Вы находитесь на странице: 1из 9

Studies in History and Philosophy of Modern Physics 44 (2013) 206 214

Contents lists available at ScienceDirect

Studies in History and Philosophy of Modern Physics


journal homepage: www.elsevier.com/locate/shpsb

Holism and nonseparability by analogy


Aristidis Arageorgis
Department of Humanities, Social Sciences, and Law, School of Applied Mathematical and Physical Sciences, National Technical University of Athens, GR-15780 Zografou Campus, Athens, Greece

art ic l e i nf o
Article history: Received 3 December 2008 Received in revised form 21 March 2013 Accepted 2 May 2013 Available online 11 July 2013 Keywords: Holism Nonseparability Quantum eld theory

a b s t r a c t
This paper explores the issues of holism and nonseparability in relativistic quantum eld theory (QFT) by focusing on an analog of the typical model featuring in many discussions of holism and nonseparability in nonrelativistic quantum mechanics. It is argued that the quantum eld theoretic model does exhibit holism in a metaphysical sense and that there are plausible grounds to view QFT holistic in an epistemological sense. However, the complexities arising from the fact that quantum elds have innite degrees of freedom prohibit the exploitation of the elaborated analogy toward demonstrating that the QFT model exhibits the kind of state nonseparability familiar from ordinary quantum mechanics. Still, it is argued that the QFT model does satisfy a rather weak epistemological criterion for state nonseparability. & 2013 Elsevier Ltd. All rights reserved.

When citing this paper, please use the full journal title Studies in History and Philosophy of Modern Physics

1. Introduction As regards the worlds canvassed by the theories of physics, holism is usually taken to be a metaphysical thesis. Paraphrasing and simplifying a denition proposed by Healey (1991; p. 402) and Healey (2008; p. 5), one might say that a compound physical system (the whole) exhibits physical property holism if it can possess under certain conditions qualitative intrinsic physical properties that do not supervene upon qualitative intrinsic physical properties and relations of its component subsystems (its basic physical parts). In the same nexus of philosophical concerns arises the issue of nonseparability. It is said that a compound system exhibits state nonseparability if there exist states that can be assigned to it under certain conditions but which do not supervene upon the states assigned under the same conditions to its component subsystems (Healey, 2008; p. 7). On an alternative approach, Seevinck (2004) has advocated instead an epistemological criterion for characterizing holism in physical theories. Roughly, a physical theory is holistic if and only if it is impossible in principle for a set of local agents, each having access to a single subsystem only, to infer the global properties of a compound system, as those properties are assigned by the theory and ascertained by global measurements, by using the resource basis available to the agentswhere the resource basis includes (at least) all local operations and classical communication (Seevinck, 2004; p. 705).

It is not my aim in this paper to explore the philosophical subtleties involved in these approaches to holism and nonseparability in physics. My only aim is to use them as a springboard in order to probe an analogy between the typical system that is taken to exhibit holism and nonseparability in ordinary quantum mechanics and a system treated in relativistic quantum eld theory (QFT).1 I shall argue that the quantum eld theoretic system does instantiate physical property holism in the metaphysical sense as well as that there are plausible grounds to deem the describing theory holistic in the epistemological sense. I shall also argue that, due to the intricacies of the quantum theory of systems having innite number of degrees of freedom, the quantum eld theoretic model cannot be shown via the elaborated analogy to exhibit state nonseparability, albeit in a rather weak epistemological sense: the compound system admits states such that no agent conned to a single subsystem can determine the state he is in or retrieve knowledge about this state from any complete knowledge about objects intrinsic to the subsystem he has access to. Clearly any conclusion concerning the holistic aspects of any quantum theory presupposes a property assignment rulea rule that species the conditions under which a quantum system
1 As it is well known, the typical system taken to exhibit holism and nonseparability in ordinary quantum mechanics is a pair of spin 1/2 particles in an entangled spin state. Redheads (1995; pp. 128133) baby ReehSchlieder theorem must be recognized as a pioneering result toward establishing a rigorous analogy between this system in ordinary quantum mechanics and a relativistic quantum eld in its vacuum state.

E-mail address: arage@central.ntua.gr 1355-2198/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.shpsb.2013.05.003

A. Arageorgis / Studies in History and Philosophy of Modern Physics 44 (2013) 206 214

207

possesses a property. And, as it is well known, property assignment rules vary from one interpretation to another. In this paper, I shall conne myself to the orthodox interpretation that employs the socalled eigenstateeigenvalue link as a property assignment rule: a quantum system has the property that the physical quantity A has the denite value a if and only if its state is an eigenstate of the selfadjoint operator representing A associated with the eigenvalue a. Of course, the philosophy of physics has witnessed, over the past few decades, an accumulation of important works probing holistic aspects of relativistic quantum elds. Summers and Werner (1988, 1995) have demonstrated the generic character of maximal violations of Bell inequalities across spacelike-separated spacetime regions in QFT. Clifton and Halvorson (2001) revealed, on the grounds of rigorous arguments, the general impossibility of disentangling, by means of local operations, the local state of a QFT system from that of its spacelike complement (see also Section 6 below). And the list can go on including arguments that are based, not only on algebraic QFT, but also on other formulations of QFT. For instance, Waynes (2002) interpretation of QFT in terms of vacuum expectation values of products of eld operators (VEVs) intimates a kind of spatiotemporal nonseparability in as much as the VEV 0jx1 :::xn j0, dened at the n distinct spacetime points x1 ; ; xn (n 1; 2; ), does not supervene on qualitative intrinsic physical properties assigned to those n points together with the spatiotemporal relations among those n points.2 Against the background of this extensive literature, this paper aspires to explore the applicability to QFT of more meticulously explicated concepts of holism and nonseparability. To this end, an effort is made to assess, not only whether QFT provides instances of the aforementioned philosophically worked-out notions of holism and nonseparability, but also whether attendant notions like whole, part or intrinsic property can appropriately be made sense of. The plan of the rest of the paper is as follows. In Section 2, I review the standard argument marshaled toward justifying the conclusion that ordinary nonrelativistic quantum mechanics exhibits physical property holism and state nonseparability. The model deployed is the usual one of a pair of spin 1/2 particles in an entangled state. In Section 3, I sketch the abstract structure of an analogous model in QFT; and in Section 4, I present in bare outline the construction of a concrete model that satises this abstract structure. The model has attracted intense interest in research on quantum eld theory in curved spacetime in connection with the thermalization of ambient vacuum states in regions of spacetimes exhibiting bifurcate Killing horizons (see, e.g., Wald (1994)). In fact, the technical work I shall rely heavily on was carried out by Kay (1985a, 1985b, 1985c) in the context of that very research. In Section 5, I draw the main conclusions concerning holism and nonseparability in the particular QFT model and in the nal Section of the paper I offer some considerations substantiating a few assumptions these conclusions rest on. 2. The standard argument in quantum mechanics The standard argument in support of nonseparability and holism in the metaphysical approach goes like this (e.g., Maudlin (1998)). Consider a pair of spin 1/2 particles produced in the singlet state of the total spin 1 ! ! ! ! j 0 p j n ; L j n ; R j n ; L j n ; R 2 1

