Randomness Amplification under Simulated PT -symmetric Evolution

Leela Ganesh Chandra Lakkaraju1, Shiladitya Mal1,2, Aditi Sen(De)1 1 Harish-Chandra Research Institute and HBNI, Chhatnag Road, Jhunsi, Allahabad - 211019, India 2 Department of Physics and Center for Quantum Frontiers of Research and Technology (QFort), National Cheng Kung University, Tainan 701, Taiwan

PT -symmetric quantum theory does not require the Hermiticity property of observables and hence allows a rich class of dynamics. Based on PT -symmetric quantum theory, various counter- intuitive phenomena like faster evolution than that allowed in standard quantum mechanics, single- shot discrimination of nonorthogonal states has been reported invoking Gedanken experiments. By exploiting open-system experimental set-up as well as by computing the probability of distinguish- ing two states, we prove here that if a source produces an entangled state shared between two parties, Alice and Bob, situated in a far-apart location, the information about the operations performed by Alice whose subsystem evolves according to PT -symmetric Hamiltonian can be gathered by Bob, if the density matrix is in complex Hilbert space. Employing quantum simulation of PT -symmetric evolution, feasible with currently available technologies, we also propose a scheme of sharing quan- tum random bit-string between two parties when one of them has access to a source generating pseudo-random numbers. We find evidences that the task becomes more efficient with the increase of dimension.

I. INTRODUCTION shown to be violated in the Gedanken experimental set-up [27]. Violation of these fundamentally signif- Among the postulates of quantum mechanics, the re- icant no-go theorems have deep consequences in the quirement of the Hermiticity property for the observ- verification of the viability of quantum theory. The ables has the least support from the perspective of phys- contradictions of PT -symmetric quantum theory has ical considerations. It was pointed out that a condition been shown to be apparent in two different ways – dictated by the fundamental discrete symmetry of the firstly, by taking care of the proper inner product of world, i.e., space-time reflection symmetry may lead to the theory, these conflicts evaporates [28–30], while in a new kind of quantum theory [1, 2] which is com- the second approach, PT -symmetric Hamiltonian was monly known as parity-time-symmetric or PT - sym- shown to be Hermitian in higher-dimensional space metric quantum theory. A Hamiltonian in this theory by employing Naimark dilation [31], which opens up has real eigenvalues in the symmetry unbroken phase the possibility of experimental simulation of the dy- and pairwise complex eigenvalues in the symmetry namics. In a recent novel experiment, a pair of spa- broken phase while in between at the exceptional point, tially separated entangled photons is employed to ad- a new type of critical behavior emerges [3–10]. This dress the problem of apparent signaling [32] in which complex extension of quantum theory not only solves a one of the two photons is evolved according to PT - long-standing problem of negative norm ’ghost’ state symmetric evolution which is effectively simulated by of renormalized Lee model in quantum field theory a joint unitary operation on the system and an auxil- [11] but also enlarges the scope of allowed dynamics, iary system and by performing a measurement on aux- thereby culminating in the exploration of varieties of iliary subsystem whereas the other spatially seperated rich phenomena. Throughout the past decade, PT - photon is governed by the standard quantum theory. In symmetric Hamiltonian has been realized in classical this process, depending upon a particular post-selection system like electronic circuit [12], waveguide [13, 14], on the auxiliary system, PT -symmetric evolution hap- microcavity [15, 16] and in photonics, capitalzsing on pens on Alice’s part of the entangled photons half of the fundamental structure of balanced loss and gain, it the time. In another work reported very recently, state creates almost a new paradigm of devices, eg., synthetic distinguishability has been experimentally verified by photonic lattices [17], single-mode laser [18, 19], high- structuring PT -symmetric quantum simulator [33] – in sensitivity sensors [20], wireless power transfer [21] to the symmetry unbroken