Entangled Harmonic Oscillators and Space-time Entanglement

The mathematical basis for the Gaussian entanglement is discussed in detail, as well as its implications in the internal space-time structure of relativistic extended particles. It is shown that the Gaussian entanglement shares the same set of mathematical formulas with the harmonic oscillator in the Lorentz-covariant world. It is thus possible to transfer the concept of entanglement to the Lorentz-covariant picture of the bound state which requires both space and time separations between two constituent particles. These space and time variables become entangled as the bound state moves with a relativistic speed. It is shown also that our inability to measure the time-separation variable leads to an entanglement entropy together with a rise in the temperature of the bound state. As was noted by Paul A. M. Dirac in 1963, the system of two oscillators contains the symmetries of O(3,2) de Sitter group containing two O(3,1) Lorentz groups as its subgroups. Dirac noted also that the system contains the symmetry of the Sp(4) group which serves as the basic language for two-mode squeezed states. Since the Sp(4) symmetry contains both rotations and squeezes, one interesting case is the combination of rotation and squeeze resulting in a shear. While the current literature is mostly on the entanglement based on squeeze along the normal coordinates, the shear transformation is an interesting future possibility. The mathematical issues on this problem are clarified.


Introduction
Entanglement problems deal with fundamental issues in physics. Among them, the Gaussian entanglement is of current interest not only in quantum optics [1,2,3,4] but also in other dynamical systems [3,5,6,7,8]. The underlying mathematical language for this form of entanglement is that of harmonic oscillators. In this paper, we present first the mathematical tools which are and may be useful in this branch of physics.
The entangled Gaussian state is based on the formula where χ n (x) is the n th excited-state oscillator wave function.
In Chapter 16 of their book [9], Walls and Milburn discussed in detail the role of this formula in the theory of quantum information. Earlier, this formula played the pivotal role for Yuen to formulate his two-photon coherent states or two-mode squeezed states [10]. The same formula was used by Yurke and Patasek in 1987 [11] and by Ekert and Knight [12] for the two-mode squeezed state where one of the photons is not observed. The effect of entanglement is to be seen from the beam splitter experiments [13,14].
In this paper, we point out first that the series of Eq.(1) can also be written as a squeezed Gaussian form which becomes 1 √ π exp − 1 2 when η = 0. We can obtain the squeezed form of Eq.(2) by replacing x and y by x ′ and y ′ respectively, where If x and y are replaced by z and t, Eq.(4) becomes the formula for the Lorentz boost along the z direction. Indeed, the Lorentz boost is a squeeze transformation [3,15]. The squeezed Gaussian form of Eq.(2) plays the key role in studying boosted bound states in the Lorentz-covariant world [16,17,18,19,20], where z and t are the space and time separations between two constituent particles. Since the mathematics of this physical system is the same as the series given in Eq.(1), the physical concept of entanglement can be transferred to the Lorentz-covariant bound state, as illustrated in Fig. 1.
We can approach this problem from the system of two harmonic oscillators. In 1963, Paul A. M. Dirac studied the symmetry of this two-oscillator system and discussed all possible transformations applicable to this oscillator [21]. He concluded that there are ten possible generators of transformations satisfying a closed set of commutation (2) applicable to quantum optics and special relativity respectively. They are the same formula from the Lorentz group with different variables as in the case of the LCR circuit and the mechanical oscillator sharing the same second-order differential equation.
relations. He then noted that this closed set corresponds to the Lie algebra of the O(3, 2) de Sitter group which is the Lorentz group applicable to three space-like and two time-like dimensions. This O(3, 2) group has two O(3, 1) Lorentz groups as its subgroups.
We note that the Lorentz group is the language of special relativity, while the harmonic oscillator is one of the major tools for interpreting bound states. Therefore, Dirac's two-oscillator system can serve as a mathematical framework for understanding quantum bound systems in the Lorentz-covariant world.
