SciELO - Scientific Electronic Library Online

 
vol.28 issue4Dynamics of the classical two dimensional XY ferromagnet at low temperatures: a memory function approachBending of light in the framework of R + R2 gravity author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand

Article

Indicators

Related links

Share


Brazilian Journal of Physics

Print version ISSN 0103-9733

Braz. J. Phys. vol.28 n.4 São Paulo Dec. 1998

http://dx.doi.org/10.1590/S0103-97331998000400024 

On the Duffin-Kemmer-Petiau algebra and the generalized phase space

 

Marco Cezar B. Fernandes(1) and J. David M. Vianna(1,2)
(1)Núcleo de Física Atômica, Molecular e Fluidos
Instituto de Física, Universidade de Brasília
70910-900 Brasília, DF, Brazil
(2)Instituto de Física, Universidade Federal da Bahia
Campus Universitário de Ondina
40210-340 Salvador, BA, Brazil

 

Received 9 September, 1998

 

 

The Duffin-Kemmer-Petiau (DKP) relativistic equation has been recently uded to study the interactions of spinless mesons with nuclei. In view of this interest and also the interest to determine the phase space picture for hadronic quantum theory, we present a derivation of the DKP equation of the generalized phase space proposed by Bohm and Hiley. Our development is based on the algebraic calculus introduced by Schönberg and uses the idea of algebraic spinors due to Riesz and Cartan. The free DKP particle and the more general case of the DKP particle in a prescribed external electromagnetic field are considered, and we obtain the DKP Liouville type equations for these cases.

 

 

I  Introduction

The description of quantum systems using a generalized form of the Liouville equation is apropriate in the context where one wants to use a classical formalism to understand the nature of physical attributes usually associated with quantum mechanics. Quantities like spin, for instance, can be given a formulation in terms of a Liouville type equation in a relativistic phase space [1]. In such cases the spinorial character of the theory is contained essentially in the algebraic structure of the Liouville superoperator. Therefore, it is important to know how the spin algebras are combined with the differential operators involved in the Liouville equation as well as the domain of these operators . This combined structure is then what we mean by a generalized Liouville equation. In the relativistic case the role played by the so-called geometric algebras has been given attention as a way of understanding the nature of the relativistic phase space [1]. Following suggestions of Schönberg and Bohm[1][2], Holland [3] derived relativistic phase space equations for a massive spin dv1.gif (116 bytes) particle in an external electromagnetic field. In his development, Holland takes the complex Dirac algebra C4 and obtains the phase space representation of the Dirac and Feynman-Gell-Mann equations.

Recently [4, and Refs. cited therein] the first-order relativistic Duffin-Kemmer-Petiau (D.K.P) equation has been analyzed as part of a program to study the interactions of spinless mesons with nuclei. Simultaneously (and independently) the D.K.P algebras which are associated to D.K.P equation have been studied from a modern perspective [5].

In view of this recent interest in the use of the D.K.P relativistic equation, the study of D.K.P algebras and also the interest to determine the phase space formulation for hadronic quantum theory [6], we propose in this paper a derivation of the phase space representation of the D.K.P equation for scalar particles. We will follow the algebraic calculus introduced by Schönberg[2] and the mathematical development of Bohm and Hiley [1] based on the idea of algebraic spinors [7]. In doing so, we show that it is also possible to write relations of classical type in a relativistic phase space by means of the D.K.P. algebra. The paper is organized as follows. In section 2, along the lines of Schönberg [2] , we begin with the D.K.P equation for the algebra in the sense that the D.K.P operator acts on a subspace of the algebra. This allows us to suggest a broader meaning for this operator aiming to generalize the equation for further integer spins. This will also yields a relation with the Dirac algebraic spinors. We will then take from that and use a relativistic version of the Wigner-Moyal transformation to make the passage to a phase space equation. Section 3 will follow with an interpretation of our D.K.P Liouville type equation corresponding to scalar particles in an external electromagnetic field. Section 4 closes with some conclusion remarks.In the Appendix the notation and some formal definitions related to spinor spaces, Grassmann, Dirac and D.K.P. algebras are presented.