phase, the distinguishability name a few. quantifier, the trace distance between states oscillates arXiv:2102.13630v1 [quant-ph] 26 Feb 2021 In a quantum domain, qualitatively distinct from while in the broken phase, it monotonically decreases PT classical behaviors has been discovered, such as worm- and hence -symmetric dynamics which is different hole like behavior in quantum brachistochrone prob- from non-Markovian evolution, allows to completely lem [22], single-shot discrimination of non-orthogonal retrieve information in the unbroken phase [34]. It was PT states [23]. Several no-go theorems, valid in standard also argued in the past that exact -symmetric quan- quantum mechanics, shown to be disturbed in PT - tum theory is equivalent to standard quantum theory symmetric quantum theory [24–26]. More prominently, [35, 36] and thus no new physics would be possible [37]. the most weaker condition, no-signaling, which is de- Therefore, to resolve the controversial issues completely sired to be satisfied in any physical theory, has been in the quantum regime, and to probe the advantage of 2

PT -symmetric quantum theory over the standard one, Hamiltonian commutes with parity and time reversal applications in the field of information processing tasks operators (i.e., for a parity operator, P HP = H and are most desirable, which is missing in the literature. for complex conjugation, T HT = H). An example Towards filling the gap, in the present work, we con- of a two-dimensional non-Hermitian PT -symmetric sider PT -symmetric quantum simulation in the open Hamiltonian is given by [27] quantum system framework [32] and show that Bob  i sin α 1  cannot have any input information from Alice’s side, H = s ,(1) if the underlying Hilbert space is real. This discovery 1 −i sin α has two immediate consequences. Firstly, it was known from the beginning era of quantum theory that system where s is the scale factor and α is the non-Hermiticity is associated with Hilbert space over complex field, al- parameter, such that when α = nπ, H is σx Pauli matrix though it was argued whether real Hilbert space is suf- and hence Hermitian. Depending upon the value of α, ficient [38–42]. This line of study has a long history and the distinct three regions exists – (i) 0 < α < π/2 is it is recently proved that complex Hilbert space is in- symmetry unbroken phase, (ii) α = π/2 is exceptional evitable. Therefore, our results along with the recent point where it crosses from unbroken phase to a broken results on complex Hilbert space [42] indicate that an one and (iii) α > π/2 is symmetry broken phase. The isolated subsystem may not evolve according to PT - initial state of the system, ρ(0), is evolved to a final ( ) = exp(−iHt)ρ(0) exp(iHt) symmetric dynamics, in general. The second one is state, ρ t Tr[exp(−iHt)ρ(0) exp(iHt)] . a direct application in devising cryptographic primi- Let us now describe the experimental set-up pro- tives, randomness amplification [43–45], where genuine posed in Ref. [32] for simulating the effective PT - randomness can be generated from an weak source of symmetric Hamiltonian in the open-system framework. randomness generator (RNG). Pseudo RNG have many A pair of space-like separated photons are generated applications in biology, economics, statistics but for by the parametric down conversion which are then sent gambling and cryptographic purposes, genuine RNG to two different protagonist say, Alice and Bob via two is most wanted apart from the foundational application different channels. After Alice encodes a single bit of in Bell test [46]. We show a pseudo-random bit string information on her photon by randomly choosing be- generated from a source provided by a supplier who tween two operations (A±), it is evolved according to may be eavesdropper, can be transformed into a gen- the PT -symmetric Hamiltonian. It is implemented by uinely random bit string of a half-length of the initial evolving the photon together with an auxiliary system in an ideal scenario only when PT -symmetric evolu- via conventional quantum gate operation and subse- tion occurs on the underlying complex Hilbert space. quently, performing a measurement on the auxiliary The novel application may lead to a better understand- system. Depending upon a particular post selection of ing of PT -symmetric quantum system in the context the auxiliary system, Alice’s photon evolved