Within this formalism, the series given in Eq.(1) can be produced from the tengenerator Dirac system. In discussing the oscillator system, the standard procedure is to use the normal coordinates defined as In terms of these variables, the transformation given in Eq.(4) takes the form where this is a squeeze transformation along the normal coordinates. While the normalcoordinate transformation is a standard procedure, it is interesting to note that it also serves as a Lorentz boost [18]. With these preparations, we shall study in Sec. 2, the system of two oscillators and coordinate transformations of current interest. It is pointed out in Sec. 3 that there are ten different generators for transformations, including those discussed in Sec. 2. It is noted that Dirac derived ten generators of transformations applicable to these oscillators, and they satisfy the closed set of commutation relations which is the same as the Lie algebra of the O(3, 2) de Sitter group containing two Lorentz groups among its subgroups. In Sec. 4, Dirac's ten-generator symmetry is studied in the Wigner phasespace picture, and it is shown that Dirac's symmetry contains both canonical and Lorentz transformations.
While the Gaussian entanglement starts from the oscillator wave function in its ground state, we study in Sec. 5 the entanglements of excited oscillator states. We give a detailed explanation of how the series of Eq.(1) can be derived from the squeezed Gaussian function of Eq.(2).
In Sec. 6, we study in detail how the sheared state can be derived from a squeezed state. It appears to be a rotated squeezed state, but this is not the case. In Sec. 7, we study what happens when one of the two entangled variables is not observed within the framework of Feynman's rest of the universe [22,23].
In Sec. 8, we note that most of the mathematical formulas in this paper have been used earlier for understanding relativistic extended particles in the Lorentz-covariant harmonic oscillator formalism [24,20,25,26,27,28]. These formulas allow us to transport the concept of entanglement from the current problem of physics to quantum bound states in the Lorentz-covariant world. The time separation between the constituent particles is not observable, and is not known in the present form of quantum mechanics. However, this variable gives its effect in the real world by entangling itself with the longitudinal variable.

Two-dimensional Harmonic Oscillators
The Gaussian form is used for many branches of science. For instance, we can construct this function by throwing dice. In physics, this is the wave function for the one-dimensional harmonic oscillator in the ground state. This function is also used for the vacuum state in quantum field theory, as well as the zero-photon state in quantum optics. For excited oscillator states, the wave function takes the form where H n (x) is the Hermite polynomial of the n th degree. The properties of this wave function are well known, and it becomes the Gaussian form of Eq.(7) when n = 0.
We can now consider the two-dimensional space with the orthogonal coordinate variables x and y, and the same wave function with the y variable: and construct the function This form is clearly separable in the x and y variables. If n and m are zero, the wave function becomes Under the coordinate rotation this function remains separable. This rotation is illustrated in Fig. 2. This is a transformation very familiar to us. We can next consider the scale transformation of the form This scale transformation is also illustrated in Fig. 2. This area-preserving transformation is known as the squeeze. Under this transformation, the Gaussian function is still separable.
If the direction of the squeeze is rotated by 45 o , the transformation becomes the diagonal transformation of Eq. (6). Indeed, this is a squeeze in the normal coordinate system. This form of squeeze is most commonly used for squeezed states of light as well as the subject of entanglements. It is important to note that, in terms of the x and y, variables, this transformation can be written as Eq.(4) [18]. In 1905, Einstein used this form of squeeze transformation for the longitudinal and time-like variables. This is known as the Lorentz boost.
In addition, we can consider the transformation of the form This transformation shears the system as is shown in Fig. 2.
After the squeeze or shear transformation, the wave function of Eq.(10) becomes non-separable, but it can still be written as a series expansion in terms of the oscillator wave functions. It can take the form

Squeezed Gaussian Function
Under the squeeze along the normal coordinate, the Gaussian form of Eq.(11) becomes which was given in Eq. (2). This function is not separable in the x and y variables. These variables are now entangled. We obtain this form by replacing, in the Gaussian function of Eq. (11), the x and y variables by x ′ and y ′ respectively, where x ′ = (cosh η)x − (sinh η)y, and y ′ = (cosh η)y − (sinh η)x.
This form of squeeze is illustrated in Fig. 3, and the expansion of this squeezed Gaussian function becomes the series given in Eq.(1) [20,26]. This aspect will be discussed in detail in Sec. 5.