 

II  Phase space formulation of the D.K.P relativistic equation

In this section, following Schönberg [2], we begin with the D.K.P equation in the algebra in order to give a broader meaning to this relativistic equation in terms of the algebraic spinors [7]. This will allow us to use the techniques introduced in [1] further on in the paper.

The D.K.P operator in the algebra combines the covariant vector /xm º m with the generator of the D.K.P algebra bm. The evolution D.K.P operator is the contraction

dv2.gif (328 bytes)

The free D.K.P particle of mass m is described by the following equation:

dvfo01.gif (383 bytes)

This is an equation defining the eigenstates Y of m bm . This equation is being seen here as a generalization of the usual D.K.P equation [2] in the sense that the Y is an element of the minimum left ideal of an extended Grassmann algebra. Following the Appendix, this Y takes the explicit form:

dvfo02.gif (838 bytes)

Now, from the general form

dvfo03.gif (1027 bytes)

it is easy to see that the action of a bm (p) on a spinor Y projects it down to the space of antisymmetric tensors of order p+1 and p. This then shows the difference between the Dirac algebraic spinors and D.K.P spinors in this algebraic context.

In the present work, we restrict our analysis to the scalar particles, i.e we take p = 0. Thus

dvfo05.gif (1019 bytes)

The Y which are elements of the space of representation of the operators above have the general form

dvfo07.gif (428 bytes)

They are expressed as a direct sum of scalars and vectors. This corresponds in the usual D.K.P. theory to the column representation

dvfo07-1.gif (788 bytes)

where f is a scalar function proportional to the projector (P) onto the scalars (see appendix) and the (fm ) are vector functions. In this case the D.K.P algebra coincides with the total matrix algebra of the (4+1) - dimensional space (see Appendix for more details).

We now aim to arrive at a phase-space D.K.P equation. Following [1] we first regard the free particle equation

dvfo08.gif (338 bytes)

and its adjoint

dvfo09.gif (346 bytes)

where

dvfo10.gif (370 bytes)

and

dvfo11.gif (374 bytes)

We define two sets of elements in the algebra:

dvfo12.gif (530 bytes)

and

dvfo13.gif (513 bytes)

It can be verified that these are generators of two D.K.P algebras satisfying the relations

dvfo14.gif (778 bytes)

and

dvfo15.gif (662 bytes)

It is convenient to relate the dvfo15-1.gif (145 bytes) and the dvfo15-2.gif (140 bytes) to the dvfo15-3.gif (148 bytes)and dvfo15-4.gif (145 bytes). For we define

dvfo16.gif (712 bytes)

where

dvfo16-1.gif (392 bytes)

The relations we want are

dvfo18.gif (552 bytes)

where

dvfo19.gif (498 bytes)

and it is verified that the element w anticommutes with all the dvfo19-1.gif (162 bytes) .

From these relations we can write

dvfo20.gif (671 bytes)

where we have adopted the convention

dvfo21.gif (625 bytes)

The construction of the above relations amounts to apply the techniques introduced in [1] to derive a Liouville type equation in the generalized phase space. The relations (20) are analogous to the relations between the Grassmann algebra and the Clifford algebra shown in [1]. Here, however, they have a different interpretation.

 

III  D.K.P equation on the generalized phase space

The above algebraic calculus is based on an algebra for vectors and covectors (see Appendix). Therefore it gives rise to a structure similar to the momentum phase space i.e the image of a local chart of a cotangent bundle [8]. It should be noted that the bilinear form B(see Appendix) used to construct the D.K.P algebra is not the symplectic bilinear form but a pseudo-euclidean form on phase space W. The space W enters the theory of the relativistic D.K.P particle in the usual (X, p) phase space in the following way: let M be the phase space of continuous variables (Xm , pm ) defined by the relativistic version of the Wigner-Moyal transformation [1].

dvfo23.gif (693 bytes)

where

dvfo23-1.gif (404 bytes)