! ! where the vector n species some direction in space and j n ; 7 L is the eigenstate of the particle that ends up in the left wing (hence, the superscript L) of an EPR-Bohm experiment corre! sponding to the eigenvalue 7 1/2 for spin in the direction of n (similarly for R).3 The singlet state is a pure state for the compound system in which the total spin !R !L S 1 1 S 2 2

has the denite value 0. However, when the pair of particles is in the singlet state, the state of each particle is mixed and described by the density operator Dk 1 ! 1 ! ! ! j n ; k n ; j j n ; k k n ; j; 2 2 k L; R 3

In such a state, no particle has a denite value of spin. In fact, such a state describes a completely unpolarized or disordered spin: its von Neumann entropy Sk tr k Dk logDk log 2; k L; R 4

assumes the maximum value it can take on a 2-dimensional Hilbert space. Further, if the pair of spin 1/2 particles were produced instead in the state 1 ! ! ! ! j 2 p j n ; L j n ; R j n ; L j n ; R 2 5

then each of the component particles would be in a state described again by the density operator Dk of (3)i.e., it would be in the same state as if the particle pair were produced in j 0 . However, j 2 is a pure state in which the total spin has the denite value 2. State nonseparability is readily established. There is a difference in the states of the pair to which no difference in the states of the individual particles corresponds. Therefore, the state assigned to the compound system does not supervene upon the states assigned at the same time to its component subsystems. Physical property holism follows suit. There is a difference in properties of the pairnamely, the values of the total spinto which there corresponds no difference in the properties of the individual particles. Thus, there are qualitative intrinsic physical properties of the compound system that do not supervene on qualitative intrinsic physical properties and relations of the component subsystems simply because there arent any such properties. The crux of the issue is that whether in j 0 or in j 2 the individual particles have no nontrivial intrinsic spin properties (cf. Healey (1991; pp. 419421)).

3. The formal analogy To construct an analog of the entangled pair of spin 1/2 particles in QFT, we can start from a mixed state and purify it by doubling the system. In the quantum mechanics of systems with nite number of degrees of freedom, there is a well-known procedure by which a statistical mixture may be regarded as the restriction to the given system of a pure state on a system twice as large. For deniteness, start with the density operator Z 1 eH ; Z treH ; 0 o o 6

2 The interpretation rests on the fact that a QFT can be fully recovered from the complete set of its VEVs. This fact is supported by the so-called reconstruction theorem, rigorously proved on the Wightman approach to QFT (see Streater & Wightman (1964; pp. 117126)). As Healey (2008; p. 20) points out, however, the claim that this interpretation of QFT is indeed characterized by spatiotemporal nonseparability hinges on whether the VEVs may be thought of as representing qualitative intrinsic physical properties of n-tuples of spacetime points or physical relations among those points.

which represents the state of thermal equilibrium at absolute temperature T 1 for a quantum system with nite number of
3 Some conventions. Throughout I use units in which c k 1. By BH I denote the set of all bounded linear operators on a Hilbert space H. By I denote the conjugate of a complex number and by An the adjoint of an element A of a C*algebra.

208

A. Arageorgis / Studies in History and Philosophy of Modern Physics 44 (2013) 206 214

degrees of freedom and Hamiltonian H on some Hilbert space H. Such states may be viewed as resulting from the adaptation of Gibbs canonical ensemble to quantum mechanics. Given that H has a positive discrete spectrum with a level density increasing less than exponentially for large eigenvalues, the operator eH is of trace-class (i.e., treH o ) and the expectation value of an observable ABH in the state is given by A trA 7

A; BBH and t , one has t AB Z 1 treH eitH AeitH B Z 1 trBeit iH AeitH Z 1 treH BeitiH AeitiH Bti A ; 12 provided that the dynamical automorphisms t of the C*-algebra BH are unitarily implemented with H being the innitesimal generator. Second, KMS states have been shown to exhibit the main characteristics of equilibrium. Specically, it has been shown that a KMS state at temperature 1 exhibits stationarity, the thermal reservoir property in the sense that it will drive any nite system to which it is coupled into thermal equilibrium at the same temperature, local thermodynamical stability in the sense no local perturbation will create a state with lower free energy, passivity in the sense that it precludes the extraction of energy from the system by means of a cyclic process, etc.6 Whats more, for nite or discrete systems, KMS states may be thought of as states having maximal entropy among the states with the same prescribed expectation values for energy and particle number (density). And it sounds reasonable to extrapolate this point of view to systems of eld theory despite the fact that relatively little rigorous work has been done in dening entropy density for continuous systems.7 Given any quantum dynamical system, one can construct a double quantum dynamical system that contains the former as one of its subsystems. ~ ; ~ t ; j conDenition 1. A double quantum dynamical system A ~ AL AR where AL , AR are two commuting sists of a C*-algebra A ~ ~ t , t , of A subalgebras, a 1-parameter group of automorphisms L R L R which map each of A , A onto A , A respectively, and an ~ which commutes with each involutory antiautomorphism j on A ~ t ( ~ t j j ~ t ) and which maps AL onto AR and AR onto AL .8 The original quantum dynamical system is then identied with ~ t jAR . The stratagem of purication consists in passing from a AR ; KMS state on a quantum dynamical system to a pure state on the associated double quantum dynamical system. The key concept here is that of a double KMS state. ~ at value 4 0 over a double Denition 2. A double KMS state ~ ; ~ t ; j is an ~ t -invariant and j-invarquantum dynamical system A ~ whose GNS triple H ~ ; ~ separable ~ has H iant state over A ~ ; and is such that ~ is cyclic for (i) ~ AR alone (similarly for R-L). ~ which (ii) There is a selfadjoint operator (total Hamiltonian) H satises ~ ~ ~ t A eit H ~ ~ AeitH ; ~ it H ~ ~ ~ 0; and ~ hence; H e ~ ~ ~ ~ AAR ~ AR domeH =2 and eH =2 ~ A ~ jA* similarly for RL and eH =2 e H =2 :
~ ~