according of very fruitful test bed of information processing tasks to PT -symmetric Hamiltonian. It is to be noted here and devising new quantum technologies. We further that the entire protocol follows the standard quantum show that increasing dimension can lead to some ad- theory and no genuine PT -symmetric Hamiltonian is vantages in the distinguishability process of quantum realised although it can be simulated. On the other states [47–49]. hand, Bob’s photon is evolved according to the identity The paper is organized as follows. In Sec.II, we channel. Finally, Alice and Bob measure locally on their describe the prerequisites that are required to present subsystems simultaneously and record some outcomes. the results. The effects of PT -symmetric is more pro- No-signaling condition. As a consequence of relativis- nounced in the case of complex Hilbert space, shown in tic causality, each spatially separated parties cannot Sec. III while the consequence on Bob’s reduced state predict the measurement choice of others looking at due to local PT -symmetric operation on Alice’s end its own measurement statistics. Mathematically, it is 0 is illustrated in Appendix. Before concluding in Sec. defined as ∑a P (a, b | A, B) = ∑a P (a, b | A , B) ∀b, B, VI, the application of the PT -symmetric dynamics to- where A, A0 and B are measurements performed by Al- wards quantum randomness generation is presented in ice and Bob respectively. Similar condition also holds Sec.IV and the effects of PT -symmetric dynamics of for interchanging Alice and Bob. In an equivalent for- higher dimensional states are studied in Sec.V. mulation, satisfaction of no-signaling principle implies that Bob’s reduced system cannot be affected by space- like separated Alice’s operation and vice-versa [50]. For II. PT -SYMMETRIC DYNAMICS AND ITS two-qubit maximally entangled shared state, by taking QUANTUM SIMULATION A± as I and σx and by permorning measurement by Alice and Bob in σy on their respective subsystems [27], Let us begin with a short discussion on a few prelim- a figure of merit for the violation of no-signaling condi- inary notions relevant for the subsequent sections. tion can be represented as PT -symmetric dynamics. In the PT -symmetric evo- PT P+ − P− = ∑ P (a, b | A+, B) − ∑ P (a, b | A−, B) ,(2) lution, the Hamiltonian satisfies H = H , i.e., the a a 3 where P (a, b | A±, B) is the joint probability of getting again that here we follow the conventional inner prod- outcome a by Alice and b by Bob measuring σy accord- uct as in open system quantum simulation, everything ing to whether Alice applies A+ or A−. In all the previ- follows according to standard quantum theory. After ous work [27, 32], violation of no-signaling was shown Alice and Bob measure in the |±yi basis, we obtain 8βγ sin α for the maximally entangled shared two-qubit state in P+ − P− = . Here we observe that Gedanken experimental set-up. We go beyond such re- (β2+γ2)(−3+cos 2α) = strictions on shared states and local measurements per- the difference vanishes when either α nπ, or when = = formed by both the parties and show that it may help to β 0 or γ 0. It implies that when the state is a 6= address an important question of differentiating quan- product state, even with α nπ, the non-Hermitian tum mechanics in real Hilbert space from that in com- Hamiltonian does not lead to signaling or as we know, plex ones. when the evolution is unitary, Bob cannot get the infor- mation about Alice’s operation even if the shared state is entangled. In case of Werner state [51], following the same prescription as before, the information about III. PT -SYMMETRIC HAMILTONIAN SINGLES OUT Alice’s subsystem f can be protected from Bob when THE REAL HILBERT SPACE 4p sin α P+ − P− = −3+cos 2α vanishes, i.e. even when α 6= nπ but p = 0, implying the shared state is a maximally Going beyond maximally entangled shared states, let mixed state. us consider (i) non-maximally entangled pure state, (ii) Condition for arbitrary two-qubit states. The canonical + + 1−p the Werner state [51], given by ρ = p|ψ ihψ | + 4 I4, form of an arbitrary two-qubit state can be written as with |ψ+i and I being the maximally entangled state 0 1 and white noise respectively and (iii) arbitrary two- ρ(m , m , C ) = (I + [m (σ ⊗ I ) + m0(I ⊗ σ ) i i ii 4 4 ∑ i i 2 i 2 i qubit density matrix. Such a consideration is also im- i=x,y,z portant due to the