In 1976 [10], Yuen discussed two-photon coherent states, often called squeezed states of light. This series expansion served as the starting point for two-mode squeezed states. More recently, in 2003, Giedke et al. [1] used this formula to formulate the concept of the Gaussian entanglement.
There is another way to derive the series. For the harmonic oscillator wave functions, there are step-down and step-up operators [17]. These are defined as If they are applied to the oscillator wave function, we have a χ n (x) = √ n χ n−1 (x), and a † χ n (x) = √ n + 1 χ n+1 (x). Thus and a χ 0 (x) = b χ 0 (y) = 0.
In terms of these variables, the transformation leading the Gaussian function of Eq. (11) to its squeezed form of Eq.(16) can be written as which can also be written as Next, we can consider the exponential form which can be expanded as If this operator is applied to the ground state of Eq.(11), the result is n (tanh η) n χ n (x)χ n (y).
This form is not normalized, while the series of Eq.(1) is. What is the origin of this difference? There is a similar problem with the one-photon coherent state [29,30]. There, the series comes from the expansion of the exponential form which can be expanded to However, this operator is not unitary. In order to make this series unitary, we consider the exponential form which is unitary. This expression can then be written as according to the Baker-Campbell-Hausdorff (BCH) relation [31,32]. If this is applied to the ground state, the last bracket can be dropped, and the result is which is the unitary operator with the normalization constant Likewise, we can conclude that the series of Eq. (27) is different from that of Eq.(1) due to the difference between the unitary operator of Eq.(23) and the non-unitary operator of Eq. (25). It may be possible to derive the normalization factor using the BCH formula, but it seems to be intractable at this time. The best way to resolve this problem is to present the exact calculation of the unitary operator leading to the normalized series of Eq. (11). We shall return to this problem in Sec. 5, where squeezed excited states are studied.

Sheared Gaussian Function
In addition, there is a transformation called "shear," where only one of the two coordinates is translated as shown in Fig. 2. This transformation takes the form which leads to This shear is one of the basic transformations in engineering sciences. In physics, this transformation plays the key role in understanding the internal space-time symmetry of massless particles [33,34,35]. This matrix plays the pivotal role during the transition from the oscillator mode to the damping mode in classical damped harmonic oscillators [36,37]. Under this transformation, the Gaussian form becomes It is possible to expand this into a series of the form of Eq.(15) [38].
The transformation applicable to the Gaussian form of Eq. (11) is and the generator is It is of interest to see where this generator stands among the ten generators of Dirac. However, the most pressing problem is whether the sheared Gaussian form can be regarded as a rotated squeezed state. The basic mathematical issue is that the shear matrix of Eq.(33) is triangular and cannot be diagonalized. Therefore, it cannot be a squeezed state. Yet, the Gaussian form of Eq.(35) appears to be a rotated squeezed state, while not along the normal coordinates. We shall look at this problem in detail in Sec. 6.

Dirac's Entangled Oscillators
Paul A. M. Dirac devoted much of his life-long efforts to the task of making quantum mechanics compatible with special relativity. Harmonic oscillators serve as an instrument for illustrating quantum mechanics, while special relativity is the physics of the Lorentz group. Thus, Dirac attempted to construct a representation of the Lorentz group using harmonic oscillator wave functions [17,21].
In his 1963 paper [21], Dirac started from the two-dimensional oscillator whose wave function takes the Gaussian form given in Eq. (11). He then considered unitary transformations applicable to this ground-state wave function. He noted that they can be generated by the following ten Hermitian operators He then noted that these operators satisfy the following set of commutation relations.