Choose now a basis of H formed by eigenvectors j of H with corresponding eigenvalues Ej so that Z eEj
j

and dene on the Hilbert space HH the vector Z 1 eEj =2 j j


j

For any observable ABH, one has tr A ; 1AHH 10 So the expectation value in of any observable A of the single system described by means of H equals the expectation value in of the observable 1A of the double system described by means of HH. Thus the mixed (thermal) state may be regarded as the restriction to one subsystem of the pure state of the double system. Kay (1985b) has worked out the generalization of this procedure of purication of thermal states so that it applies as well to quantum systems with innite numbers of degrees of freedom and Hamiltonians with continuous spectra. For such systems, including those of QFT in innitely extended regions of spacetime, the operator eH is not of trace-class and consequently thermal equilibrium states cannot be dened via (6) and (7). So before we proceed to Kays work per se, it is expedient to review some preliminaries.4 A quantum dynamical system A; t consists of a C*-algebra A and a 1-parameter group of automorphisms t , t , of A. The C*algebra A contains the (bounded) observables of the system (selfadjoint elements of A) and the states of the system are represented by normalized positive linear functionals over A. The 1-parameter group ft gt prescribes the dynamics. A state over A is called an t ; -KMS state for some 4 0 if and only if t AB Bt i A
5

11

for all t and all A; BA. According to the received interpretation, which has been developed since the seminal paper by Haag, Hugenholtz, and Winnik (1967), such a state describes thermal equilibrium of the quantum dynamical system at temperature 1 . The interpretation is rmly supported by several considerations. First, it has been demonstrated that, for quantum systems for which eH is of trace-class, the canonical Gibbs state dened by means of (6) and (7) is the only state that satises the KMScondition (11) plus the analyticity and continuity properties that accompany (11) in the rigorous treatmentsconditions I have simply omitted here. The verication that it does satisfy (11) is matter of a simple calculation that relies upon the cyclicity and the invariance under similarity transformations of the trace. For all
4 Most of the arguably less familiar background material is part of the algebraic formulation of the quantum theory of systems with innite number of degrees of freedom. Standard references include, among others, Bratteli & Robinson (1979, 1981) and Haag (1992). Kadison & Ringrose (1997) is addressed mainly to mathematicians. Halvorson (2007) has achieved an admirable balance between mathematical detail and philosophical perspective. Lastly, much of the relevant material is included in an excellent review published in this journal (Rdei & Summers, 2007). 5 Here and elsewhere I omit signicant mathematical niceties, which, however, play no role in the sequel. For more rigor in the denition of a KMS state the reader may consult the references cited in the next footnote.

(iii) There is a unique antiunitary operator J such that ~ ~ and J ; eitH ~ ~ AJ ; J 0: ~ jA J

~ guarantees, in the GNS ~ t -invariance of Note that the ~ by ~ , the existence of a unique representation of A
6 Some of the relevant proofs rest on additional physically motivated hypotheses. For details see Bratteli & Robinson (1981; chapter 5), Haag (1992; chapter V), or Emch (2007; pp. 11141154). 7 See, however, Narnhofer (1994). Buchholz & Junglas (1989) had previously demonstrated the existence of KMS states for a large class of QFT models. 8 ~ -A ~ is involutory means that j2 is the identity and that it is an That j : A ~ and one has jA B jA jB, antiautomorphism means that for all A; BA jA A, jAB jAjB, jAn jAn, and j1 1, where 1 here denotes the C*algebra unit.

A. Arageorgis / Studies in History and Philosophy of Modern Physics 44 (2013) 206 214

209

~ onto ~ t and mapping 1-parameter unitary group implementing ~ . The additional assumption involved in the rst part of (ii) is that the aforementioned group is strongly continuous so that the ~ is implied via Stones existence of its innitesimal generator H ~ theorem. Note also that (iii) is entailed by the j-invariance of ~ t. and the fact that j commutes with Now, the restriction to AR of a double KMS state at value is a KMS state (in the usual sense) at inverse temperature with respect to the restriction to AR of the dynamical automorphism group of the double system. The following simple computation, which utilizes the properties of the various objects introduced by Denition 2, indicates the truth of this assertion. For all A; BAR and t , one has ~ ; eitH ~ ~ ~ t AB ~ Aeit H ~ B ~ it H ~ ~ ~ A* ; e ~ B ~ ; eitH eH =2 J ~ ~ A ~ B* eH =2 J
~ ~ ~ =2 ~ ; eH ~ eitH eH =2 ~ B* ~ A ~ ~ ; eitiH ~ ~ A ~ B* ~ B ~ ti A: ~ ~ ~ ~ ~

R ~ be a pure with eigenvalue zero (so that ML , M be factors and ~ , courtesy of Theorem 1) and (ii) the von Neumann state over A R algebras ML , M are type III. This will not affect the validity of the nal conclusions since both assumptions are provably true for the concrete model constructed in Section 4. R Since ML and M are von Neumann factors and are commuR tants of one another, the pair ML ; M satises the so-called Schlieder condition and is therefore C*-independent. This means R R that for every state L on ML and every state on M there exists ~ such that ~ on BH a state

~ AB L AR B ~ A ~ B AML
13 BMR .

15

13

That is, the KMS-condition (11) is satised. Loosely speaking, then, a double KMS state at value on the compound system becomes a state of thermal equilibrium at temperature 1 when restricted to each of the component subsystems. The important point is that the double KMS state itself is (under suitable assumptions) a pure state over the compound system. This is part of the content of the following theorem (Kay, 1985b; pp. 1033 1035). To avoid excess clutter in the notation, I shall denote in ~ by what follows the von Neumann algebras ~ AL , ~ AR , ~ A 9 L R ~ M , M , M . ~ be a double KMS state over a double quantum Theorem 1. Let ~ ; ~ ; ~ be its associated GNS ~ t ; j and H ~ ; dynamical system A
R triple. Then the von Neumann algebras ML and M are commutants ~ is the only (up to phase, normalof one another. Furthermore, if ~ with eigenvalue zero,10 then ML and MR are ized) eigenvector of H

and all C*-independence is an important for all aspect of the situation when it comes to investigating issues of R holism and nonseparability. Given that ML and M contain the algebras of observables pertaining to the two subsystems of the double quantum dynamical system, the C*-independence of the R pair ML ; M guarantees that no preparation of a specic state on the one subsystem precludes the preparation of any state on the ~ is other subsystem. Whats more, assuming that every state of M physically realizable, any two partial states on the two subsystems may be physically realized via a common preparation procedure. ~ Consider now the faithful normal state R induced by on R M , ~ ~ R A ; A ; AMR : 16