fact that if it turns out that violation +Cii(σi ⊗ σi)]),( 3) of no-signaling condition at the subsystem level only 0 happens for maximally entangled state, it will never be where mi(mi) = tr(σi ⊗ Iρ) (tr(I ⊗ σiρ)) and Cii = observed since in a practical scenario, one can have only tr(σi ⊗ σiρ) denote the magnetizations and classi- state close to a maximally entangled state and viola- cal correlators respectively. Following the similar 0 tion of no-signaling under PT -symmetric evolution be- procedure, we obtain that when Cyy = mymy or comes pathological. Interestingly, such a general frame- when Cyy = my = 0, the difference P+ − P− = 0 work provides us a necessary condition for the peace- 2(Cyy−mymy)(−3+cos 2α) sin α 2 vanishes even when ful coexistence of the genuine PT -symmetric dynam- (−3+cos 2α+4my sin α)(1+2my sin α+sin α) ics and no-signaling principle, i.e., we find that the al- the PT -symmetric evolution occurs at Alice’s port. It lowed states and observables have to be defined over indicates that the probabilities exclusively depend on 0 real Hilbert space to avoid the information gain of Bob state variables in the y direction such as my, my and Cyy. about Alice’s random operation. This is due to the fact that the measurement is per- Before presenting the main results for arbitrary two- formed in the direction of |±yih±y|. Only the σy com- qubit states with arbitrary measurement, let us illus- ponent is contracting with the measurement setting, trate our findings with shared non-maximally entan- hence the corresponding variables are visible in the gled, Werner, and two-qubit states by performing the probabilities. local projective measurement in the y-direction, i.e. in Removing such biases, i.e., by performing arbitrary |0i±i|1i rank-1 projective measurements, we can arrive to the the basis |±yi = √ as described in Ref. [27]. 2 result that if the shared state has elements from com- Non-maximally entangled and Werner states under PT - plex Hilbert space, Bob can distinguish the information symmetric local evolution. Suppose Alice and Bob ini- of Alice’s random operation. |++i+ |−−i tially share the state, |ψ+i = β √ γ , where |±i β2+γ2 Theorem 1. Random input information about Alice’s oper- are eigenstates of σx. Depending upon her random in- ation can be predicted by Bob with a finite probability when formation, Alice applies either A+ = I or A− = σx Alice’s subsystem is governed by PT -symmetric Hamilto- which is followed by the non-unitary evolution U(t), nian successfully and when the underlying Hilbert space cor- generated by the non-Hermitian Hamiltonian in Eq. responds to the state is in complex Hilbert space.  0 0  −itH 1 cos (t − α) −i sin t (1), U(t) ≡ e = cos α 0 0 . Proof. By considering an arbitrary two-qubit state, −i sin t cos (t + α) . 0 0 ∆E π ρ(m , m , C ), Alice applies I and σ (without loss of Putting t = t, t = E where δE = E+ − E−, we i i ii x 2 δ generality, we can choose such a fixed operation, since  sin α −i  consequently get U(t0) = . After the at the end, Alice performs arbitrary measurements), fol- −i − sin α lowed by an application of local PT -symmetry oper- PT -symmetric evolution, the normalised joint state be- E ation at Alice’s node. Finally, Alice locally performs ± =  ( ) ⊗ −iIt  | i ⊥ ⊥ comes ψf U τ A± e I ψ . It is to be noted measurements in the basis {|φihφ|, |φ ihφ |} with 4

y iv y ⊥ y |φi = cos 2 |0i + e sin( 2 )|1i and |φ i = sin 2 |0i − it can only be discriminated probabilistically and hence −iv y ⊥ ⊥ with a nonvanishing probability, Bob can gain informa- e cos 2 |1i while Bob measures {|ϕihϕ|, |ϕ ihϕ |} z iu z tion about Alice’s randomly chosen operation. with |ϕi = cos 2 |0i + e sin 2 |1i. The difference be- tween the probability of obtaining |ϕi at Bob’s end, The trace distance between two density matrices, ρ reads and σ can quantify the maximum probability of mini- mum error discrimination between the states by the best a a 0 P+ − P− =−2(7 sin α − sin 3α)[mymx cos u sin z quantum measurement strategy [52] and is defined as h i + sin u sin z(−C + m m0 ) + m m0 cos z] 1 p 2 1 yy y y y z T(ρ, σ) = 2 Tr (ρ − σ) = 2 ∑i |λi|, where λis are 2 2 2 p 2 /((−3 + cos 2α) − 16my sin α), the eigenvalues of the operator (ρ − σ) . In this sit- (4) uation, ρ and σ represent the state of Bob when I is applied by Alice, and when σx is applied respectively. where superscript, ”a”, is used to indicate arbitrary For arbitrary two-qubit density matrix, it reduces to measurements. We observe that the numerator vanishes when 7 sin α − sin 3α = 0, which is when α = nπ and q √ C2 + m02m2 − 2C m m0 + m2m02 + m2m02 −1 p yy x y yy y y y y y z also when α = 2πn ± 2i tanh ( 3 ± 2 2), n ∈ Z T = − + − 2
