidence of nonclassicality in linear optics. Since the Hong-Ou-Mandel dip and Mach-Zehnder interference appear in classical and sawtooth theory too, these effects cannot by themselves certify quantum behaviour. This connects to the paper's motivation that photonic boson sampling has been used to demonstrate quantum computational advantage, and to the observation that effects such as Bell inequality violation remain genuinely quantum in contrast to linear-optical interference. The framework also extends prior GPT work on continuous-variable systems (the harmonic oscillator and the hydrogen atom) into practical circuit-level modelling.

Real-world applications (as motivated by the paper):

• Benchmarking and validating photonic quantum computers and boson samplers built from linear-optical elements.
• Designing interferometric sensors, given the role of interferometers in quantum sensing.
• Interaction-free measurement protocols, which the paper cites as a canonical interferometric setup.
• Certified randomness or nonclassicality tests based on contextuality inequalities rather than interference alone.

Industry relevance. Any hardware platform whose advantage claims rest on linear-optical interference — photonic quantum computing, boson sampling demonstrators, and precision interferometry — benefits from knowing which observable signatures actually rule out classical or other non-quantum descriptions of the device.

Future Directions

1. Complete and analyse the contextuality comparison. The joint numerical range construction (Fig. 11) is set up but its quantitative output is not available in the provided content; establishing which theories violate the Clauser-Horne-Shimony-Holt-derived contextuality inequalities, and by how much, is the natural next test of a genuinely qualitative separation.

2. Extend beyond the three theories and the two-mode case. The framework is presented as a foundation for generalised theories of complex photonic setups; larger circuits, more modes, and other phase-space theories remain to be explored.

3. Exploit negative phase-space states in sawtooth theory. The authors explicitly note that the $V = 2/\pi \approx 64%$ ceiling applies only to the delta-distributed states they chose, and that negative phase-space states might reach higher visibilities at the cost of moving further from classical theory.

4. Incorporate sources, detectors and loss more completely. The abstract positions the work as a foundation for modelling photonic quantum computers; a full operational account would need to treat non-ideal components in the same language as beam splitters and phase shifters.

Target Audience

Quantum foundations researchers working on generalised probabilistic theories and continuous-variable systems; quantum optics experimentalists and theorists interested in what linear-optical interference actually certifies; and photonic quantum computing / boson sampling teams who need to justify nonclassicality claims. Readers should be comfortable with phase-space methods and linear-optical transfer matrices, though the comparative results and figures are usable by a broader physics audience.

Authors’ abstract

We use the analogy between the quadrature operators in quantum optics and the phase-space coordinates in a mechanical harmonic oscillator to generalise the theory of linear optics beyond quantum theory. For this, we introduce a framework for analysing photonic circuits within generalised probabilistic theories based on quadrature operators. Using this framework, we ask whether phenomena like the Hong-Ou-Mandel effect, Mach-Zehnder interferometry are specific to quantum theory or also occur across a broader class of theories. We find the latter: the vanishing photon coincidences at the output of a beam splitter also occur for classical light and for a modified version of quantised light. Similarly, for Mach-Zehnder interferometry we find that there is no qualitative difference between quantum theory and these alternative theories. Our framework can serve as a foundation to study generalised theories of complex photonic set-ups and photonic quantum computers.