Revealing and concealing entanglement with non-inertial motion

Marko Toroš,1, ∗ Sara Restuccia,2 Graham M. Gibson,2 Marion Cromb,2 Hendrik Ulbricht,3 Miles Padgett,2 and Daniele Faccio2, † 1Department of Physics and Astronomy, University College London, Gower Street, WC1E 6BT London, UK 2School of Physics and Astronomy, University of Glasgow, Glasgow, G12 8QQ, UK 3Department of Physics and Astronomy, University of Southampton, SO17 1BJ, UK Photon interference and bunching are widely studied quantum effects that have also been pro- posed for high precision measurements. Here we construct a theoretical description of photon- interferometry on rotating platforms, specifically exploring the relation between non-inertial mo- tion, relativity, and quantum mechanics. On the basis of this, we then propose an experiment where photon entanglement can be revealed or concealed solely by controlling the rotational motion of an interferometer, thus providing a route towards studies at the boundary between quantum mechanics and relativity.

Introduction. The notions of space and time are at covers the results for the Sagnac effect in the quantum the core of modern physics and remain an area of in- regime [17] as well as for the recent demonstration of tense research [1, 2]. A striking example of how elemen- photon bunching in a rotating reference frame [16]. tary notions of space and time lead to surprising conse- Theoretical model. We consider an experimental quences is the derivation of Lorentz transformations, a platform rotating at angular frequency Ω depicted in cornerstone of quantum field theory, utilizing only basic Fig. 1(a). The system is confined to move on a circle assumptions [3, 4]. of radius r in the equatorial plane normal to the rotation The exploration of the special-relativistic regime is his- axis. We further suppose there is a co-rotating medium torically strongly linked to investigations of the prop- with refractive index n. For the co-rotating observer, agation of light [5], e.g. the Michelson–Morley exper- the light propagation speed is c/n in both directions as iment [6]. More recent experiments have also started can be deduced from symmetry considerations. It is also to probe the quantum nature of light, e.g. the Hong- instructive to describe the same experiment from the in- Ou-Mandel (HOM) experiment [7], indirectly testing the ertial frame of the laboratory: one can formally map the underpinning spacetime symmetries. Quantum optical circular trajectories to straight-line motions as shown in interference effects, either one-photon or two-photon, are Fig. 1(b) [18]. In this latter case one has to account for thus of fundamental importance [8], as well as providing the light-dragging effect [13], i.e. the Fizeau effect (see paths to technological applications [9]. supplementary material). A further test of special relativity, moving towards the To describe quantum mechanically the evolution of the domain and ideas of general relativity, is possible in sit- system we choose for convenience the viewpoint of the uations where linear acceleration or rotational motion is co-rotating observer (see Fig. 1(a)). To account for the present [10]. A notable example is the classical Sagnac non-inertial motion when Ω 6= 0 we start from the Born experiment where an interferometer is placed on a rotat- chart and exploit the methods of symplectic Hamiltonian ing platform [11–13]. More recent experiments include mechanics [19, 20]. In particular, exploiting the so-called experimental tests of photonic entanglement in accel- co-moment map from the generators of the Poincaré al- erated reference frames [14], the demonstration of how gebra to Hilbert space operators one finds the following to overcome the shot-noise limit using an entanglement- Hamiltonian (see supplementary material): enhanced optical gyroscope [15] and the extension of ˆ ˆ ˆ HOM interference to rotating platforms [16]. HBorn = Γ(H + rΩP ), (1) In this letter, we propose a new experimental plat- 1 Ωr 2 − 2 where Γ = (1−( c ) ) is the Lorentz factor. The term arXiv:1911.06007v1 [quant-ph] 14 Nov 2019 form based on the Mach-Zehnder interferometer that ex- ˆ plores the relation between interference, entanglement, ∼ P keeps track of the non-inertial motion of the detec- ˆ and non-inertial rotational motion. In particular, we dis- tor, P , which is the generator of translations, changes cover that by simply setting the apparatus in rotational the relative distance between the detector and the sys- motion one can detect or conceal entanglement. We first tem, e.g. in a time δt the relative distance changes by provide a theoretical description of quantum experiments rΩδt. on a rotating platform starting from the Hamiltonian on We now apply the Hamiltonian in Eq. (1) to photon- a generic Hilbert space which we then apply to study interferometry. We use the Abraham relation between photon-interferometry experiments. The model also re- kinetic momentum and energy [21, 22]: Hˆ = nc|Pˆ|, (2) We combine Eqs. (1) and (2) to find: ∗ [email protected] † [email protected] ˆ (±) ˆ HBorn = Γ (1 ± β) H, (3) 2