Dirac then determined that these commutation relations constitute the Lie algebra for the O(3, 2) de Sitter group with ten generators. This de Sitter group is the Lorentz group applicable to three space coordinates and two time coordinates. Let us use the notation (x, y, z, t, s), with (x, y, z) as space coordinates and (t, s) as two time coordinates. Then the rotation around the z axis is generated by The generators L 1 and L 2 can be also be constructed. The K 3 and Q 3 generators will take the form From these two matrices, the generators K 1 , K 2 , Q 1 , Q 2 can be constructed. The generator S 3 can be written as The last five-by-five matrix generates rotations in the two-dimensional space of (t, s). If we introduce these two time variables, the O(3, 2) group leads to two coupled Lorentz groups. The particle mass is invariant under Lorentz transformations. Thus, one Lorentz group cannot change the particle mass. However, with two coupled Lorentz groups we can describe the world with variable masses such as the neutrino oscillations. In Sec. 2, we used the operators Q 3 and K 3 as the generators for the squeezed Gaussian function. For the unitary transformation of Eq.(23), we used However, the exponential form of Eq. (25) can be written as which is not unitary, as was seen before. From the space-time point of view, both K 3 and Q 3 generate Lorentz boosts along the z direction, with the time variables t and s respectively. The fact that the squeeze and Lorentz transformations share the same mathematical formula is well known. However, the non-unitary operator iK 3 does not seem to have a space-time interpretation.
As for the sheared state, the generator can be written as leading to the expression given in Eq. (37). This is a Hermitian operator leading to the unitary transformation of Eq.(36).

Entangled Oscillators in the Phase-Space Picture
Also in his 1963 paper, Dirac states that the Lie algebra of Eq. (39) can serve as the four-dimensional symplectic group Sp(4). This group allows us to study squeezed or entangled states in terms of the four-dimensional phase space consisting of two position and two momentum variables [15,40,41]. In order to study the Sp(4) contents of the coupled oscillator system, let us introduce the Wigner function defined as [42] W (x, y; p, q) If the wave function ψ(x, y) is the Gaussian form of Eq.(11), the Wigner function becomes The Wigner function is defined over the four-dimensional phase space of (x, p, y, q) just as in the case of classical mechanics. The unitary transformations generated by the operators of Eq. (38) are translated into Wigner transformations [40,39,41]. As in Figure 4: Transformations generated by Q 3 and K 3 . As the parameter η becomes larger, both the space and momentum distribution becomes larger.
the case of Dirac's oscillators, there are ten corresponding generators applicable to the Wigner function. They are and These generators also satisfy the Lie algebra given in Eq. (38) and Eq. (39). Transformations generated by these generators have been discussed in the literature [15,39,41]. As in the case of Sec. 3, we are interested in the generators Q 3 and K 3 . The transformation generated by Q 3 takes the form This exponential form squeezes the Wigner function of Eq. (47) in the x y space as well as in their corresponding momentum space. However, in the momentum space, the squeeze is in the opposite direction as illustrated in Fig. 4. This is what we expect from canonical transformation in classical mechanics. Indeed, this corresponds to the unitary transformation which played the major role in Sec. 2. Even though shown insignificant in Sec. 2, K 3 had a definite physical interpretation in Sec. 3. The transformation generated by K 3 takes the form This performs the squeeze in the x q and y p spaces. In this case, the squeezes have the same sign, and the rate of increase is the same in all directions. We can thus have the same picture of squeeze for both x y and p q spaces as illustrated in Fig. 4. This parallel transformation corresponds to the Lorentz squeeze [20,25]. As for the sheared state, the combination generates the same shear in the p q space.

Entangled Excited States
In Sec. 2, we discussed the entangled ground state, and noted that the entangled state of Eq.(1) is a series expansion of the squeezed Gaussian function. In this section, we are interested in what happens when we squeeze an excited oscillator state starting from In order to entangle this state, we should replace x and y respectively by x ′ and y ′ given in Eq. (17).
The question is how the oscillator wave function is squeezed after this operation. Let us note first that the wave function of Eq.(53) satisfies the equation This equation is invariant under the squeeze transformation of Eq. (17), and thus the eigenvalue (n − m) remains invariant. Unlike the usual two-oscillator system, the x component and the y component have opposite signs. This is the reason why the overall equation is squeeze-invariant [25,3,43].