+ + 2
or when all the y components of Alice, i.e., my and Cyy 1 2my sin α sin α 1 2my sin α sin α vanish which gives the proof. ×|(sin α + sin3 α)|,( 5)

Remark. Since the time evolution operator at the given which again shows that if both Cyy and my vanish, the specific time can be written as U = sin(α)σz − iσx, we distance vanishes. It also implies that if the shared state a a find that in P+ − P−, mx, mz, Cxx and Czz are not present. has no imaginary component, the information about Alice’s operations cannot be gathered by Bob, even Corollary 1. For maximally entangled state and for the probabilistically, thereby confirming Theorem 1. Werner state, arbitrary measurements at Bob’s end is re- quired to be complex for gathering information about Alice’s operation by Bob. IV. RANDOMNESS AMPLIFICATION VIA Proof. When |ϕi at Bob’s side clicks, the condition PT -SYMMETRIC EVOLUTION a a for the non-maximally entangled state is P+ − P− = 8βγ sin u sin z sin α (β2+γ2)(−3+cos 2α) while for the Werner state, it can be 4p sin u sin z sin α a a L/2 given by P+ − P− = −3+cos 2α (in Eq. (4), putting [0,1] my =0, and Cyy = −p). Clearly, Alice’s information cannot be obtained by Bob if the measurement at Bob’s σy side is real, i.e., u = 0 or measurement is along the z-direction or α = nπ. For pure states, we also get the condition that when the state is product, signaling cannot occur while for Werner states, the condition gives the state to be maxi- mally mixed states. Interestingly, note that the informa- �/σx tion gain at Bob’s end is not related to entanglement for SOURCE mixed shared states since the Werner state is entangled L WEAK with p > 1/3 (see Appendix and Fig. 3). [0,1] RNG

Distinguishing Bob’s states via Trace distance after PT- λ symmetric evolution at Alice’s node. To exhibit the results further, let us find the distance between the Bob’s state FIG. 1. Schematic diagram of randomness amplification with PT -symmetric evolution. The source produces a two-party corresponding to Alice’s action of I or σx on the shared state. It was shown that for maximally entangled state state shared between Alice (A) and Bob (B). Alice possess a weak random number generator, and accordingly, operates σ |ψ+i, at α = π , Bob’s state is |± ih± | corresponding x 2 y y or nothing randomly. After that, by using the auxiliary sys- to either I or σx operation on Alice’s side [27], which tem followed by the measurement and classical information indicates that Bob can distinguish his own subsystem transmission from A to B, Alice’s subsystem evolves accord- perfectly, thereby distinguishing Alice’s operation with ing to PT -symmetric Hamiltonian. If the initial shared den- unit probability. Here notice that the probability of suc- sity matrix is in complex Hilbert space, Bob can always create cess in distinguishing Alice’s operation also depends on a smaller string of genuine randomness with a finite proba- the probability involved in post-selection. In general, bility which is dictated by the minimum error discrimination Bob’s state may not always be perfectly distinguishable, bound. 5

Let us now describe how the above result enables us interesting further study how to make this protocol se- to devise novel applications of PT -symmetric dynam- cure even when entangled photon source is provided ics in quantum information processing task, specifically, by an eavesdropper. in the quantum cryptographic domain. It is known that Instead of maximally entangled state, if Alice and in the classical regime, only pseudo-random number Bob share an arbitrary two-qubit density matrix, ρ, such can be obtained while in the quantum domain, gen- random number amplification is successful probabilis- uine randomness can be certified by using quantum tically, and Bob can have a smaller length of bit string no-go theorems [53, 54]. In randomness amplification, than half, provided the classical correlator and magneti- a shorter genuine random bit-string is generated from zation in the y-direction, i.e., Cyy, my are nonvanishing. a pseudo-long random string [43]. Towards executing The length of the genuine random