~ is irreducible (i.e., ~ ). ~ is a pure state over A factors and ~ A

~ is cyclic for ~ AR (Denition 2) By combining the fact that R with the fact that ML and M are commutants (Theorem 1) and by ~ exploiting the symmetry L2R,11 one can infer that the vector is cyclic and separating for each of the von Neumann algebras ML and MR . This allows for the application of the TomitaTakesaki ~ modular theory to the pair MR ; : the modular conjugation is ~ identied with J and the modular operator with eH . Then ~ ~ ~ and Tomitas theorem yields J
L J MR J M
12

Clearly R is invariant under the modular automorphism group R ~ of MR relative to . And because M is a type III factor, the R ~ modular automorphism group of M ; is not an inner group of t automorphisms. This means that the unitary operators i , t , do 14 not belong to MR . In this sense, the total dynamics of the double quantum dynamical system is not described by objects belonging to the component subsystems. Summing up, a double quantum dynamical system, regarded as a compound system, admits pure states whose restrictions to the component subsystems may be viewed as states of maximal entropy. This constitutes the rst aspect of the analogy with the standard spin example in ordinary quantum mechanics. Of course, this much falls short from establishing any direct conclusion as to holism or nonseparability. Whats more, the entire discussion so far has been carried out in the abstract. It is time to provide some content by presenting a concrete model of a double quantum dynamical system exhibiting double KMS states. Such a model is furnished by the quantum theory of a free scalar eld on Minkowski spacetime satisfying the KleinGordon equation. The rigorous mathematical work is due, again, to Kay (1985c). What follows is a mere sketch of the construction.

and

R i t t i MR M

14 4. The realization in quantum eld theory Consider the KleinGordon eld m2 0; ab a b ; m 4 0


4

for all t . In what follows, I shall conne attention to models in which (i) ~ ~ is indeed the only (up to phase, normalized) eigenvector of H
Given any subset S of the set BH of all bounded linear operators on a Hilbert space H, by S we denote its commutant, i.e., the set of all bounded linear operators on H which commute with every element of S. If S is a selfadjoint set (i.e., contains along with each element also its adjoint), then its double commutant, S S, is the smallest von Neumann algebra containing S. 10 ~ , H ~ ~ in the domain of H ~ 0 Strictly, this means that for every vector ~ ~ for some . implies 11 Recall that a vector is cyclic for a von Neumann algebra if and only if it is separating for its commutant and that for every selfadjoint subset S of BH for a Hilbert space H, S S. 12 For the TomitaTakesaki modular theory consult almost any book on operator algebras and their applications to physics (e.g., Bratteli & Robinson (1979; chapter 2.5) or Haag (1992; chapter V.2)) or the brief and up-to-date overview by Summers (2006).
9

17

on Minkowski spacetime ; ab endowed with a global inertial coordinate system T ; X ; Y ; Z in which diag 1; 1; 1; 1. Let be the Cauchy surface corresponding to T 0 and introduce
13 For the relevant mathematical concepts and facts see Summers (1990) or Florig & Summers (1997). 14 A von Neumann algebra M with a faithful normal state is type III if, and only if, for any nonzero projection E in the center MM the restriction to ME of the modular automorphism group of M associated to is not an inner group of automorphisms of ME(Bratteli & Robinson, 1981; p. 130). Of course, if M is a factor, then its center MM consists only of the complex multiples of the unit element.

210

A. Arageorgis / Studies in History and Philosophy of Modern Physics 44 (2013) 206 214

the antisymmetric nondegenerate bilinear form Z s1 ; 2 1 a 2 2 a 1 na d ;

18

where na is the unit normal to , on the space S of smooth real solutions to (17) with compact support on . Following tradition,15 let us call right Rindler wedge the part of Minkowski spacetime demarcated by X 4 jT j and denote it by R and let us call left Rindler wedge the part of Minkowski spacetime demarcated by X o jT j and denote it by L. Lastly, let us call L , R the portions of lying in L, R respectively and denote by SL , SR the subspaces of smooth real solutions to (17) with compact support on L , R respectively. In the symplectic space S; s we dene a 1-parameter symplectic group T t , t , via T t T ; X ; Y ; Z t T ; X ; Y ; Z 19 where t represents the Lorentz boost in the X -direction with speed tanh t so that 0 1 0 10 1 T T cosh t sinh t 0 0 B X C B sinh t cosh t 0 0 CB X C B C B CB C t B C B 20 CB C @Y A @ 0 0 1 0 A@ Y A Z 0 0 0 1 Z in matrix notation. The structure SR ; s; T t constitutes a classical linear dynamical system and one can deploy a well-known procedure to dene the corresponding quantum dynamical system AR ; t . AR is dened as the associated Weyl algebra of the canonical commutation relations (CCRs)i.e., as the C*-algebra generated by elements W , SR ; satisfying W 1 W 2 ei s1 ;2 =2 W 1 2 and W W :
R
n

Fig. 1. The double wedge system in Minkowski spacetime.

~ AL AR . The 1-parameter group ~ t , t , of automorphisms of A ~ A is dened with the help of (19) via its action on the Weyl operators ~ t W W T t : 26

It remains to dene the involutory antiautomorphism j on the double wedge algebra. Geometrically, it corresponds to the wedgereversal map that takes points T ; X ; Y ; Z onto T ; X ; Y ; Z . At the level of the double classical linear dynamical system, it is implemented by the antisymplectic involution , T ; X ; Y ; Z T ; X ; Y ; Z
L R R L

27

which maps S onto S and S onto S . And at the level of the double quantum dynamical system, it is dened by the corresponding action on the Weyl operators jW W : 28 ~ ; ~ t ; j is a double quantum dynamical system as per Thus A Denition 1.16 In order to pass over to the Hilbert space theory, Kay (1985c; pp. 7073) constructs a regular ground 1-particle structure for the classical linear dynamical system SR ; s; T t and with its aid ~ at value denes explicitly, for each 4 0, a double KMS state ~ ; ~ t ; j(Kay 1985c; pp. 6263). If h is the pertinent 1 over A particle Hilbert space, h is the 1-particle Hamiltonian17 and F h is the corresponding symmetric Fock space, then the GNS represen~ associated with tation of the double wedge algebra A ~ acts on F h F h and the total Hamiltonian is expressed by ~ d h1 1d h; H 29

21

22

The 1-parameter group of automorphisms of A , which prescribes the quantum dynamical evolution, is given by t W W T t 23

for each t . The quantum dynamical system AR ; t portrays the dynamics in the right Rindler wedge R once time-evolution is specied by the timelike Killing vector eld a X =T a T =X a
a