D1D2 (-) To analyze photon-interferometry experiments we will (+) (-) t t(+) (a) (b1) i (-) vv (+) i work in the Schrödinger picture [26], where we will de- H Ω H D1 D2 note the initial (final) state with the subscripts i (f). We (-) (+) consider the experimental situation where at time ti = 0 (b2) t a t c/n c/n v (-) (+) a D1 vv D2 v the photon is prepared in a state |ψii, and then con- counter co-rotating L strained to move in a circular motion for a time tf = n rotating c resulting in a state |ψf i, where L is the traveled distance. Although one can always postulate a given initial state Figure 1. Conceptual setup. (a) Description from the view- point of the co-rotating observer. The counter-rotating (co- it is nonetheless instructive to compare the state |ψii, rotating) quantities are shown on the left (right). The which is assumed to be generated by the apparatus co- counter-rotating (co-rotating) Hamiltonians are different, rotating with the platform, with the state generated by while the light-speed in the two directions is equal (c/n). the same apparatus when the platform is not rotating, D1 (D2) indicate the detectors for the counter-rotating (co- i.e. when Ω = 0. In particular, it is reasonable to assume rotating) direction, respectively. (b) Description from the that the frequencies of the initial states generated in the inertial laboratory frame represented in a straight line (see Ω two experimental situations differ by at most ∼ 2π . How- supplemental material). Here both the Hamiltonian and the ever, such a difference in the initial state produces only (−) (+) speed of light differ in the two directions, i.e. v 6= v sub-leading effects which are not amplified during time- and v(±) 6= c . (b1) Only the light-drag effect is taken into n evolution, as can be explicitly verified using the formulae (+) (−) account; ti , ti denote how long it would take to reach the we will develop. One can thus approximate the initial detectors assuming they would not have moved; the subscript states generated on the rotating platform with the states i stands for “initial”. (b2) Both light-drag effect and the mo- (+) (−) that would be generated at Ω = 0. tion of detectors is taken into account. ta , ta denote how The time-evolution is given by the usual Schrödinger long it takes to reach the detectors; the subscript a stands for equation with the Hamiltonian in Eq. (4), i.e. “actual”. v = rΩ is the speed of the detectors as seen from the inertial laboratory reference frame. Z (2) h (+) † (−) † i Hˆ =~ dω ω aˆ (ω)ˆa(ω) + ω ˆb (ω)ˆb(ω) , (5) where the positive (negative) superscript denotes the (±) ˆ counter-rotating (co-rotating) motion to be measured by where we have defined ω = (1 ± β)ω, and aˆ (b) is the rΩ counter (co-rotating) mode. the detector D1 (D2), and β = nc . We note that Eq. (3) suggests a simple physical interpretation of the Hamil- The state |ψf i then interferes at a beam-splitter and tonian as the relativistic Doppler-shifted energy of the one measures the outputs using two-detectors: the input ˆ system. modes are aˆ and b and we denote the output modes by We can finally write the total Hamiltonian of the sys- cˆ and dˆ. Here we consider the following relation between tem: the input and output modes: Hˆ = Hˆ (+) ⊗ + ⊗ Hˆ (−) , (4)  cˆ(ω)  1  1 1   aˆ(ω)  Born Born I I Born = √ . (6) dˆ(ω) 2 1 −1 ˆb(ω) where we have assumed that the photons moving in oppo- site directions do not interact, and I denotes the identity In particular, we are interested in the probability of de- operators. In summary, the total Hilbert space can be ˆ (+) (−) (+) (−) tecting photons in the modes cˆ or d. To this end, it is written as H = H ⊗ H , where H (H ) denotes convenient to define the temporal modes [27]: the Hilbert space of the counter-rotating (co-rotating) modes. ˆ ˆ cˆ(t) = Ft[ˆc(ω)], d(t) = Ft[d(ω)], (7) The time parameter t that keeps track of the dynamics through Schrödinger’s equation is ticking as a clock fol- where F [ · ] = √1 R dω · e−iωt. In particular, we define t 2π lowing the detectors’ motion; this is a direct consequence the single-photon probability of detection as of the quantization procedure that leads to the Hamilto- Z nian in Eq. (4). However, here we are mainly interested (1) † in the dominant effects where one can approximate the Pc = dthψf |cˆ (t)ˆc(t)|ψf i, (8) Lorentz factor as Γ ∼ 1 and the distances and times co- incide