The generalized relativistic phase space of the D.K.P particle can modulo relativistic constraints be viewed as the product space

dvfo23-2.gif (410 bytes)

where dvfo23-3.gif (245 bytes) is the spin space of the D.K.P particle in W. T is the total space. A state of the particle is represented by the Wigner function defined on T. The set of these states will then form the domain of the operators dvfo23-4.gif (424 bytes) that will be involved in the Liouville equation. Next, we derive this equation for the free D.K.P scalar particle. We begin with the D.K.P equation with no mass. The following steps serve to give a brief summary of the techniques used in [1] to obtain the Liouville operator. We go back to equations (8) and (9) and write it down for the density matrix V in the usual x-representation:

dvfo24.gif (854 bytes)

V is a function defined on the space T. The dependence on the spin indices has been dropped for convenience. Multiplying (25) by w and using the definitions (18) we get

dvfo26.gif (838 bytes)

Now we use the relations (20) to write:

dvfo28.gif (697 bytes)

We change variables according to

dvfo30.gif (474 bytes)

Thus (28) and (29) become

dvfo31.gif (1209 bytes)

The addition and subtraction of the above equations give

dvfo33.gif (1144 bytes)

Under the Wigner-Moyal transformation (23) these equations become

dvfo35.gif (1331 bytes)

These are the massless free-particle D.K.P equations on the generalized phase space T. The functions F(Xm, pm) has the following module structure

dvfo37.gif (494 bytes)

where F(Xm , pm ) is the Wigner function defined on the continuous variables only.

The square of the operator

dvfo38.gif (548 bytes)

is the Liouville operator on F. In fact, we have

dvfo38-1.gif (568 bytes)

or

dvfo38-2.gif (1125 bytes)

And, by using the relations

dvfo38-3.gif (897 bytes)

where [A, B]+ = AB+BA, we obtain

dvfo39.gif (611 bytes)

This is the classical Liouville equation for a free particle. Notice that the equation has two terms. The first one accounts for the scalar part of F and the second for the vector part of F. The operator

dvfo40.gif (539 bytes)

leaves the space of states stable which makes the equation algebraicaly consistent. It is noteworth to point out that besides the scalar part, there is also a term involving dvfo40-1.gif (191 bytes) . This is because the D.K.P is a matrix algebra of the tensors of order p and p+1. In the case in consideration, we are taking p = 0 which justify the term dvfo40-1.gif (191 bytes) corresponding to p+1. The making of the equation (39) opens the possibility of deriving the Liouville operator in a more general situation; we shall regard the D.K.P particle with mass m in an external electromagnetic field (in units dvh.gif (112 bytes) = 1 and metric (1,-1,-1,-1)). The equations for V are:

dvfo41.gif (677 bytes)

dvfo42.gif (670 bytes)

Multiplying (42) by w and using (18) we get:

dvfo43.gif (691 bytes)

dvfo44.gif (676 bytes)

We now use the relations (20) to write

dvfo45.gif (778 bytes)

dvfo46.gif (755 bytes)

Changing the variables to Xm and xm , according to (30), and further expanding in the potentials yields

dvfo46-1.gif (693 bytes)

Following the same steps which led to (33) and (34) one gets

dvfo46-2.gif (1203 bytes)

dvfo46-3.gif (1199 bytes)

which under the Wigner-Moyal transformation become

dvfo47.gif (1193 bytes)

dvfo48.gif (1187 bytes)

These are the D.K.P equations in the generalized phase space. The corresponding Liouville operator is the square of the operator on F given in (47). The square of(48) leads to the same equation. In order to compute the square of the above operator we first rewrite (48) taking into account the fact that w2 = 1. Thus,

dvfo49.gif (1052 bytes)

This equation is an eigenvalue equation for F. The Liouville operator LDKP is the square of the operator

dvfo50.gif (810 bytes)

with m2 as the eigenvalue. In order to compute the square we define the following notation

dvfo50-1.gif (833 bytes)

