Generalised quantum interference
Sebastian Malewicz, Laura Serino, Marcelo P. de Almeida, Markus Rambach, Ming Su, Jihun Cha, Till J. Weinhold, Austin Lund, A. G. White · 2026
Read Generalised quantum interference by Sebastian Malewicz, Laura Serino, Marcelo P. de Almeida, Markus Rambach, Ming Su, Jihun Cha, Till J. Weinhold, Austin Lund, A. G. White (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-31053.txt.
Introduction
Malewicz and eight colleagues report an experiment that breaks a famous symmetry of quantum optics. In the Hong-Ou-Mandel effect, two identical photons entering a beam splitter always leave together through the same port, because destructive interference cancels the outcome where they split up. The team shows this is only one possibility within a broader family of two-photon interference.
Using a programmable integrated photonic processor, they interfere imbalanced two-photon states on a variable beam splitter and tune the output distribution continuously. At the extreme, destructive interference suppresses bunching at one output port instead: the photons may leave separately or bunch at the other port, but never together at the forbidden one. The result widens the known toolbox for characterizing photon sources and building quantum sensors.
Catalog record
| CREATOR | Sebastian Malewicz, Laura Serino, Marcelo P. de Almeida, Markus Rambach, Ming Su, Jihun Cha, Till J. Weinhold, Austin Lund, A. G. White |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 17 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.31053
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Generalised quantum interference Sebastian Malewicz, Laura Serino, Marcelo P. de Almeida, Markus Rambach, Ming Su, Jihun Cha, Till J. Weinhold, Austin Lund, A. G. White Abstract Hong-Ou-Mandel interference is one of the clearest signatures of quantum behaviour: two identical photons meeting at a beam splitter always leave together. Here we demonstrate that this symmetry can be broken. Using a programmable integrated photonic processor, we interfere imbalanced two-photon states on a variable beam splitter and show that the output state of quantum interference can be continuously tuned, reaching the extreme case where destructive interference suppresses bunching at one output port. The asymmetry creates a forbidden bunching channel: photons may exit together at one port or separately at both ports, but cannot emerge together at the other port. The Hong-Ou-Mandel (HOM) effect 14 ; 31 ; 10 ; 33 is a quintessential manifestation of quantum interference 6 ; 9 . When two indistinguishable photons impinge on different input ports of a balanced beam splitter, they interfere such that they always leave together. Since this behaviour depends directly on photon indistinguishability, HOM interference has evolved from a foundational demonstration of nonclassical physics into a standard tool for assessing indistinguishability 37 ; 34 ; 18 ; 16 ; 13 , characterising photon sources 17 ; 15 ; 5 ; 25 ; 28 ; 35 ; 4 ; 36 ; 27 and optical components and interfaces 32 ; 41 ; 1 ; 12 , and performing Bell-state analysis 19 ; 7 ; 3 . Beyond characterisation, HOM-based techniques have also found practical application in high-resolution imaging and microscopy 30 ; 39 , precision metrology 22 ; 8 ; 26 and quantum-enhanced sensing 29 ; 2 ; 23 . Nevertheless, the familiar HOM effect is not the only possibility for two-photon quantum interference. In the standard case, destructive interference suppresses the coincidence outcome, in which the photons leave in different output ports. More generally, however, linear optics allows the interference to be controlled by changing the beam splitter transformation, enabling generalised quantum interference 21 . In this framework, destructive interference can be tuned to the extreme case in which the suppressed output is instead one of the two bunching terms, giving rise to asymmetric quantum interference. For suitably prepared two-photon input states, the photons can emerge separately or bunch in one output port, but not bunch in the other. Here we experimentally demonstrate generalised quantum interference on a programmable integrated photonic processor 24 . We prepare imbalanced two-photon states and interfere them on a variable beam splitter, showing how the full two-photon output distribution can be reshaped by tuning the input-state imbalance and the beam splitter reflectivity. Within this family, we realise the asymmetric quantum interference regime in which one bunching output is suppressed while the remaining bunching output and the coincidence output are equally probable. Figure 1: Conceptual comparison of Hong-Ou-Mandel (HOM) interference and generalised two-photon