to those measured by a ruler and a clock in the (1) with a similar definition for the probability Pd for the inertial laboratory frame. output mode dˆ. In addition, we also define the the two- Photon-interferometry experiments. We now fur- photon probability of detection ther develop the model by adopting Glauber’s theory of photo-detection [23]. Here we will focus on the experi- Z Z (2) ˆ† † ˆ mental situation of photons with a coherence time that P = dt1 dt2hψf |d (t1)ˆc (t2)ˆc(t2)d(t1)|ψf i, (9) is short compared to the time resolution of the detec- tors, but temporal aspects could be easily taken into ac- which gives the coincidence probability. For the case count [24, 25]. P (2) < 0.5 we speak of coalescence or HOM photon 3 bunching and for P (2) > 0.5 we speak of photon anti- coalescence or anti-bunching. Classically, one is lim- ited to values 0.25 < P (2) < 0.5, making coincidence PD2 PD1 probabilities a valuable tool to assess the quantum na- BS ture of the electromagnetic field. Importantly, anti- symmetrization and photon anti-coalescence reveals hid- Ω BS den entanglement, as has already been demonstrated ex- b a perimentally in a non-rotating setup [28]. BBO We consider first the experimental situation with a generic single-photon input state: laser Z h † ˆ† i |ψf i = dω ψa,f (ω)ˆa (ω) + ψb,f (ω)b (ω) |0i, (10) Figure 2. Layout of proposed quantum Sagnac/Hong-Ou- where ψa,f (ω), ψb,f (ω) are one-photon wavefunctions. Mandel interferometer on a rotating platform. Two entangled From Eqs. (8) and (10), exploiting Eqs. (6) and (7), we photons are emitted from the BBO crystal: photon a (purple find arrow) enters a Sagnac interferometer and exits towards the Z upper 50/50 beamsplitter (BS) where 2-photon HOM interfer- (1) 1 1  ∗  ence occurs with photon b (green arrow) that circles around P = ± dω ψa,f (ω)ψb,f (ω) + c.c. , (11) c,d 2 2 the setup (in order to maintain the same overall path length as photon a). Coincidence counts are measured between de- where we have imposed the normalization of the state, i.e. tectors PD1 and PD2 as a function of the rotation frequency hψf |ψf i = 1. As an example let us consider the Quan- Ω. tum Sagnac experiment [17]: a photon is prepared in a superposition of counter-propagating modes before inter- fering at the beam splitter (see supplementary material). Using Eq. (14) we then immediately find the coincidence Specifically, we consider the initial state in Eq. (10) with probability: the one-photon wavefunctions ψa,i(ω) = ψb,i(ω) = g(ω), 1 1 2 2 where g(ω) is a Gaussian with mean frequency µ and P (2) = − e−σ (ts−δt) . (16) bandwidth σ. After the time-evolution using Eq. (5) we 2 2 iωβt have the state in Eq. (10) with ψa,f (ω) = g(ω)e f and where ts is the classical Sagnac delay. −iωβt ψb,f (ω) = g(ω)e f . Using Eq. (11), and making the In the previous paragraph we have considered identi- further approximation |g(ω)|2 ∼ δ(µ − ω), we find the cal photons with a separable spectrum [7], but one could single-photon detection probability: also consider identical frequency-entangled photons. For example, if we consider spontaneous parametric down (1) 1 Pc,d = (1 ± cos (µts)) , (12) conversion (SPDC) type I two-photon generation [29] we 2 again find the coincidence probability in Eq. (16). It 2 where ts = 8πAf/c is the classical Sagnac delay, f = would thus seem that entanglement in combination with Ω/2π is the rotation frequency, A = πr2 is the encircled rotational motion leaves no trace on the photon coinci- area, r is the circle radius. dence rate, P (2), measurement. We now further explore We next consider the two-photon state this question. Z Z Manifestation of entanglement through rota- † ˆ† |ψf i = dω1 dω2 ψf (ω1, ω2)ˆa (ω1)b (ω2)|0i, (13) tion. Entanglement can manifest itself in a HOM co- incidence rate measurement through anti-coalescence, i.e. P (2) > 0.5. In particular, for a completely anti- where ψf (ω1, ω2) is the two-photon spectrum. From Eqs. (9) and (13), exploiting Eqs. (6), (7), we find symmetric spectrum, i.e. ψ(ω1, ω2) = −ψ(ω2, ω1), one obtains perfect anti-coalescence, but even with a par- 1 1 Z tially anti-symmetric spectrum one can have P (2) > 0.5, P (2) = − dωψ∗(ω , ω )ψ (ω , ω ), (14) 2 2 f 1 2 f 2 1 thus witnessing entanglement. As we show below, this manifestation of entanglement may be susceptible to the where we have imposed the normalization hψf |ψf i = 1. motion of the interferometer. As an example we consider the Hong-Ou-Mandel exper- We consider the experimental setup depicted in Fig. 2. iment on a rotating platform [16]: two identical photons As an initial state we consider a SPDC type I two-photon counter-propagate before interfering at a beam-splitter state (i.e. two photons with the same polarisation) with (see supplementary material). The experimentalist con- spectrum trols the initial time-delay δt of the mode aˆ; the initial state is given by Eq. (13) with the two-photon spectrum ψi(ω1, ω2) = δ(ω1 + ω2 − 2µ)g(ω1)g(ω2), (17) −iω1δt ψi(ω1, ω2) = g(ω1)g(ω2)e . After the time-evolution we find the final state in Eq. (13) with where we have omitted the normalization. We note that the initial spectrum in Eq. (17) is completely symmetric, −iω1δt iβ(ω1−ω2) (2) ψf (ω1, ω2) = g(ω1)g(ω2)e e . (15) i.e. ψ(ω1, ω2) = ψ(ω2, ω1), and gives P = 0 at Ω = 0. 4