Now we compute

dvfo50-2.gif (577 bytes)

which gives

dvfo50-3.gif (1956 bytes)

Clearly one can distinguish two terms. Firstly, the scalar part, i.e,

dvfo51.gif (1042 bytes)

and secondly, the vector part

dvfo52.gif (1695 bytes)

The total Liouville operator is dvfo52-1.gif (168 bytes)+dvfo52-2.gif (166 bytes)-2im2. Due to the presence of the projector P0 º P (see Appendix) in (51), dvfo52-1.gif (168 bytes) operates only on the scalar part of F. We can leave P out of dvfo52-1.gif (168 bytes) since it is just a unit on its range. Thus (51) can be read without P. Hence dvfo52-1.gif (168 bytes) is just the Liouville operator for the Klein-Gordon equation with Euclidean metric dmn [3]. The two terms in dvfo52-1.gif (168 bytes) represent: the modified contribution to LDKP coming from the motion of the trajectory in phase space and the electromagnetic force, respectively.

In (53) we have the vector part of F being modified by two similar terms. Contrary to (51), these terms operate only on the vector part since they vanish on the scalar part. The third term shows the effect of the spin of the DKP particle which appears coupled with the electromagnetic field only in the vector part. Here there is a complete similarity with the Liouville operator for the Dirac equation derived in [1]. The term (dvfo52-3.gif (178 bytes)- dvfo40-1.gif (191 bytes)) is just the spin operator Smn exhibited in [9]. Notice that despite we are analyzing the scalar case (p = 0), there is always the p+1 term involved. As aforementioned, this is because the D.K.P algebra gives rise to a total matrix algebra of order p and p+1.

 

IV  Conclusions

We have shown how to apply the phase space approach proposed by Bohm and Hiley [1] in order to find a formalism describing bosonic particles. In such an approach the physical property of spin appears in a classical relativistic algebraic formalism in phase space. This space can be viewed as a product space which naturally embodies new kind of algebraic operators besides the usual differential operators that feature in the plain Liouville formalism. The generators of the D.K.P algebras are constructed in the standard fashion used for deriving Clifford algebras out of bilinear forms.

Via a relativistic version of the Wigner-Moyal transformation we have arrived at two Liouville type equations. One for free particles and the other for particles in the presence of an electromagnetic field. In previous works on this subject Bohm and Hiley studied the Dirac equation and its non-relativistic case [10]. Therefore our results add to the program since we were able to exhibt an algebraic development for the relativistic phase space describing bosons as well. As a consequence, the interpretation of the solutions of the Liouville equation as spin fields having certain independence of motion, and the phase trajectories as whole depending on the interaction of these fields with the electromagnetic field has also been verified for bosonic systems.

So far we have obtained the Liouville equation for D.K.P particles corresponding to p = 0 in the D.K.P algebra. However it is possible to generalize this result for any allowed values of the label p. A further work in this direction is in progress and should be published in a forthcoming paper.

 

V  Appendix

V-I The extended Grassmann algebra of a phase space

Let V be a n-dimensional vector space and V* its dual space. A phase space W is defined as

W = V ÅV* = {(v,u), v Î V, u Î V*}.
(54)

Let B be a bilinear form defined on W, i.e. B: W×W® R  or  C   (Real or Complex field) given by

2B( (v,u),(v¢,u¢)) = áu,v¢ñ+áu¢,vñ
(55)

where á·,·ñ is the natural pairing of vectors and covectors. This bilinear form is non-singular and establishes an isomorphism between W and W*; the choice of this bilinear form on W embeds it into a Clifford algebra but in this case a Clifford algebra of the phase space W.

The standard way to construct the algebra from B is as follows: Let T(W) be the tensor algebra over W i.e.

T(W) = ¥
Å
i = 0 
T(i)(W)

where T(i) = WÄWļÄW (i times) and W is identified with the 1-piece T(1)(W) of T(W).