quantum interference at the asymmetric point. (a) In HOM interference, two indistinguishable photons entering different input ports of a balanced beam splitter interfere such that the coincidence output is suppressed, leaving a superposition of the two bunched outputs. (b) In asymmetric quantum interference, an imbalanced two-photon input interferes on a beam splitter such that one of the two bunching outputs is suppressed, while the coincidence output and the other bunching output remain equally probable. For two photons in two modes—such as the two possible output paths of a beam splitter—the possible output configurations are \ket{1,1} , in which the photons leave in different ports, and \ket{2,0} and \ket{0,2} , in which they bunch into the same port. In the standard HOM effect, two indistinguishable photons entering different input ports of a balanced beam splitter interfere such that the \ket{1,1} term is suppressed, leaving a superposition of the two bunched outputs (Fig. 1 a). HOM interference is therefore associated with destructive interference in the coincidence channel. Ref. 21 showed that the standard HOM effect is only one particular case of a broader class of generalised two-photon interference effects accessible through linear optics. More generally, two-photon states can be classified into equivalence classes, such that two states in the same class can be transformed into each other using only linear optical elements. Each class has a canonical form, which in the two-mode case can be written as the superposition of bunched states a\ket{2,0}+\sqrt{1-a^{2}}\ket{0,2}\,, (1) with 0\leq a\leq 1 . By preparing a two-photon input state in this form and then interfering it on a variable beam splitter, as shown in Fig. 1 , one can access any target state within the same equivalence class. This means that, starting from input states with different imbalance coefficients a and varying the beam splitter reflectivity and the phase shifts, one can continuously tune the relative weights and phases of the three output terms, thereby obtaining an arbitrary two-photon, two-mode state 21 . In particular, for a=(\sqrt{3}+1)/2\sqrt{2}\approx 0.966 and beam splitter reflectivity R=1/(3-\sqrt{3})\approx 78.9\% with a transmission phase shift of \pi/2 , the output becomes \ket{\psi_{\mathrm{out}}}=\frac{1}{\sqrt{2}}\left(\ket{2,0}+i\ket{1,1}\right), (2) so that the \ket{0,2} term is suppressed, while the \ket{2,0} and \ket{1,1} terms are equally probable (Fig. 1 b). By symmetry, the complementary reflectivity R\approx 21.1\% would suppress the opposite bunching term \ket{2,0} . Figure 2: Optical circuit implemented in the photonic processor Ascella. Two single photons interfere on a 50:50 beam splitter to produce a two-photon bunched state. A beam splitter with reflectivity 1-\gamma introduces loss in one of the two channels, thereby preparing the unbalanced input state of Eq. ( 1 ), which is interfered on a variable beam splitter with reflectivity R . The three possible two-photon outputs are resolved using pseudo-photon-number-resolving (PPNR) detectors implemented with a 50:50 beam splitter in each output channel and detector pairs D_{0} - D_{1} and D_{3} - D_{4} , following the mode numbering of the implemented subcircuit. To realise generalised quantum interference experimentally, we use the programmable photonic quantum processing unit (QPU) Ascella 24 . We program the chip such that two single photons are first interfered on a balanced beam splitter to generate a HOM bunched state, after which controlled attenuation in one arm prepares the imbalanced input of Eq. ( 1 ). This state is then interfered on a variable beam splitter, and the three two-photon outputs are resolved using pseudo-photon-number-resolving (PPNR) detection. The five-mode subcircuit of Ascella used for this experiment is shown in Fig. 2 . We begin with two single photons generated by Ascella’s quantum dot source and temporal demultiplexer 24 . Interfering these photons on a balanced beam splitter produces the typical HOM bunched state, and controlled attenuation of one arm then prepares the imbalanced superposition. After postselection on two-photon events, the resulting state at this point takes the form of Eq. ( 1 ) with a=\left[1+(1-\gamma)^{2}\right]^{-1/2} , where \gamma is the transmissivity of the beam splitter that couples one arm to a loss channel. The state is then sent to a variable beam splitter with reflectivity R . To resolve the three possible two-photon outputs, we use PPNR detection at the outputs of the variable beam splitter. Each output arm is split on an additional balanced beam splitter and monitored by a pair of superconducting nanowire detectors, so that coincidences within one