bit string is related the protocol in our set-up, we replace the short-delay to the distinguishability of states as measured in the quantum random pulse generator employed in the orig- previous section. inal protocol [32] by an weak random number genera- tor (WRNG) like Santha-Vazirani source [43, 55]. The choice is due to the fact that we want to show ampli- V. PT SYMMETRY IN HIGHER DIMENSION fication of random number generation (RNG), thereby showing perfect quantum random number source and hence initially if parties have quantum random num- 0.7 α = π/2, d = 2 α = π/2, d = 3 ber generator in their possession, the protocol does not α = π/2, d = 4 make sense. Here we also assume that this source pro- 0.6 vided by a supplier may be an eavesdropper. 0.5 Let us first give the protocol (in Fig. 1) for maxi- mally entangled shared state and then we go for an 0.4 arbitrary shared two-qubit states. Before starting the 0.3 protocol, we assume that Alice and Bob’s actions are predecided. Alice encodes one bit of information ob- Trace Distance 0.2 tained from WRNG on her subsystem of the shared 0.1 of maximally entangled photon via applying σx when she gets 0 and do nothing when she has 1. Using the 0 open system approach, she jointly evolves her photon 0 0.2 0.4 0.6 0.8 1 together with the auxiliary system by appropriate uni- p tary followed by postselection on auxiliary subsystem (for detailed see [32]). After a short delay, Bob mea- FIG. 2. Trace distance between the initial states and the state PT sures σy on his part and record his outcome. After that after -symmetric evolution (vertical axis)) against p (hori- zontal axis) of the Werner state, given by ρW = p|ψ+ihψ+ + he receives a phone call from Alice and learns the inci- + dence when she succeeds in post selection to obtain a (1 − p)I where |ψ i is the maximally entangled state for a particular outcome. Bob can keep only those outcomes given dimension while I denotes the identity operator of that α and discard all others. As on average half of the times dimension. We choose different values for demonstration. Both the axis is dimensionless. Alice succeeds about the simulation of PT -symmetric evolution, at the end of the protocol, Bob has a genuine random bit string which has length half of the initial Upto now, we consider the shared states to be string which Alice obtained from some WRNG. Thus two-qubits. Let us now move to a higher di- generated randomness is genuine as we do not allow mensional situation. The PT -symmetric Hamilto- nian, with spin-1 matrices, takes the form as [56] any interference in the lab of two trusted parties, Alice   and Bob, including source of entanglement. i sin α 1 0 H = √1  1 0 1 . Suppose Alice and The above scheme can lead to several directions de- 2 0 1 −i sin α pending upon how we run the protocol. For example, 3 1 one can also think that at the end of protocol, Alice and Bob share a two-qutrit density matrix, ρ = 9 (I9 + 0 Bob share a random string of bits which can not be pre- ∑i=x,y,z miSi ⊗ I + mi I ⊗ Si + CiiSi ⊗ Si), where Si de- 0 dicted by the outsider, thereby manifesting a QKD pro- notes the spin-1 matrices, and mi, mi, Cii are corre- tocol. On the other hand, we can employ this model to spondingly local and global correlators defined in terms certify entanglement of the source instead of key gen- of spin operator, Si. If we now compare the Bob’s state eration protocol. If Alice and Bob verifies their random before and after applying PT -symmetric evolution on bit string to be equal, they can confirm about the source Alice’s side, we find that they coincide when my and producing maximally entangled state and otherwise, Cyy are vanishing even with α 6= nπ which is similar to the source is unable to create. In the ideal scenario, the qubit case as shown in Appendix. However, unlike this protocol succeeds in the asymptotic limit of non- the qubit case, we also require mz and Czz to be zero in Hermiticity parameter to the exceptional point. It is an order to avoid signaling. 6

Let us now analyze the effects of dimensionality on APPENDIX: EFFECTS OF ALICE’S OPERATIONS ON distinguishability i.e., with the increase of dimension, BOB’S REDUCED STATE we compute the trace distance between states before and after PT -symmetric evolution, for different α val- ues when the shared state is the Werner-like state in the 1 α = 0.2 α = 0.5 respective dimension. We observe that the difference in- α = π/2 α = 2.5 creases with the increase of the dimension as depicted 0.8 α = 3.1 in Fig. 2, when α is chosen to be close to π/2. 0.6