From the universal property of T(W) one has

dvapp.gif (1855 bytes)

where C(W) is an associative algebra. Let I be the two-sided ideal

wÄw¢+w¢Äw-2B(w,w¢)1T(W).

in T(W). Define C(W) = T(W)/I. Therefore one has

f( wÄw¢+w¢Äw-2B(w,w¢)1T(W)) = 0,

since f is an homomorphism. This implies that

[f(w),f(w¢)]+ = 2B(w,w¢)1C(w)
(56)

Writing w as w = v Å u, v Î V and u Î V* one gets the extended Grassmann algebra of vectors and covectors

[f(u),f(u¢)]+ = [f(v),f(v¢)]+ = 0
(57)
[f(v),f(u)]+ = áu,vñ1C(w)
(58)

where one may have the identifications f(v) = v, f(u) = u.

Writing v and u in a basis i.e.

v = viei,       u = uiei

and using linearity of f one gets

[ei, ej]+ = dij 1C(w)
(59)
[ei, ej]+ = [ei, ej]+ = 0
(60)

Therefore C(W) is an algebra generated by vectors and covectors with basis comprised by the 22n elements

(e1)r1 ¼(en)rn(en)sn ¼(e1)s1 ;     r,s = 0,1

In order to build up a canonical basis for C(W) one defines the following primitive idempotent

P = N1 ¼Nn;     n = dim V
(61)

and

Nj = ejej,        no   sum.

It is easily verified that P2 = P. The canonical basis for C(W) is written as

dvfo61-1.gif (521 bytes)

which implies

dvfo62.gif (1018 bytes)

Hence these 22n elements are linearly independents and a general element of C(W) can be expressed as a linear combination

dvfo63.gif (862 bytes)

or

dvfo64.gif (705 bytes)

where we have adopted the multi-index notation

Jp = j1 ¼jp
Kq = k1¼kq

and sum in repeated indices is also understood.

 

V-II The spinor spaces of C(W)

We shall now consider the regular representation of C(W) . The regular representaion of an algebra is a representation of the algebra on itself. An irreducible representation of C(W) can be constructed on its minimum left ideals [7]. According to Cartan [12], the minimum left ideal of the Clifford algebra can be taken as the space of spinors of the algebra. The group of inner-automorphisms of the algebra acts irreducibly on the minimum left ideals thus yielding a spinorial representation. A way to construct minimum left ideals in an algebra is to use a primitive idempotent. In our case the element P introduced in (61) happens to be a primitive idempotent[13] since,

P2 = P, P* = P,     PGP = cP;       G Î C(W), c Î C.

Hence a minimum left ideal of C(W) can be formed by the projection

GP

for every G Î C(W). We denote the projected space by CP(W).

Analogously one can form the space PG denoted by CP(W), which is a minimun right ideal of C(W). It turns out that C(W) is a matrix algebra and has a representation on the subspaces CP(W) and CP(W). Let us see how this result works in terms of components. For take

dvfo64-1.gif (617 bytes)

and project it down into CP(W),i.e

dvfo64-2.gif (690 bytes)

By taking into account the relations ei(P) = (P)ej = 0 as well as rules (62) we get

dvfo64-3.gif (773 bytes)

Now the space CP(W) of elements Y plays the role of the space of representations of C(W). Indeed it can be easily verified that

G: Y¢
GY = Y¢;     G Î C(W);    Y,  Y¢ Î CP(W).

 

V-III The Dirac algebra

The algebra C(W) developed above is an affine algebra since it is generated by a set of vectors and covectors defined independently. When a metric is available one can distinguish two metric subalgebras in C(W) generated by

gm (±) = em ±gmnen ,

one for each sign of the metric.

Using relations (60) it follows that

[gm(±),gn(±)]+ = ±2 gmn 1C(W)
(65)
[gm(+),gn(-)]+ = 0
(66)

Here an important point on the notation is in order[2]. The element noted by gmnen does not mean to lower the indices n with the metric. We just mean that there is a sum on n thus leaving free the indice m to balance the equality. In this regard we should point out that the multiplication rules of the algebra depend on the covariant or contravariant nature of the generators.