24

where is some positive constant. The vector eld corresponds geometrically to the generator of the Lorentz boosts t . Its integral curves are the hyperbolas X 2 T 2 c 1 ; Y c2 ; Z c3 25

with asymptotes {X 7 T ; Y c2 ; Z c3 } (c1 ; c2 ; c3 are real constants) and represent the worldliness of uniformly accelerated observers in Minkowski spacetime. Clearly, an exactly analogous construction can be applied to the left Rindler wedge L provided that due notice is taken of the fact that the timelike Killing vector eld a is past-directed in L. The geometry of the double system is depicted in Fig. 1. The arrows indicate the direction of a ; O is a 2-dimensional surface on which a vanishes; is a Cauchy hypersurface corresponding to T 0; and hA , hB are two null hypersurfaces dened by T X 0, T X 0 respectively. The double quantum dynamical system can now be dened as follows. The tensor product algebra is the double wedge algebra
15 The tradition is traced back to Rindler (1969) who was the rst to propose to look at Minkowski spacetime from the point of view of an accelerating observer.

where d denotes the second quantization map for selfadjoint operators. The comparison of this formula with (2) is suggestive of the analogy with the standard spin example in ordinary quantum mechanics. Finally, Kay (1985c) showed that in the GNS representation of the double wedge algebra induced by a double KMS state ~ is the unique (up to phase, normalized) ~ , the vector
16 Let me interject here a comment on the way the concrete QFT model has been presented so far. One might be tempted to regard the implementation of terminology germane to the study of quantum elds on curved spacetimes (Cauchy surface, Killing vector eld, and the like) as superuous, given that what is discussed, after all, is the simplest quantum eld on at Minkowski spacetime. However, it is important for our purposes to view the right Rindler wedge R as a spacetime in its own right that allows the application of standard quantization procedures for the KleinGordon eldviz. as a globally hyperbolic static spacetime. I shall return to this point in Section 6 below in connection with whether the quantum eld in R may indeed be viewed as a component subsystem (part) of the compound double dynamical system (whole) enjoying intrinsic properties. 17 Note that h is the 1-particle Hamiltonian corresponding to the t-time ow prescribed by the future-directed timelike Killing vector eld a in R that generates the Lorentz boosts t . In short, the regular ground 1-particle structure mentioned here is the one that belongs to what is commonly referred to in the literature as Fulling or Rindler quantization.

A. Arageorgis / Studies in History and Philosophy of Modern Physics 44 (2013) 206 214

211

~ to the eigenvalue zero and eigenvector of the total Hamiltonian H the left- and right-wedge von Neumann algebras are type III1 in the classication of Connes. This implies that the spectrum of the ~ extends from to .18 And this fact modular Hamiltonian H constitutes an important mathematical detail for the conclusion to be drawn on the subject of holism.

~ ; ; ~ of AR is unitarily equivalent to that the representation H the GNS representation H ; ; of AR associated with . In ~ that confers the unimore detail, the isomorphism U : H -H tary equivalence, A U AU n ; AAR ; 32

is dened by extending 5. Holism and nonseparability in the quantum eld theoretic model To turn now to the issues of holism and nonseparability in QFT, I shall regard the double quantum dynamical system of the Klein Gordon eld on Minkowski spacetime as a compound system with component subsystems its restrictions to the two Rindler wedges, L and R. This presupposes that the quantum eld in each of the two wedges may appropriately be viewed as part of the whole. I shall say more about the tenability of this presupposition in the nal section of the paper. To substantiate physical property holism in the metaphysical sense, it sufces to pinpoint a qualitative intrinsic physical property of the double quantum dynamical system that does not supervene on qualitative intrinsic physical properties and relations of the two single-wedge quantum dynamical systems. The sought for property is the energy of the eld.19 ~ at some value 4 0 over the Fix any double KMS state ~ assigns the double quantum dynamical system. The state property zero energy to the compound system: in the GNS ~ AL AR induced representation of the double wedge algebra A ~ ~ , the cyclic vector representing ~ is the eigenvector of by ~ that corresponds to the eigenvalue 0. the total Hamiltonian H Are there intrinsic energy properties of the eld in each of the two wedges upon which the total energy supervenes? Specically, what is the energy of the eld in the right wedge R when ~ ? Take the restriction the double wedge system is in the state ~ to R, of ~ 1A; A AAR ;
R

~ U A A

33

~ . and thus satises U The natural question to ask now is whether the restriction of ~ to AR assigns a denite value to the part the double KMS state of the total Hamiltonian (29) contributed by R. In other words, is an eigenvector of U n 1d hU ? The answer is negative. For thanks to the aforementioned unitary equivalence of representations, the norm of the vector U n 1d hU in H should equal ~ . And the latter becomes ~ in H the norm of the vector 1d h innite! By utilizing the identities supplied by Denition 2, one formally arrives at the divergent expression
~ =2 ~ jj2 jjeH ~ jj2 jj1d h 1d h

~ ; eH 1d h ~ : 1d h

34

30

and consider the GNS representation H ; ; of A by . Since is a mixed state over AR , the representation is reducible. Further, it can be shown that it is unitarily equivalent to the ~ dened by representation of AR on H ~ : A A ~ 1A : A R - B H 31

The sketch of the proof proceeds as follows. First, one veries, by direct appeal to the denition of the representation of a C*algebra and to the structure of the tensor product of C*-algebras, ~ ; is a representation of AR . Second, one uses the fact that H ~ that is a cyclic vector for ~ AR (recall Denition 2 (i)) to show ~ ; ; ~ is a cyclic representation of AR with cyclic vector that H R ~ . Third, if ~ denotes the vector state over A dened by ~ , one shows by direct calculation with the aid of (30) and (31) R that ~ on A . It then follows by application of a wellknown theorem concerning cyclic representations of C*-algebras20
18 On the other hand, for the spectra of and d h one has sp 0; spd h (cf. Kay (1985c; p. 69)). The subclassication of type III factors by Connes appears in any advanced book on operator algebras and their application to physics. Still, a concise presentation of the denitions and facts necessary here can be found in Haag (1992; pp. 224225), Summers (2006; pp. 253-254) or Halvorson (2007; pp. 737740). 19 The energy of the eld is represented by the innitesimal generator of the unitary group implementing the dynamics once time-evolution is prescribed by the timelike Killing vector eld a (associated with the Lorentz boosts in the X -direction). The scare quotes are prompted by the fact that the spectrum of the total Hamiltonian is not bounded below. 20 See, e.g., Proposition 4.5.3 in Kadison & Ringrose (1997; p. 279).