We then have to write this squeezed oscillator in the series form of Eq. (15). The most interesting case is of course for m = n = 0, which leads to the Gaussian entangled state given in Eq. (16). Another interesting case is for m = 0 while n is allowed to take all integer values. This single-excitation system has applications in the covariant oscillator formalism where no time-like excitations are allowed. The Gaussian entangled state is a special case of this single-excited oscillator system.
The most general case is for nonzero integers for both n and m. The calculation for this case is available in the literature [20,45]. Seeing no immediate physical applications of this case, we shall not reproduce this calculation in this section.
For the single-excitation system, we write the starting wave function as There are no excitations along the y coordinate. In order to squeeze this function, our plan is to replace x and y by x ′ and y ′ respectively and write χ n (x ′ )χ 0 (y ′ ) as a series in the form Since k ′ − k = n or k ′ = n + k, according to the eigenvalue of the differential equation given in Eq. (54), we write this series as This coefficient is This calculation was given in the literature in a fragmentary way in connection with a Lorentz-covariant description of extended particles starting from Ruiz's 1974 paper [44], subsequently by Kim et al. in 1979 [26] and by Rotbart in 1981 [45]. In view of the recent developments of physics, it seems necessary to give one coherent calculation of the coefficient of Eq.(59).
We are now interested in the squeezed oscillator function As was noted by Ruiz [44], the key to the evaluation of this integral is to introduce the generating function for the Hermite polynomials [46,47]: and evaluate the integral The integrand becomes one exponential function, and its exponent is quadratic in x and y. This quadratic form can be diagonalized, and the integral can be evaluated [20,26]. The result is We can now expand this expression and choose the coefficients of r n+k , s k , r ′n for H (n+k) (x), H n (y), and H n (z ′ ) respectively. The result is Thus, the series becomes If n = 0, it is the squeezed ground state, and this expression becomes the entangled state of Eq.(16).

E(2)-sheared States
Let us next consider the effect of shear on the Gaussian form. From Fig. 5 and Fig. 3, it is clear that the sheared state is a rotated squeezed state.
In order to understand this transformation, let us note that the squeeze and rotation are generated by the two-by-two matrices respectively. We can then consider This matrix has the property that S 2 = 0. Thus the transformation matrix becomes exp (−iαS) = 1 2α 0 1 .
Since S 2 = 0, the Taylor expansion truncates, and the transformation matrix becomes the triangular matrix of Eq.(34), leading to the transformation The shear generator S of Eq.(68) indicates that the infinitesimal transformation is a rotation followed by a squeeze. Since both rotation and squeeze are area-preserving transformations, the shear should also be an area-preserving transformations.
In view of Fig. 5, we should ask whether the triangular matrix of Eq.(69) can be obtained from one squeeze matrix followed by one rotation matrix. This is not possible mathematically. It can however, be written as a squeezed rotation matrix of the form resulting in cos ω e λ sin ω −e −λ sin ω cos ω .
If we let (sin ω) = 2αe −λ (73) If λ becomes infinite, the angle ω becomes zero, and this matrix becomes the triangular matrix of Eq.(69). This is a singular process where the parameter λ goes to infinity. If this transformation is applied to the Gaussian form of Eq.(11), it becomes The question is whether the exponential portion of this expression can be written as The answer is Yes. This is possible if tan(2θ) = 1 α , e 2η = 1 + 2α 2 + 2α √ α 2 + 1, In Eq.(74), we needed a limiting case of λ becoming infinite. This is necessarily a singular transformation. On the other hand, the derivation of the Gaussian form of Eq.(75) appears to be analytic. How is it possible? In order to achieve the transformation from the Gaussian form of Eq.(11) to Eq.(75), we need the linear transformation If the initial form is invariant under rotations as in the case of the Gaussian function of Eq.(11), we can add another rotation matrix on the right hand side. We choose that rotation matrix to be cos(θ − π/2) − sin(θ − π/2) sin(θ − π/2) cos(θ − π/2) , write the three matrices as with The multiplication of these three matrices leads to (cosh η) sin(2θ) sinh η + (cosh η) cos(2θ) sinh η − (cosh η) cos(2θ) (cosh η) sin(2θ) . (81) The lower-left element can become zero when sinh η = cosh(η) cos(2θ) and consequently this matrix becomes 1 2 sinh η 0 1 .