1 The shaded region corresponding to P (2) > 0.5 indicates the presence of entanglement that manifests as photon anti-coalescence. Short bandwidth, i.e. long coherence photons show a periodic series of revivals of entangle- ment with increasing rotation frequency. For larger pho- ton bandwidths, i.e. shorter coherence lengths, increas- 0.5 ing the relative photon delay by increasing the rotation speed leads to a reduction of the coincidence peak val- ues and of the overall fringe visibility. This is a result of coincidence probability probability coincidence the loss of mutual coherence between the two interfering photons. 0 The anti-symmetrization of the photon spectrum, 0 5 10 15 20 25 30 which leads to a modification of the coincidence prob- Ω (Hz) ability, is a direct consequence of the non-inertial motion of platform. More generally, this shows that rotational Figure 3. Coincidence plot as a function of the angular motion can activate dormant asymmetries in the experi- frequency Ω = 2πf. We have set the interferometer area mental setup leading to an anti-symmetric spectrum. It 2 15 A = 22.7m , µ = 2.36 × 10 Hz, corresponding to a typical is also interesting to consider an initial anti-symmetric photon carrier wavelength of 800 nm. Two curves are shown 13 spectrum ψi; the proposed experiment shows that ψi can for two different bandwidths, σ = 1.47 × 10 Hz (blue solid become symmetrized during time-evolution, fully con- curve) and σ = 1.18×1014Hz (dashed red curve), correspond- ing to 5 nm and 40 nm bandwidths, respectively. The shaded cealing the anti-coalescence signature of entanglement. region corresponding to P (2) > 0.5 indicates the region where These effects can be traced to the impossibility of clock measurements imply photon entanglement. synchronization along a closed loop on the rotating plat- form. In particular, the Hamiltonian in Eq. (5) can be linked to the effect of clock desynchronization [10]. It is We now consider the same setup with the interferome- important to note that this is a genuine relativistic effect, ter in a constant rotational motion with frequency Ω 6= 0. which is not expected to arise in a Newtonian theory, al- The final spectrum of the two-photon state changes to though it imprints a non-negligible experimental trace in the regime typically associated with the latter. This is different from the observer-dependent entanglement ef- −iβω2tf ψf (ω1, ω2) = ψi(ω1, ω2)cos(βω1tf )e , (18) fect in non-inertial reference frames [30], expected to arise as a consequence of the Unruh radiation [31, 32] which where the factor cos(βω1tf ) results from the interference vanishes at low accelerations. of the mode aˆ; by ‘final state’ we again mean the state Conclusions. We have developed a formalism that arrives at the last beam-splitter. Using Eq. (14) we for describing interferometry experiments on rotat- then find the coincidence probability ing platforms. We have first analyzed two recent photon-interferometry experiments, namely, the quan- 1 2 2 1  1 2 2  − 8 σ ts − 2 σ ts tum Sagnac and the Hong-Ou-Mandel experiment on a 1 cos(µts)e + 2 1 + e P (2) = − . (19) rotating platform. We have then proposed a modified − 1 σ2t2 2 2(1 + cos(µts)e 8 s ) Hong-Ou-Mandel interferometer where entanglement can be revealed or concealed depending on the rotational fre- We have plotted P (2) as a function of the angular fre- quency. These results indicate new directions for inves- quency Ω in Fig. 3 with the interferometer area A = tigating the notions of space and time as well as its con- 22.7 m2 (assuming the photons travel through a 100 m sequences in a quantum mechanical regime. long fibre, wound 35 times along a 0.9 m diameter loop) Acknowledgments. The authors acknowledge sup- and µ = 2.36 × 1015 Hz, corresponding to a typical pho- port from the EU H2020FET project TEQ (Grant ton carrier wavelength of 800 nm. Two curves are shown No. 766900), from the EPSRC (UK, Grant No. for two different photon bandwidths, σ = 1.47 × 1013 Hz EP/M009122/1) and from the European Union’s Hori- (blue solid curve) and σ = 1.18 × 1014 Hz (dashed red zon 2020 research and innovation programme under grant curve), corresponding to 5 nm and 40 nm, respectively. agreement No 820392.