In this sense, the part of the total Hamiltonian pertaining to the right wedge cannot even be expressed in the GNS representation of the right wedge algebra associated with the restriction to that algebra of a double KMS state. This is just another facet of the fact mentioned in Section 3 that the modular automorphism group ~ of MR ; cannot be implemented by inner automorphisms. A physicist might understand the situation thus.21 When the double quantum dynamical system is in a double KMS state and hence has zero energy, the expectation value and the uctuations of the energy for the wedge subsystems become innite. Indeed, the vector representing the KMS state over the right wedge R is not expected on physical grounds to be an eigenvector of the corresponding energy operator exactly because the energy in a Gibbs ensemble is not sharply dened. Moreover, the expression (34) is expected on physical grounds to be innite since it measures the energy uctuations of an innitely extended eld in a state of thermal equilibrium at nonzero temperature. The upshot is that the restriction of a double KMS state does not assign any denite energy property to the wedge quantum dynamical systems. Therefore, there are qualitative intrinsic properties of the compound system that do not supervene on qualitative intrinsic properties of its component subsystemsi.e., we do have an instantiation of physical property holism in the metaphysical sense. Of course, on the face of it, there is a signicant difference with the standard spin example in ordinary quantum mechanics. One might be inclined to say that the uncertainty as to the values of the energy in the subsystems is of classical as opposed to quantal origin. For it may be attributed to the statistical uctuations afforded by the thermal equilibrium nature of the KMS state over AR . And, theoretically, quantum and statistical uctuations should not be confounded. Quantum uctuations indicate the random deviation of the measured value of an observable from its expectation value in a quantum state due to the fact that the state in question is not an eigenstate of the observable; whereas statistical uctuations indicate the random deviation of the measured value of an observable due to the presence of stochastic inuences.
21 Cf. Haag et al. (1967; p. 224) as well as Haag (1992; p. 209) on the explanation of the divergence of total energy uctuations at the thermodynamic limit of a Bose system in a Gibbs (KMS) state and initially enclosed in box (i.e., a nite volume).

212

A. Arageorgis / Studies in History and Philosophy of Modern Physics 44 (2013) 206 214

Be this as it may, three considerations should give us pause and make us hesitant to endorse promptly the above line of thought. First, as it is well known, the KMS condition does not furnish an adequate characterization of thermal equilibrium for classical (commutative) systems. For no nontrivial 1-parameter group of automorphisms of an abelian algebra does admit faithful KMS states (Proposition 5.3.28 in Bratteli and Robinson (1981; p. 117)). Second, the uncertainty as to the values of uctuating observables in a thermal equilibrium state in classical statistical physics is typically attributed to ignorance as to the microstates of the members of the ensemble. But the ignorance interpretation of mixtures in quantum theory encounters severe difculties. The situation is much worse in relativistic QFT where the local von Neumann algebras are generically of type III and thus do not admit pure states. So it is impossible, in this context, to understand a local thermal state as prepared by mixing pure states (Clifton & Halvorson, 2001; p. 25). Finally, it is generally admitted that it is not as yet well understood what exactly is it that the horizon entropy does count in such models of QFT in curved spacetime or noninertial frames.22 So much for holism in the metaphysical sense. What about holism in the epistemological sense? In this vicinity of concerns, the relevant question to ask is this. Can two agents, call them Lisa and Rob, each having access only to one wedge, to L and to R respectively, infer the global properties of the compound system by using the resource basis (in Seevincks, 2004 terminology) available to them? I have no denitive answer to this question. Yet, the answer must be negative if we demand that every local operation performed by Rob on a global state in order to prepare an eigenstate of some observable should produce a state that is empirically distinguishable from the original one by Lisa. Indeed, consider any global state represented by a normalized vector ~ . And suppose that Rob wishes to turn this state into H R eigenstate of a nonzero projection MR . Since M is a type III von Neumann factor, there exists an operator W MR such that E WW*and W*W 1.23 So Rob can achieve the desired result by means of the local to R operation ; : ; W n :W 35

Indeed, , W* EW , W* WW*W 1. However, Lisa cannot empirically detect this change of state because for every AML , ; W n AW ; W n WA ; A 36

In short, the distinguishability of global states by local agents is delimited by the possibility afforded by type III factors to change any global state into an eigenstate of a projection via a local operation in a region without disturbing the expectation values of any observables in that regions spacelike complement. And it is such global states that determine the global (dynamical) properties of the system via the orthodox property ascription rule. Let us lastly turn to the issue of state nonseparability. To establish state nonseparability, one would need to nd two distinct states on the double quantum dynamical system whose restrictions to the component subsystems are identical. However, this is not the case for double KMS states on the double wedge ~ 1 and ~ 2 , at different algebra. Take any two double KMS states, values, 1 2 . Their restrictions, 1 and 2 , to the right-wedge quantum dynamical system will be KMS states at different temperatures. And given that the von Neumann algebras MR k
22 A tentative suggestion is this: for an observer conned to the outside of the horizon, the horizon entropy measures the number of ways that the world beyond the horizon can affect the world outside (see Jacobson & Parentani (2003)). 23 Cf. Yngvason (2005; pp. 139140) to whose paper I owe the idea of this application of the physics of type III factors to the exploration of the holistic nature, in the epistemological sense, of QFT.

(k 1; 2) are type III, 1 and 2 will be disjoint.24 This means that 1 and 2 cannot even be written as density operators in one anothers GNS representation and are, thus, highly nonidentical. Of course, this argument alone does not warrant the conclusion that the QFT model does not exhibit state nonseparability since it deals only with a particular class of states for the compound system, that of double KMS states. Still, the analogy with the standard spin example in ordinary quantum mechanics fails in this respect. In a sense, this is the price to be paid for dealing with systems having innite number of degrees of freedom. In the quantum theory of systems with nite number of degrees of freedom satisfying CCRs, the Stone-von Neumann theorem ensures uniqueness (up to unitary equivalence and multiplicity) of the representation of the algebra. But the quantum theory of systems with innite number of degrees of freedom encounters the representation problem: quantum systems with innite number of degrees of freedom admit many unitarily inequivalent irreducible Hilbert space representations of the same abstract algebraic structure and the choice of the appropriate representation (assuming that such a choice does have physical signicance) must be ne-tuned to the dynamics and the relevant macroscopic constraintsin our case, the temperature. However, some considerations may tone down the existing differences with the standard spin example in ordinary quantum mechanics and suggest a tempered epistemological version of state nonseparability. To begin with, given any double KMS state ~ , the question Is the eld in the right wedge R in the state ? is not decidable within R. For this to be the case there should exist ~ ~ a nontrivial projection operator P MR such that ; P 1. But, ~ is a separating vector for MR , this condition due to the fact that implies via an elementary argument that P 1. Of course, the restriction of ~ to R cannot be viewed as a maximal specication of the state of the eld in R. If it could, then the GNS representation of AR induced by would be unitarily equivalent up to multiplicity (i.e., quasi-equivalent) to the GNS representation of AR associated with a pure state over AR . But this is impossible. Let be any pure state over AR and H ; ; the associated GNS representation of AR . The assumption that is pure entails that AR is irreducible and hence that the von Neumann algebra AR is type I as AR BH .25 On the other hand, AR is isomorphic to MR and thus of type III. And quasi-equivalence of the two representations would demand,26 per impossible, that AR and AR be isomorphic. The preceding considerations suggest that when the double quantum dynamical system is in some double KMS state, an agent conned to the right Rindler wedge R, our Rob, cannot determine with certainty which state he owns. Nor can he retrieve knowledge of the eld in R from any complete knowledge of objects intrinsic to R. In addition, if Rob moves along the integral curves of the Killing vector eld a in R, then he cannot deploy any classical means to probe the state of the eld outside R since hA and hB constitute event horizons for his worldline: no event in L can be witnessed by such an observer in R (and vice versa). And the situation appears even more complex in the small. Thermal eld theories at different temperatures, although they must be viewed as disjoint from the spatiotemporally global point of view, they do not look that different from the spatiotemporally local point of view.27 Under physically plausible assumptions,