Feynman's Rest of the Universe
We need the concept of entanglement in quantum systems of two variables. The issue is how the measurement of one variable affects the other variable. The simplest case is what happens to the first variable while no measurements are taken on the second variable. This problem has a long history since von Neumann introduced the concept of density matrix in 1932 [51]. While there are many books and review articles on this subject, Feynman stated this problem in his own colorful way. In his book on statistical mechanics [22], Feynman makes the following statement about the density matrix. When we solve a quantum-mechanical problem, what we really do is divide the universe into two parts -the system in which we are interested and the rest of the universe. We then usually act as if the system in which we are interested comprised the entire universe. To motivate the use of density matrices, let us see what happens when we include the part of the universe outside the system.
Indeed, Yurke and Potasek [11] and also Ekert and Knight [12] studied this problem in the two-mode squeezed state using the entanglement formula given in Eq. (16). Later in 1999, Han et al. studied this problem with two coupled oscillators where one oscillator is observed while the other is not and thus is in the rest of the universe as defined by Feynman [23].
Somewhat earlier in 1990 [27], Kim and Wigner observed that there is a time separation wherever there is a space separation in the Lorentz-covariant world. The Bohr radius is a space separation. If the system is Lorentz-boosted, the time-separation becomes entangled with the space separation. But, in the present form of quantum mechanics, this time-separation variable is not measured and not understood.
This variable was mentioned in the paper of Feynman et al. in 1971 [43], but the authors say they would drop this variable because they do not know what to do with it. While what Feynman et al. did was not quite respectable from the scientific point of view, they made a contribution by pointing out the existence of the problem. In 1990, Kim and Wigner [27] noted that the time-separation variable belongs to Feynman's rest of the universe and studied its consequences in the observable world.
In this section, we first reproduce the work of Kim and Wigner using the x and y variables and then study its consequences. Let us introduce the notation ψ n η (x, y) for the squeezed oscillator wave function given in Eq.(65): with no excitations along the y direction. For η = 0, this expression becomes χ n (x)χ 0 (y). From this wave function, we can construct the pure-state density matrix as ρ n η (x, y; r, s) = ψ n η (x, y)ψ n η (r, s), which satisfies the condition ρ 2 = ρ, which means ρ n η (x, y; r, s) = ρ n η (x, y; u, v)ρ n η (u, v; r, s)dudv.
As illustrated in Fig. 6, it is not possible make measurements on the variable y. We thus have to take the trace of this density matrix along the y axis, resulting in ρ n η (x, r) = ψ n η (x, y)ψ n η (r, y)dy The trace of this density matrix is one, but the trace of ρ 2 is T r ρ 2 = ρ n η (x, r)ρ n η (r, x)drdx which is less than one. This is due to the fact that we are not observing the y variable. Our knowledge is less than complete.
The standard way to measure this incompleteness is to calculate the entropy defined as [51,52,53] which leads to Let us go back to the wave function given in Eq.(84). As is illustrated in Fig. 6, its localization property is dictated by its Gaussian factor which corresponds to the groundstate wave function. For this reason, we expect that much of the behavior of the density matrix or the entropy for the n th excited state will be the same as that for the ground state with n = 0. For this state, the density matrix is and the entropy is The density distribution ρ η (x, x) becomes The width of the distribution becomes cosh(2η), and the distribution becomes widespread as η becomes larger. Likewise, the momentum distribution becomes wide-spread as can be seen in Fig. 4. This simultaneous increase in the momentum and position distribution widths is due to our inability to measure the y variable hidden in Feynman's rest of the universe [22].
In their paper of 1990 [27], Kim and Wigner used the x and y variables as the longitudinal and time-like variables respectively in the Lorentz-covariant world. In the quantum world, it is a widely accepted view that there are no time-like excitations. Thus, it is fully justified to restrict the y component to its ground state as we did in Sec. 5.