The elements gm+ generates a Clifford algebra corresponding to the positive metric gmn. When n = 4 and gmn is the Minkowski metric, we have the Dirac algebra of space-time.

 

V-IV The Duffin-Kemmer-Petiau algebras

The elements

dvfo64-4.gif (323 bytes)

are idempotent elements of C(W). The unity of the algebra has an idempotent decomposition

1C(W) = n
å
p = 0 
Pp;        P0 : = P

The elements Pp satisfy some important properties, namely,

(Pp)(Pq) = dp,q (Pp)
(e j )(Pp) = (Pp+1)(e j )
(Pp)(ej) = (ej)(Pp+1)

The generators of the D.K.P algebra which is denoted by b(p)m satisfy the following fundamental relations:

b(p)mb(p)nb(p)g +b(p)gb(p)nb(p)m = gmnb(p)g + ggn bm(p)
(67)

These generators have an expression within C(W) as follows

dvfo68.gif (1285 bytes)

where "p'' is the order of the linear space into which Pp projects the Y Î C(W).

The element (Pp) + (Pp+1) is idempotent and commutes with all the bj(p). Thus (Pp) + (Pp+1) is the unity of the D.K.P algebra.

Despite one has a metric in the defining relations (67), the D.K.P algebra is not always a metric subalgebra of C(W). In fact,one can introduce the affine algebra Dn,p [2] generated by

(Pp)(ej);     (e j )(Pp)     and     (Pp+1)+ (Pp)

It is clear that Dn, p contains D.K.P as a subalgebra. In fact, Dn, p coincides with D.K.P algebra when dvfo68-1.gif (198 bytes) [2]. In this case, D.K.P is an affine algebra even if the components of the bj(p) are defined in terms of the components of the metric tensor. As far as C(W) is concerned the algebra Dn, p is the total matrix algebra of the direct sum of the linear subspaces of CP(W) corresponding to covariant antisymmetric tensors of order "p'' and "p+1''. This total matrix algebra is the D.K.P algebra when dvfo68-1.gif (198 bytes). We shall denote the space of matrices of order p and p+1 by Dpp+1(W)

 

References

[1]
D. Bohm and B.J. Hiley, in ''Old and New Questions in Physics, Cosmology, Philosophy, and Theoretical Biology'', A. van der Merwe, ed. (Plenum, New York, 1983), p.67.
[2]
M. Schönberg, Anais Acad. Bras. Cienc. 28, 11(1956), 473 (1957); 30, 1, 117, 259, 429 (1958); Suppl. Nuovo Cimento (X) 6, 356 (1957).
[3]
P.R. Holand, Found. Phys. 16, 701(1986)
[4]
Y. Nedjadi and R. C. Barrett, J. Phys. G.: Nucl. Part. Phys. 19, 87 (1993)
[5]
O. A. Sánchez-Valenzuela and R. E. Zuazua-Vega, J. Phys. A: Math. Gen. 23, 4967 (1993)
[6]
Y. S. Kim and M. E. Noz, ''Phase Space Picture of Quantum Mechanics'', World Scientific, London (1991)
[7]
M. Riesz, in Comptes Rendus 12me Cong. Math. Scand. (Lund, 1953), p. 241
[8]
L. H. Loomis and S. Sternberg, ''Advanced Calculus'', Addison-Wesley, Reading, Mass. (1968).
[9]
N. Kemmer, Proc. R. Soc. A 173, 91 (1939)
[10]
D. Bohm and B.J. Hiley, Found. Phys. 11, 179(1981).
[11]
R. Hermann,''Spinors, Clifford and Cayley Algebras'', Interdisciplinary Mathematics Vol. VII. Math. Sci. Press. USA (1974)
[12]
E. Cartan, ''The Theory of Spinors'', Dover Publications, New York (1966)
[13]
M. C. B. Fernandes, ''Geometric Algebras and the Foundations of Quantum Theory'', Ph.D. Thesis, Birkbeck College, Dept. Phys., London University (1995)