See Theorem 5.3.35 in Bratteli & Robinson (1981; p. 130). If a selfadjoint set S of bounded operators on a Hilbert space H is irreducible, then its commutant S consists of multiples of the identity operator (Proposition 2.3.8 in Bratteli & Robinson (1979; p. 47) and thus S BH. 26 See Theorem 2.4.26 in Bratteli & Robinson (1979; p. 80). 27 Here I use the qualication spatiotemporally local in order to refer specically to bounded regions of spacetime.
25

24

A. Arageorgis / Studies in History and Philosophy of Modern Physics 44 (2013) 206 214

213

Jkel (2004) has shown that given any KMS state 1 at inverse temperature 1 over the quasi-local algebra of a quantum eld and any bounded open region O, there exists a state that satises locally the KMS condition at any inverse temperature 2 1 in every appropriately slightly smaller than O region Oo but coincides with 1 in the spacelike complement of every appropri^ .28 In this sense, KMS states at ately slightly larger than O region O different inverse temperatures cannot be distinguished even as thermal states in the small.

6. Concluding remarks The entire preceding discussion has rested on a tacit assumption: in the context of eld theory, proper subregions of a spacetime region constitute subsystems with the subsystem relation corresponding to inclusion. Familiarity with the intricacies of nonlocality and entanglement in QFT casts doubt on the viability of such an assumption, despite the initial apparent plausibility it may enjoy. I shall not try here to justify this assumption in its full generality. I shall only offer some arguments to the effect that the restricted use I have made of it is warranted. More precisely, I shall argue that the double quantum dynamical system of the KleinGordon eld on Minkowski spacetime offers a model of QFT in which the concepts like part, whole or intrinsic property, featuring in philosophical accounts of holism, can be made sense of. So, may the left and the right dynamical systems of the specic model be regarded as component subsystems (parts) of the compound double dynamical system (whole) in a sense appropriate to the philosophical analysis of holism? A positive answer requires the substantiation of two points. First, the double dynamical system must be such that it may be understood as wholly composed of the left and the right dynamical systems during the entire process of dynamical evolution.29 And second, it should make sense to talk about intrinsic properties of the left and the right dynamical systems. It is not hard to see that the rst requirement is satised. At the classical level, both SL and SR are invariant under the (respective restrictions of the) 1-parameter dynamical symplectic group T t , t , and at the quantum level, both AL and AR are invariant under the (respective restrictions of the) corresponding 1-parameter ~ t , t . dynamical automorphism group In order to delve into the issue raised by the second requirement, one would ideally need a sharply dened distinction between extrinsic and intrinsic properties of an object. However, no philosophical elaboration of the distinction is free of objections. Let us just say, then, that the intrinsic properties of an object are those properties that the object can possess independently of the existence or state of any other contingent object. And since in the present context we are interested in dynamical (i.e., state-dependent) properties, the talk of intrinsic properties of the left and the right dynamical systems presupposes that independent state assignments be possible.
A state over a quantum dynamical system A; t in Minkowski spacetime is said to satisfy the local KMS condition at inverse temperature in some bounded spacetime region O if for every subregion Osub O whose closure is contained in the interior of O, there exists a sub 4 0 such that for all A; BAOsub , t AB Bt i A for jt j o sub . The result I glossed over in the text is Proposition 3.4 in Jkel (2004; p. 594). 29 The underlying intuition has been nessed admirably by Healey (1991) in the philosophical scrutiny of the notion of a basic physical part of an object. The gist can be stated thus: an object of the class Part is a basic physical part of an object of the class Whole subject to a process of the class Process just in case every member of Whole continues to be wholly composed of a xed set of members of Part during any member of Process.
28

This presupposition is also satised. Classically, the right Rindler wedge R may be viewed on its own right as a at globally hyperbolic static spacetime,30 often dubbed Rindler spacetime. Accordingly, R is deterministically benign in the sense that the initial value formulation for the KleinGordon equation is well posed. Therefore the classical linear dynamical system SR ; s; T t is a prima facie good candidate for the application of the standard quantization procedures. Of course, the Rindler spacetime R is not inextendible because it can be isometrically embedded as a proper subset of another spacetime. But the restriction of quantization procedures to elds inhabiting only inextendible spacetimes seems unreasonable: it would rule out models of signicant physical interest such as QFT on Schwarzschild spacetime. At the level of the quantum theory now, the right- and the leftR wedge von Neumann algebras, ML and M , associated with a double KMS state of the double quantum dynamical system are, as it was shown in Section 3, C*-independent. Therefore any state of R ML is compatible with any state of M and vice versa. There are no logical relations between assignments of states to the right and to the left quantum dynamical systemshence, between predications of intrinsic dynamical properties to each of those systems. Of course, as so often in QFT, things are not that simple. And R one aspect of the complications is that the algebras ML and M are type III. It has been argued by several authors that systems of QFT that exhibit this type of von Neumann algebras must be viewed as open systems. The issue has been elaborated upon by Clifton and Halvorson (2001). Roughly, Clifton and Halvorson showed that if the von Neumann algebra MO over an open (not necessarily bounded) region of Minkowski spacetime is type III, then no local operation in O can disentangle the state over MO from the state of the eld in regions spacelike separated from Othe environment of O. Applied to the model discussed here, the Clifton and Halvorson result says that there is no operation performable in R (i.e., no positive weak*-continuous bounded linear map Z MR satisfying 0Z 11) that would turn a state entangled across L R M ; M (i.e., by denition, a state that is not a weak*-limit of R convex combinations of product states on ML M ) into a disen31 tangled one. The main obstacle is that there exists no normal R L product state across ML ; M in mathematical jargon, M and 32 R M are not W*-independent in the product sense. Still, W*-independence in the product sense sounds too stringent a requirement to impose in this context. After all, in view of the generic fact that the local von Neumann algebras in relativistic QFT are type III, it would then be hard even to conceive of a nontrivial formulation of holism for relativistic quantum eld theoretic systems.