Space-time Entanglement
The series given in Eq.(1) plays the central role in the concept of the Gaussian or continuous-variable entanglement, where the measurement on one variable affects the quantum mechanics of the other variable. If one of the variables is not observed, it belongs to Feynman's rest of the universe.
The series of the form of Eq.(1) was developed earlier for studying harmonic oscillators in moving frames [24,20,25,26,27,28]. Here z and t are the space-like and time-like separations between the two constituent particles bound together by a harmonic oscillator potential. There are excitations along the longitudinal direction. However, no excitations are allowed along the time-like direction. Dirac described this as "c-number" time-energy uncertainty relation [16]. Dirac in 1927 was talking about the system without special relativity. In 1945 [17], Dirac attempted to construct space-time wave functions using harmonic oscillators. In 1949 [18], Dirac introduced his light-cone coordinate system for Lorentz boosts, telling that the boost is a squeeze transformation. It is now possible to combine Dirac's three observations to construct the Lorentz covariant picture of quantum bound states, as illustrated in Fig. 7.
If the system is at rest, we use the wave function which allows excitations along the z axis, but no exciations along the t axis, according to Dirac's c-number time-energy uncertainty relation.
If the system is boosted, the z and t variables are replaced by z ′ and t ′ where z ′ = (cosh η)z − (sinh η)t, and t ′ = −(sinh η)z + (cosh η)t.
This is a squeeze transformation as in the case of Eq. (17). In terms of these space-time variables, the wave function of Eq.(84), can be written as and the series of Eq.(65) then becomes Since the Lorentz-covariant oscillator formalism shares the same set of formulas with the Gaussian entangled states, it is possible to explain some aspects of space-time physics using the concepts and terminologies developed in quantum optics as illustrated in Fig. 1.
The time-separation variable is a case in point. The Bohr radius is a well-defined spatial separation between the proton and electron in the hydrogen atom. However, if the atom is boosted, this radius picks up its time-like separation. This time-separation variable does not exist in the Schrödinger picture of quantum mechanics. However, this variable plays the pivotal role in the covariant harmonic oscillator formalism. It is gratifying to note that this "hidden or forgotten" variable plays its role in the real First of all, does the wave function of Eq.(96) carry a probability interpretation in the Lorentz-covariant world? Since dzdt = dz ′ dt ′ , the normalization This is a Lorentz-invariant normalization. If the system is at rest, the z and t variables are completely dis-entangled, and the spatial component of the wave function satisfies the Shcrödinger equation without the time-separation variable. However, in the Lorentz-covariant world, we have to consider the inner product The evaulation of this integral was carried out by Michael Ruiz in 1974 [44], and the result was In order to see the physical implications of this result, let us assume that one of the oscillators is at rest with η ′ = 0 and the other is moving with the velocity β = tanh(η). Then the result is Indeed, the wave functions are orthnormal if they are in the same Lorentz frame. If one of them is boosted, the inner product shows the effect of Lorentz contraction. We are familiar with the contraction √ 1 − β 2 for the rigid rod. The ground state of the oscillator wave function is contracted like a rigid rod.
The probability density |ψ 0 η (z)| 2 is for the oscillator in the ground sate, and it has one hump. For the n th excited state, there are (n+ 1) humps. If each hump is contracted like √ 1 − β 2 , the net contraction factor is √ 1 − β 2 n+1 for the n th excited state. This result is illustrated in Fig. 8.
With this understanding, let us go back to the entanglement problem. The ground state wave function takes the Gaussian form given in Eq.(11) where the x and y variables are replaced by z and t respectively. If Lorentz-boosted, this Gaussian function becomes squeezed to [24,20,25] leading to the series 1 cosh η k (tanh η) k χ k (z)χ k (t). According to this formula, the z and t variables are entangled in the same way as the x and y variables are entangled.
Here the z and t variables are space and time separations between two particles bound together by the oscillator force. The concept of the space separation is well defined, as in the case of the Bohr radius. On the other hand, the time separation is still hidden or forgotten in the present form of quantum mechanics. In the Lorentz-covariant world, this variable affects what we observe in the real world by entangling itself with the longitudinal spatial separation.