[1] Hermann Minkowski. Raum und zeit. Jahresbericht der [3] WA Von Ignatowsky. Einige allgemeine bemerkungen zum Deutschen Mathematiker-Vereinigung, vol. 18, p. 75-88, relativitätsprinzip. Verh. Deutsch. Phys. Ges, 12:788–796, 18:75–88, 1909. 1910. [2] Vesselin Petkov et al. Minkowski spacetime: A hundred [4] Stefano Liberati. Tests of lorentz invariance: a 2013 years later, volume 11. Springer, 2010. update. Classical and Quantum Gravity, 30(13):133001, 5

2013. photon interference. Advances In Atomic, Molecular, and [5] Howard P Robertson. Postulate versus observation in the Optical Physics, 53:253–289, 2006. special theory of relativity. Reviews of modern Physics, [25] Holger P Specht, Jörg Bochmann, Martin Mücke, Bern- 21(3):378, 1949. hard Weber, Eden Figueroa, David L Moehring, and Ger- [6] Albert A Michelson and Edward W Morley. On the rel- hard Rempe. Phase shaping of single-photon wave packets. ative motion of the earth and of the luminiferous ether. Nature Photonics, 3(8):469, 2009. Sidereal Messenger, vol. 6, pp. 306-310, 6:306–310, 1887. [26] Kaige Wang. Quantum theory of two-photon wavepacket [7] Chong-Ki Hong, Zhe-Yu Ou, and Leonard Mandel. interference in a beamsplitter. Journal of Physics B: Measurement of subpicosecond time intervals between Atomic, Molecular and Optical Physics, 39(18):R293, two photons by interference. Physical review letters, 2006. 59(18):2044, 1987. [27] KJ Blow, Rodney Loudon, Simon JD Phoenix, and [8] Leonard Mandel. Quantum effects in one-photon and two- TJ Shepherd. Continuum fields in quantum optics. Phys- photon interference. In More Things in Heaven and Earth, ical Review A, 42(7):4102, 1990. pages 460–473. Springer, 1999. [28] Alessandro Fedrizzi, Thomas Herbst, Markus As- [9] Ashley Lyons, George C Knee, Eliot Bolduc, Thomas pelmeyer, Marco Barbieri, Thomas Jennewein, and Anton Roger, Jonathan Leach, Erik M Gauger, and Daniele Fac- Zeilinger. Anti-symmetrization reveals hidden entangle- cio. Attosecond-resolution hong-ou-mandel interferome- ment. New Journal of Physics, 11(10):103052, 2009. try. Science advances, 4(5):eaap9416, 2018. [29] Marco Barbieri, Emanuele Roccia, Luca Mancino, Marco [10] Éric Gourgoulhon. Special relativity in general frames. Sbroscia, Ilaria Gianani, and Fabio Sciarrino. What hong- Springer, 2016. ou-mandel interference says on two-photon frequency en- [11] Georges Sagnac. L’éther lumineux démontré par l’effet tanglement. Scientific reports, 7(1):7247, 2017. du vent relatif d’éther dans un interféromètre en rotation [30] Ivette Fuentes-Schuller and Robert B Mann. Alice falls uniforme. CR Acad. Sci., 157:708–710, 1913. into a black hole: entanglement in noninertial frames. [12] Georges Sagnac. Sur la preuve de la réalité de l’éther Physical review letters, 95(12):120404, 2005. lumineux par l’expérience de l’interférographe tournant. [31] Paul CW Davies. Scalar production in schwarzschild and CR Acad. Sci., 157:1410–1413, 1913. rindler metrics. Journal of Physics A: Mathematical and [13] Evert Jan Post. Sagnac effect. Reviews of Modern General, 8(4):609, 1975. Physics, 39(2):475, 1967. [32] William G Unruh. Notes on black-hole evaporation. [14] Matthias Fink, Ana Rodriguez-Aramendia, Johannes Physical Review D, 14(4):870, 1976. Handsteiner, Abdul Ziarkash, Fabian Steinlechner, Thomas Scheidl, Ivette Fuentes, Jacques Pienaar, Timo- thy C Ralph, and Rupert Ursin. Experimental test of pho- tonic entanglement in accelerated reference frames. Nature communications, 8:15304, 2017. [15] Matthias Fink, Fabian Steinlechner, Johannes Hand- steiner, Jonathan P Dowling, Thomas Scheidl, and Rupert Ursin. Entanglement-enhanced optical gyroscope. New Journal of Physics, 21(5):053010, 2019. [16] Sara Restuccia, Marko Toroš, Graham M. Gibson, Hen- drik Ulbricht, Daniele Faccio, and Miles J. Padgett. Pho- ton bunching in a rotating reference frame. Phys. Rev. Lett., 123:110401, Sep 2019. [17] Guillaume Bertocchi, Olivier Alibart, Daniel Barry Os- trowsky, Sébastien Tanzilli, and Pascal Baldi. Single- photon sagnac interferometer. Journal of Physics B: Atomic, Molecular and Optical Physics, 39(5):1011, 2006. [18] Angelo Tartaglia and Matteo Luca Ruggiero. The sagnac effect and pure geometry. American Journal of Physics, 83(5):427–432, 2015. [19] Ana Cannas Da Silva and A Cannas Da Salva. Lectures on symplectic geometry, volume 3575. Springer, 2001. [20] Peter Woit, Woit, and Bartolini. Quantum Theory, Groups and Representations. Springer, 2017. [21] Miles J Padgett. On diffraction within a dielectric medium as an example of the minkowski formulation of optical momentum. Optics express, 16(25):20864–20868, 2008. [22] Stephen M Barnett. Resolution of the abraham- minkowski dilemma. Physical review letters, 104(7):070401, 2010. [23] Roy J Glauber. The quantum theory of optical coherence. Physical Review, 130(6):2529, 1963. [24] Thomas Legero, Tatjana Wilk, Axel Kuhn, and Gerhard Rempe. Characterization of single photons using two- 1