Acknowledgments An early version of the paper was read at the 1st European Philosophy of Science Conference (EPSA07, November 2007, Madrid, Spain). I would like to thank very much Professor Richard
The hypersurfaces of constant t are Cauchy and a is the Killing vector eld. R L R ML M is the von Neumann algebra generated by M and M , i.e., the R intersection of all von Neumann algebras containing ML and M . 32 This remark is likely to generate some confusion since it seems to run into R conict with the claim that ML and M are C*-independent [cf. also Eq. (15)]. The R situation has as follows (Summers, 1990; p. 213). If ML and M were W*R independent in the product sense, then ML M would be isomorphic to the von R L R Neumann tensor product ML M . But given that M and M are factors and R commutants of one another, ML M is the algebra of all bounded linear operators ~ , hence a type I algebra. Yet ML MR inherits the type III property from its on H R L R L R components. Therefore ML M and M M cannot be isomorphicM M is R L R strictly larger than ML M . So M , M are not W*-independent in the product sense despite their being C*-independent.
31 30

214

A. Arageorgis / Studies in History and Philosophy of Modern Physics 44 (2013) 206 214 Jkel, C. D. (2004). The relation between KMS states for different temperatures. Annales Henri Poincar, 5, 579606. Kadison, R. V., & Ringrose, J. R. (1997). Fundamentals of the theory of operator algebras. Volume I: Elementary theory. Volume II: Advanced theory. Providence, RI: The American Mathematical Society. Kay, B. S. (1985a). A uniqueness result for quasi-free KMS states. Helvetica Physica Acta, 58, 10171029. Kay, B. S. (1985b). Purication of KMS states. Helvetica Physica Acta, 58, 10301040. Kay, B. S. (1985c). The double-wedge algebra for quantum elds on Schwarzschild and Minkowski spacetimes. Communications in Mathematical Physics, 100, 5781. Maudlin, T. (1998). Part and whole in quantum mechanics. In: E. Castellani (Ed.), Interpreting bodies: classical and quantum objects in modern physics. Princeton: Princeton University Press pp. 4660. Narnhofer, H. (1994). Entropy density for relativistic quantum eld theory. Review of Mathematical Physics, 6, 11271145. Rdei, M., & Summers, S. J. (2007). Quantum probability theory. Studies in History and Philosophy of Modern Physics, 38, 390417. Redhead, M. (1995). More ado about nothing. Foundations of Physics, 25, 123137. Rindler, W. (1969). Essential relativity: Special, general, and cosmological. New York: Springer-Verlag. Seevinck, M. P. (2004). Holism, physical theories and quantum mechanics. Studies in History and Philosophy of Modern Physics, 35, 693712. Streater, R. F., & Wightman, A. S. (1964). PCT, spin and statistics, and all that. 4th printing. New York: Addison-Wesley. Summers, S. J. (1990). On the independence of local algebras in quantum eld theory. Reviews in Mathematical Physics, 2, 201247. Summers, S. J. (2006). TomitaTakesaki modular theory. In: J. -P. Franoise, G. L. Naber, & T. S. Tsun (Eds.), Encyclopedia of mathematical physics, Vol. 5. Amsterdam: Elsevier pp. 251257. Summers, S. J., & Werner, R. F. (1988). Maximal violation of Bells inequalities for algebras of observables in tangent spacetime regions. Annales de l' Institut Henri Poincar, 49, 215243. Summers, S. J., & Werner, R. F. (1995). On Bells inequalities and algebraic invariants. Letters in Mathematical Physics, 33, 321334. Wald, R. M. (1994). Quantum eld theory in curved spacetime and black hole thermodynamics. Chicago: The University of Chicago Press. Wayne, A. (2002). A nave view of the quantum eld. In: M. Kuhlmann, H. Lyre, & A. Wayne (Eds.), Ontological aspects of quantum eld theory. New Jersey: World Scientic pp. 127133. Yngvason, J. (2005). The role of type III factors in quantum eld theory. Reports on Mathematical Physics, 55, 135147.

Healey and Professor Miklos Rdei for the discussion and encouraging comments at that juncture. I also thank them for taking the time to read earlier drafts. Of course, Richard Healeys seminal work on holism and nonseparability has been all along an inspiration for this work. Lastly, I express my gratitude to two anonymous referees of this journal for reading carefully the manuscript and offering valuable suggestions for improvement. References
Bratteli, O., & Robinson, D. W. (1979). Operator algebras and quantum statistical mechanics I: C*- and W*-algebras, symmetry groups, decomposition of states. New York: Springer-Verlag. Bratteli, O., & Robinson, D. W. (1981). Operator algebras and quantum statistical mechanics II: Equilibrium states, models in quantum statistical mechanics. New York: Springer-Verlag. Buchholz, D., & Junglas, P. (1989). On the existence of equilibrium states in local quantum eld theory. Communications in Mathematical Physics, 121, 255270. Clifton, R., & Halvorson, H. (2001). Entanglement and open systems in algebraic quantum eld theory. Studies in the History and Philosophy of Modern Physics, 32, 131. Emch, G. G. (2007). Quantum statistical physics. In: J. Buttereld, & J. Earman (Eds.), Philosophy of physicsHandbook of the philosophy of science. Amsterdam: Elsevier pp. 10751182. Florig, M., & Summers, S. J. (1997). On the statistical independence of algebras of observables. Journal of Mathematical Physics, 38, 13181328. Haag, R. (1992). Local quantum physics: elds, particles, algebras. New York: Springer-Verlag. Haag, R., Hugenholtz, N. M., & Winnik, M. (1967). On the equilibrium states in quantum statistical mechanics. Communications in Mathematical Physics, 5, 215326. Halvorson, H. (2007). Algebraic quantum eld theory. In: J. Buttereld, & J. Earman (Eds.), Philosophy of physicsHandbook of the philosophy of science. Amsterdam: Elsevier pp. 731864. Healey, R. A. (1991). Holism and nonseparability. Journal of Philosophy, 88, 393421. Healey, R. A. (2008). Holism and nonseparability in physics. In: E. Zalta (Ed.), Stanford encyclopedia of philosophy. o plato.stanford.edu 4 . Jacobson, T., & Parentani, R. (2003). Horizon entropy. Foundations of Physics, 33, 323348.

Вам также может понравиться