In Chapter 16 of their book [9], Walls and Milburn wrote down the series of Eq.(1) and discussed what would happen when the η parameter becomes infinitely large. We note that the series given in Eq.(104) shares the same expression as the form given by Walls and Milburn, as well as other papers dealing with the Gaussian entanglement. As in the case of Wall and Milburn, we are interested in what happens when η becomes very large.
As we emphasized throughout the present paper, it is possible to study the entanglement series using the squeezed Gaussian function given in Eq.(103). It is then possible to study this problem using the ellipse. Indeed, we can carry out the mathematics of entanglement using the ellipse shown Fig. 9. This figure is the same as that of Fig. 6, but it tells about the entanglement of the space and time separations, instead of the x and y variables. If the particle is at rest with η = 0, the Gaussian form corresponds to the circle in Fig. 9. When the particle gains speed, this Gaussian function becomes squeezed into an ellipse. This ellipse becomes concentrated along the light cone with The point is that we are able to observe this effect in the real world. These days, the velocity of protons from high-energy accelators is very close to that of light. According to Gell-Mann [54], the proton is a bound state of three quarks. Since quarks are confined in the proton, they have never been observed, and the binding force must be like that of the harmonic oscillator. Furthermore, the observed mass spectra of the hadrons exhbit the degeneracy of the three-dimentional harmonic oscillator [43]. We use the word "hadron" for the bound state of the quarks. The simplest hadron is thus the bound state of two quarks.
In 1969 [55], Feynman observed that the same proton, when moving with its velocity close to that of light, can be regarded as a collection of partons, with the following peculiar properties.
1. The parton picture is valid only for protons moving with velocity close to that of light.
2. The interaction time between the quarks become dilated, and partons are like free particles.
3. The momentum distribution becomes wide-spread as the proton moves faster. Its width is proportional to the proton momentum.
4. The number of partons is not conserved, while the proton starts with a finite number of quarks.  Indeed, Fig. 10 tells why the quark and parton models are two limiting cases of one Lorentz-covariant entity. In the oscillator regime, the three-particle system can be reduced to two independent two-particle systems [43]. Also in the oscillator regime, the momentum-energy wave function takes the same form as the space-time wave function, thus with the same squeeze or entanglement property as illustrated in this figure. This leads to the wide-spread momentum distribution [20,56,57].
Also in Fig. 10, the time-separation between the quarks becomes large as η becomes large, leading to a weaker spring constant. This is why the partons behave like free particles [20,56,57].
As η becomes very large, all the particles are confined into a narrow strip around the light cone. The number of particles is not constant for massless particles as in the case of black-body radiation [20,56,57].
Indeed, the oscillator model explains the basic features of the hadronic spectra [43]. Does the oscillator model tell the basic feature of the parton distribution observed in high-energy laboratories? The answer is YES. In his 1982 paper [58], Paul Hussar compared the parton distribution observed in high-energy laboratory with the Lorentzboosted Gaussian distribution. They are close enough to justify that the quark and parton models are two limiting cases of one Lorentz-covariant entity.
To summarize, the proton makes a phase transition from the bound state into a plasma state as it moves faster, as illustrated in Fig. 10. The un-observed time-separation variable becomes more prominent as η becomes larger. We can now go back to the form of this entropy given in Eq.(92) and calculate it numerically. It is plotted against (tanh η) 2 = β 2 in Fig. 11. The entropy is zero when the hadron is at rest, and it becomes infinite as the hadronic speed reaches the speed of light.
This temperature is also plotted against (tanh η) 2 in Fig. 11. The temperature is zero if the hadron is at rest, but it becomes infinite when the hadronic speed becomes close to that of light. The slope of the curvature changes suddenly around (tanh η) 2 = 0.8, indicating a phase transition from the bound state to the plasma state.
In this section, we have shown how useful the concept of entanglement is in understanding the role of the time-separation in high energy hadronic physics including Gell-Mann's quark model and Feynman's parton model as two-limiting cases of one Lorentz-covariant entity.