Supplementary material to “Revealing On the other hand, in the laboratory inertial reference and concealing entanglement with frame, one has to account for the motion of the detectors non-inertial motion” as well as of the medium (again through the Hamiltoni- ans). In short, the advantage of the co-rotating reference frame is the absence of the light-dragging effect. S1: LABORATORY REFERENCE FRAME The Sagnac delay can be obtained also from the per- spective of the co-rotating observer where it arises from In this section we briefly discuss the relation between clock desynchronization [S2]. the co-rotating reference frame and the inertial labora- tory reference frame, focusing on the classical effects. We start by noting that the speed of light in the co-rotating S2: DERIVATION OF THE HAMILTONIAN reference frame, where the co-rotating medium is station- ary, is the same in the co-rotating and counter-rotating c In this section we derive the Hamiltonian for the ex- directions; specifically we have that the velocities are ± n . periments depicted in Fig. 1(a): we consider a rotating Using the relativistic velocity addition formula we find (±) platform, which spins at angular frequency Ω, and we the corresponding velocities u in the inertial labora- restrict to the dynamics on a circle of radius r, centered tory frame: on the symmetry axis of the rotating platform. Specif- ically, we will adopt the methods of representation the- ± c + v c ory [S3] and symplectic Hamiltonian mechanics [S4] to u(±) = n ≈ ± + αv, (S1) 1 ± c v n map the time-evolution generator of the Poincaré algebra n c2 to a Hilbert space operator. One could of course make where v = rΩ, and α = 1− 1 . The corresponding speeds an ad-hoc quantization in a non-inertial reference frame, n2 and obtain a Hamiltonian, but the results might be incon- are given by v(±) ≈ c ± αv. See Fig. 1(b) for a graphical n sistent with basic symmetry requirements. Anyhow, we illustration of this light-dragging effect, which coincides choose the former method which constructs the Hamilto- with the Fizeau effect, but could in principle differ [S1]. nian starting from basic symmetry considerations of the We assume an initial spatial-distance L between the Poincaré group. As argued in the supplemental material systems and the detectors. We can convert L into a time- S1 we will for convenience describe the experiment in the distance, i.e. the initial time-distance (see Fig. 1(b1)), co-rotating reference frame. (±) L which is given by ti = v(±) . Furthermore, exploiting One typically starts describing the experiments by set- Eq. (S1), we find: ting up a chart, e.g. a Cartesian chart. We note that the chosen spacetime coordinates critically reflect the motion of the observer which affects the resulting description of (±) L vL t ≈ n ∓ αn2. (S2) the dynamics. For example, two observers moving with i c c2 different speeds or accelerations will use different charts, However this is different from the actual time it takes the and hence ascribe different energies to the same system, signals to reach the detectors (see Fig. 1 (b2)): we need and hence care must be taken with the choice of the co- to take into account also the motion of the detectors. In ordinate system. In the following we will assume that (±) (±) (±) particular we have the condition v ta = L ± vta , the detectors are stationary in the observer’s chart: with which after some algebra readily gives this choice there is a simple relation between observables in the description and the quantities measured by the detectors. L L Lv t(±) = ≈ n ± . (S3) We start by recalling the quantization procedure in an a (±) 2 v ∓ v c c inertial reference frame, i.e. with Ω = 0. In our case, we In this way we immediately recover the classic Sagnac will use the polar chart for the laboratory inertial refer- (+) (−) ence frame. Specifically, the line-element in an inertial delay given by t = t − t = 2Lv . Specifically, to s a a c2 reference frame expressed in the polar chart is given by find the usual expression of the Sagnac delay we set L = 2πrN, where N denotes the winding number, and define the encircled area as A = Nπr2. Using v = rΩ = r2πf ds2 = c2dt2 − r2dφ2, (S5) we then immediately find [S1, S2]:

8πAf where r is a constant in our case (we have restricted the ts = . (S4) analysis to a 1 + 1 spacetime). From the line-element in c2 Eq. (S5) it is then possible to immediately read the time- 1 µ From this discussion it is clear why the description in evolution Killing vector ( c ∂t) [S5]. The time-evolution the co-rotating reference frame is slightly more conve- Killing vector (more precisely a vector field) is the tan- nient: there only the non-inertial motion of the detectors gent vector to the flow lines of a system in free motion. has to be taken into account (through the Hamiltonians). Using the co-moment map we can then map the time- 2 evolution Killing vector to the Hilbert space Hamilto- can now map the latter Killing vectors to operators on a nian [S3, S4]: Hilbert space using the usual prescription

∂t → H,ˆ (S6) ∂t → H,ˆ (S14) where Hˆ is the representation of the time-evolution gen- 1 ∂ → P.ˆ (S15) erator on the considered Hilbert space. For example, for r φ a single mode aˆ we would have Hˆ = ~ωaˆ†aˆ, where ω is the frequency of the oscillator. Exploiting Eqs. (S13)-(S15), we can now finally write the To apply the quantization procedure in a non-inertial time-evolution operator reference frame, i.e. with Ω 6= 0, we have to make an additional step: we have to relate the co-rotating ref- Hˆ = Hˆ + rΩP,ˆ (S16) erence frame to the laboratory inertial reference frame. Born There reason is simple: we do not know how to directly which we name Born Hamiltonian. Note that Eq. (S16) quantize in a non-inertial reference frame, but only in captures the idea that the dynamics on a rotating plat- the inertial reference frame. We denote the laboratory form can be fully explained in terms of the non-inertial (polar) and the co-rotating (Born) coordinates by the motion of the detector [S7]: the term rΩPˆ describes the 0 unprimed (xµ = (ct, rφ)) and primed (xµ = (ct0, rφ0)) non-inertial motion of the detector, i.e. at each instant of labels, respectively. In particular, we have the following time the detector is translated with respect to the system relation [S6]: which evolves freely on a circle. We note that the transformation in Eq. (S7) leaves the time coordinate unchanged. This can be seen as a dt = dt0, (S7) Galilean-type transformation on a circle, which we now dφ = dφ0 + Ωdt0. (S8) generalize to a Lorentz-type transformation. In particu- lar, we consider It is straightforward to find the corresponding transfor- mation matrix r2Ω dt = Γdt0 + AΓ dφ0, (S17) c2 ∂xµ  1 0  0 0 = , (S9) dφ = BΓdφ + ΓΩdt , (S18) ∂xµ0 rΩ 1 c 1 Ωr 2 − 2 where Γ = (1 − ( c ) ) . If we set A = 1 and B = 1 the and to express the Minkowski metric in the two coordi- transformation is formally equivalent to a Lorentz boost nates systems, i.e. with speed v = rΩ, while if we set A = 0 and B = Γ−1 we obtain the transformation considered by Post [S1]. We

2 2 2 2 2 find that the Hamiltonian is insensitive to the value of A, ds =c dt − r dφ (S10) but depends on the chosen value of B. In the following Ω2r2 we set B = 1 which leads to the Hamiltonian =c2(1 − )dt02 − 2Ωr2dt0dφ0 − r2dφ02. (S11) c2 ˆ rel ˆ ˆ From Eqs. (S10) and (S11) we can immediately find HBorn = Γ(H + rΩP ). (S19) 1 µ0 > 1 µ the relevant Killing vectors ( ∂ 0 ) = (1, 0) , ( ∂ ) = c t c t Eq. (S19) can be seen as a relativistic Born Hamiltonian, (1, 0)>, and ( 1 ∂ )µ = (0, 1)>. Using Eq. (S9) we can r φ which generalizes Eq. (S16). 1 µ0 then express ( c ∂t0 ) in the laboratory coordinates, i.e. We can also analyze non-uniform rotations using the above formalism by considering a time-dependent angu- µ lar frequency Ωt (we remark that the time-evolution vec- 1 µ ∂x 1 µ0 ( ∂t0 ) = 0 ( ∂t0 ) , (S12) tor does not need to be generally a Killing vector). We c ∂xµ c repeat the derivation in this section with the formal re- 1 µ rω > placements which gives ( c ∂t0 ) = (1, c ) , and thus