0.4

VI. CONCLUSION Trace Distance 0.2 In a quantum world, evolution of the system accord- ing to parity-time aka PT symmetry theory reveals sev- 0 eral intriguing features ranging from minimum evolu- 0 0.2 0.4 0.6 0.8 1 tion time in quantum brachistochrone problem, sponta- P neous oscillations in the distinguishability of quantum states to the no-signaling condition. It was also shown FIG. 3. Trace distance between Bob’s state before and af- that if two observers share a quantum state, an ob- ter PT -symmetric dynamics (ordinate) vs. p (abscissa) of the server can predict the operations occurred at the other two-qubit Werner state. Different α of PT -symmetric evolu- end evolved by the PT -symmetric Hamiltonian – such tion are chosen. Both the axes are dimensionless. a consequence can never occur when one adopts the proper inner product or simulates such systems in open There is a straight forward way to check whether the quantum system framework. Bob can draw information about Alice’s operation, i.e., We proved that for arbitrary two-qubit shared states, signaling happens or not by comparing Bob’s state be- information about randomly chosen operations per- fore and after local PT -symmetric Hamiltonian as also formed by a subsystem on which PT -symmetric evo- shown in Sec.V. lution takes place can be gathered by another subsys- Case 1. Non-maximally entangled states. When the tem provided the density matrix is defined in complex shared state is non-maximally entangled, Bob’s state Hilbert space. For example, considering the maximally (ρB) is initially entangled state and Werner state, we demonstrated that  2  1 − 2β information about operations can be obtained when ei- 2 1 β2+γ2 ther the state is a product or the measurement has a ρB =  2  (6) 1 − 2β 1 complex component. We then showed that such a con- β2+γ2 2 sideration has a direct connection in designing quan- PT tum random number generator from a pseudo random while after the application of local -symmetric bit-string. In particular, we showed that systems un- Hamiltonian on Alice’s subsystem, Bob’s state becomes der PT -symmetric evolution can indeed amplify the  1 Unm  shared randomness which has direct implications in ρ0 = 2 7 B ¯nm 1 ,( ) quantum state certification as well as quantum cryp- U 2 tographic protocols. Finally, we indicated that such a ( 2− 2)(− + ( ))+ ( ) Unm = β γ 3 cos 2α 8iβγ sin α condition for obtaining information about the operation where 4(β2+γ2)(1+sin(α)2) The condi- performed by a spatially separated party holds even in 0 tion that ρB = ρB leads to the condition γ = 0 even higher-dimensional systems. when α 6= nπ, which is similar to the condition reached in Sec. III. Case 2. Mixed states. For Werner state, Bob’s state after PT -symmetry dynamics reads ACKNOWLEDGEMENT  1 ì p Sin (α)  2 2 1+sin(α) .(8)  2 i p sin(α) 1  We acknowledge the support from the Interdisci- −3+cos(2α) 2 plinary Cyber Physical Systems (ICPS) program of the Department of Science and Technology (DST), India, from its initially maximally mixed subsystem. We plot Grant No.: DST/ICPS/QuST/Theme- 1/2019/23 and the trace distance between these two states and find that SM acknowledges Ministry of Science and Technology it vanishes only when p = 0 and the maximum distance in Taiwan (Grant no. 110-2811-M-006 -501). is achieved when α = nπ/2 (see Fig. 3). 7

 2 0  Interestingly, we find the similar condition obtained 1 (1±sin(α) )mz with R± = 1 + 2 and U = in Eq. (4) if we compare Bob’s state before the dynam- 2 1+2my sin(α)+sin(α) 2 0 2 0 ics, (1+sin(α) )mx −i(2Cyy sin α+(1+sin(α) )my ) 2 . Comparing 2(1+2my sin α+sin α) 0 0 im0 ! 1 mz mx y each element of the matrix, we find that R± reduces 2 + 2 2 − 2 0 ρB = 0 im0 0 (9) 1 mz mx y 1 mz to 2 ± 2 when my vanishes while if both Cyy and my 2 + 2 2 − 2 are vanishing, U reduces to the off-diagonal term of ρB, and after Alice applies local PT -symmetric operation, thereby arriving to the same condition as obtained in Theorem 1.  R U  ρ0 = + (10) B UR¯ −