Ω → Ωt, (S20) 1 1 0 Ωtr 2 − ∂t = ∂t + rΩ ∂φ . (S13) Γ → Γ = (1 − ( ) ) 2 . (S21) r t c We have now expressed the time-evolution Killing vector At the end we obtain in place of Eq. (S19) the following ∂t0 , which generates the dynamics in the co-rotating ref- Hamiltonian: 1 erence frame, in terms of the Killing vectors ∂t and r ∂φ, which generate time-evolution and space-translation in ˆ rel ˆ ˆ the inertial laboratory reference frame, respectively. We HBorn(t) = Γt(H + rΩtP ). (S22) 3

tected. The Hong-Ou-Mandel experiment on a rotating (a) 1 3 c (b) c platform is depicted in Fig. S1(b): two identical photons C b 2 counter-propagate before interfering at a beam-splitter. a Ω Ω One detects the arrival of the photons and extracts the b coincidence probability, P (2). In both cases, the setups a are placed on a rotating platform. d d The proposal shown in Fig. 2 can be seen as a com- Figure S1. (a) Quantum Sagnac experiment [S8]. The ele- bination of the setups shown in Fig. S1, which can be ment C denotes the circulator which allows only the paths exploited to gain an intuitive understanding of the re- 1 to 2 and 2 to 3. (b) Hong-Ou-Mandel experiment on a sults. For example, the asymptotic value for the coinci- rotating platform [S9]. dence probability in Fig. 3 can be intuitively understood in terms of Quantum Sagnac and the HOM setups. From Eq. (19) we find at high rotation frequency the value In this case one expects two physical effects, one re- P (2) ∼ 1/4; this is halfway between a full dip, P (2) ∼ 0, lated to the (geometrical) Sagnac phase, and a possible and the case without bunching or anti-bunching, P (2) ∼ new contribution related to a dynamical phase, which 1/2. The paths denoted by the purple and blue arrows in typically emerges in situations where there is a time- Fig. 2 can be related to the Quantum Sagnac: at high ro- dependence in the Hamiltonian. tation frequency the counter-propagating mode (blue ar- row) no longer interferes with the co-rotating modes due to non-overlapping frequency spectra. Loosely speak- S3: NOTES ON EXPERIMENTAL SETUPS ing, the counter-propagating mode (blue arrow) can be heuristically associated to half of the initial mode aˆ, In this section we summarize the two experimental while the other half of the mode aˆ co-rotates (purple schemes that form the building blocks, conceptually as arrow) and interferes with the other co-rotating mode ˆb well as experimentally, for the new proposal to demon- (green arrow). As the frequency spectrum of the two strate how entanglement can be revealed or concealed co-rotating modes is identical, i.e. no time-delay exists with non-inertial motion. The Quantum Sagnac experi- between them, they bunch together at the output ports ment is depicted in Fig. S1(a): a photon enters through of the beam-splitter. Hence the counter-rotating (co- the path 1, is directed into path 2 and then interferes rotating) part of the mode aˆ is associated with P (2) ∼ 1/2 for the first time with the beam splitter. After evolving (P (2) ∼ 0). The coincidence probability at high rotation in counter-progating directions, the photon then inter- frequency can be thus seen as the average behaviour, i.e. feres again at the beam splitter, after which it is de- P (2) ∼ (0 + 1/2)/2 = 1/4.

[S1] Evert Jan Post. Sagnac effect. Reviews of Modern tions of Physics, 32(10):1525–1556, 2002. Physics, 39(2):475, 1967. [S7] Angelo Tartaglia and Matteo Luca Ruggiero. The sagnac [S2] Éric Gourgoulhon. Special relativity in general frames. effect and pure geometry. American Journal of Physics, Springer, 2016. 83(5):427–432, 2015. [S3] Peter Woit, Woit, and Bartolini. Quantum Theory, [S8] Guillaume Bertocchi, Olivier Alibart, Daniel Barry Os- Groups and Representations. Springer, 2017. trowsky, Sébastien Tanzilli, and Pascal Baldi. Single- [S4] Ana Cannas Da Silva and A Cannas Da Salva. Lectures photon sagnac interferometer. Journal of Physics B: on symplectic geometry, volume 3575. Springer, 2001. Atomic, Molecular and Optical Physics, 39(5):1011, 2006. [S5] Eric Poisson. A relativist’s toolkit: the mathematics of [S9] Sara Restuccia, Marko Toroš, Graham M. Gibson, Hen- black-hole mechanics. Cambridge university press, 2004. drik Ulbricht, Daniele Faccio, and Miles J. Padgett. Pho- [S6] Guido Rizzi and Matteo Luca Ruggiero. Space geometry ton bunching in a rotating reference frame. Phys. Rev. of rotating platforms: an operational approach. Founda- Lett., 123:110401, Sep 2019.