[1] C. M. Bender and S. Boettcher, Phys. Rev. Lett. 80, 5243 [24] S.-L. Chen, G.-Y. Chen, and Y.-N. Chen, Phys. Rev. A 90, (1998). 054301 (2014). [2] C. M. Bender, D. C. Brody, and H. F. Jones, Phys. Rev. [25] A. K. Pati, arXiv:1404.6166 (2014). Lett. 89, 270401 (2002). [26] M. Znojil, Acta Polytechnica (Prague) 56, 254 (2016). [3] C. M. Bender, S. Boettcher, and P. N. Meisinger, J. Math. [27] Y.-C. Lee, M.-H. Hsieh, S. T. Flammia, and R.-K. Lee, Phys. 40, 2201 (1999). Phys. Rev. Lett. 112, 130404 (2014). [4] P. E. Dorey, C. Dunning, and R. Tateo, J. Phys. A: Math. [28] D. C. Brody, J. Phys. A: Math. Theor. 49, 10LT03 (2016). Gen. 34, 5679 (2001). [29] G. Japaridze, D. Pokhrel, X-Q Wang, J. Phys. A: Math. [5] P. E. Dorey, C. Dunning, and R. Tateo, J. Phys. A: Math. Theor. 50, 18503 (2017). Theor. 40,R205 (2007). [30] Chia-Yi Ju, Adam Miranowicz, Guang-Yin Chen, and [6] C. M. Bender and T. T. Wu, Phys. Rev. Lett. 21, 406 (1968). Franco Nori, Phys. Rev. A 100, 062118 (2019). [7] C. M. Bender and T. T. Wu, Phys. Rev. 184, 1231 (1969). [31] U. Gunther, and B. F. Samsonov, Phys. Rev. Lett. 101, [8] W. D. Heiss, J. Phys. A: Math. Theor. 45, 444016 (2012). 230404 (2008). [9] C. M. Bender, N. Hassanpour, D. W. Hook, S. P. Klevan- [32] J.-S. Tang, et al., Nat. Photonics 10, 642 (2016). sky, C. Sünderhauf, and Z. Wen, Phys. Rev. A 95, 052113 [33] Y.-T. Wang et al., Phys. Rev. Lett. 124, 230402 (2020). (2017). [34] K. Kawabata, Y. Ashida, M. Ueda, Phys. Rev. Lett. 119 [10] Y. Ashida, S. Furukawa and M. Ueda, Nat. Comm. 8, 190401 (2017). 15791 (2017). [35] A. Mostafazadeh, J. Math. Phys. 43, 205 (2001). [11] C. M. Bender, Rep. Prog. Phys. 70, 947 (2007). [36] A. Mostafazadeh, Phys. Rev. Lett. 99, 130502 (2007). [12] N. Bender, S. Factor, J. D. Bodyfelt, H. Ramezani, D. N. [37] S. Croke, Phys. Rev. A 91, 052113 (2015). Christodoulides, F. M. Ellis, and T. Kottos, Phys. Rev. [38] E. C. G. Stueckelberg, Quantum Theory in Real Hilbert Lett. 110, 234101 (2013). Space. Helvetica Physica Acta 33, 727 (1960). [13] A. Guo, G. J. Salamo, D. Duchesne, R. Morandotti, M. [39] M. McKague, M. Mosca, and N. Gisin, Phys. Rev. Lett. Volatier-Ravat, V. Aimez, G. A. Siviloglou, and D. N. 102, 020505 (2009). Christodoulides, Phys. Rev. Lett. 103, 093902 (2009). [40] L. Hardy and W. K. Wootters, Found. Phys. 42, 454 [14] J. Doppler, A. A. Mailybaev, J. Böhm, U. Kuhl, A. (2012). Girschik, F. Libisch, T. J. Milburn, P. Rabl, N. Moiseyev, [41] A. Aleksandrova, V. Borish, and W. K. Wootters, Phys. and S. Rotter, Nature (London) 537, 76 (2016). Rev. A 87, 052106 (2013). [15] B. Peng, ¸S.K. Ozdemir, F. Lei, F. Monifi, M. Gianfreda, [42] M.-O. Renou, D. Trillo, M. Weilenmann, Le P. Thinh, G. Lu Long, S. Fan, F. Nori, C. M. Bender, and L. Yang, A. Tavakoli, N. Gisin, A. Acin, and M. Navascues, Nat. Phys. 10, 394 (2014). arXiv:2101.10873. [16] L. Chang, X. Jiang, S. Hua, C. Yang, J. Wen, L. Jiang, G. [43] R. Colbeck and R. Renner, Nat. Phys. 8, 450 (2012). Li, G. Wang, and M. Xiao, Nat. Photonics 8, 524 (2014). [44] M.-H. Li, et. al., Phys. Rev. Lett. 126, 050503 (2021). [17] A. Regensburger, et. al., Nature 488, 167–171 (2012). [45] R. Ramanathan, et. al., arXiv:1810.11648. [18] L. Feng, Z. J. Wong, R.-M. Ma, Y. Wang, and X. Zhang, [46] J. S. Bell, Speakable and Unspeakable in Quantum Mechanics, Science 346, 972 (2014). (Cambridge University Press, New York, 1987). [19] P. Miao, Z. Zhang, J. Sun, W. Walasik, S. Longhi, N. M. [47] D. Kaszlikowski, P. Gnacinski, M. Zukowski, W. Mik- Litchinitser, and L. Feng, Science 353, 464 (2016). laszewski, and A. Zeilinger, Phys. Rev. Lett. 85, 4418 [20] W. Chen, S. K. Ozdemir, G. Zhao, J. Wiersig, and L. Yang, (2000). Nature (London) 548, 192 (2017). [48] D. Collins, N. Gisin, N. Linden, S. Massar, and S. [21] S. Assawaworrarit, X. Yu, and S. Fan, Nature (London) Popescu, Phys. Rev. Lett. 88, 040404 (2002). 546, 387 (2017). [49] S. Roy, A. Kumari, S. Mal, and A. Sen (De), arXiv: [22] C. M. Bender, D. C. Brody, H. F. Jones, and B. K. Meister, 2012.12200. Phys. Rev. Lett. 98, 040403 (2007). [50] A. Kumari, and U. Sen, arXiv:2007.13461. [23] C. M. Bender, D. C. Brody, J. Caldeira, U. Gunther, B. K. [51] R.F. Werner, Phys. Rev. A 40, 4277 (1989). Meister, and B. F. Samsonov, Phil. Trans. R. Soc. A 01, 371 [52] C. W. Helstrom, Quantum Detection and Estimation Theory (2013). (Academic, New York, 1976). 8

[53] S. Pironio, et. al., Nature 464, 1021 (2010). tem Sciences 33, 75 (1986). [54] S. Mal, M. Banik and S. K. Choudhary, Quant. Inf. Proc. [56] Z. Bian, et. al., Phys. Rev. A 102, 030201(R) (2020) and 15, 2993 (2016). references therein. [55] M. Santha and U. Vazirani, Journal of Computer and Sys-