6. CPT reversal l
Précis. On the representation view, there may be weak and strong arrows of time. A weak arrow exists. A strong arrow might too, but not in current physics due to CPT symmetry.
There are many ways to reverse time besides the time reversal operator T. Writing P for spatial reflection or ‘parity’, and C for matter-antimatter exchange or ‘charge conjugation’, we find that PT, CT, and CPT are all time reversing transformations, when they exist. More generally, for any non-time-reversing (unitary) transformation
U, we find that UT is time reversing. Are any of these transformations relevant for the arrow of time? Feynman (1949) captured many of our imaginations with the proposal that to understand the direction of time, we must actually consider the exchange of matter and antimatter as well.1 Writing about it years later, he said:
“A backwards-moving electron when viewed with time moving forwards appears the same as an ordinary electron, except it’s attracted to normal electrons — we say it has positive charge. For this reason it’s called a ‘positron’. The positron is a sister to the electron, and it is an example of an ‘anti-particle’. This phenomenon is quite general. Every particle in Nature has an amplitude to move backwards in time, and therefore has an anti-particle.”(Feynman 1985, p.98)
Philosophers Arntzenius and Greaves (2009, p.584) have defended Feynman’s proposal,
1In his Nobel prize speech, Feynman (1972, pg.163) attributes this idea to Wheeler in the development of their absorber theory (Wheeler and Feynman 1945). Stueckelberg (1942) had arrived earlier at a similar perspective.
151 interpreting it as the claim that “the operation that ought to be called ‘time reversal’ — in the sense that it bears the right relation to spatiotemporal structure to deserve that name — is the operation that is usually called TC”. However, a more common physics adage today is perhaps the one expressed by Wallace, that,
“in quantum field theory, it is the transformation called CPT, and not the one usually called T, that deserves the name”. (Wallace 2011b, p.4)
The intuition behind this is that in the standard model of particle physics, only CPT is guaranteed to be a symmetry. CPT symmetry is moreover deeply connected to what it means to be a relativistic quantum field theory of any kind, by a collection of results known as the CPT Theorem. It is also connected to the spin statistics connection, that half-integer spin systems obey Fermi-Dirac statistics, while integer spin systems obey Bose-Einstein statistics.2 This chapter will evaluate these ‘alternative’ arrows of time, and argue in par- ticular that CPT symmetry does have implications for time’s arrow. How important it is depends on how one interprets two different perspective on time’s arrow. First: experimental evidence shows that T symmetry is violated, which establishes one sense in which time is asymmetric. I call this the weak arrow of time. However, there is stronger sense of time asymmetry, which is to say that no time reversing symmetry is available at all: neither T, nor any non-time-reversing transformation composed with it. I call this a strong arrow of time. One’s philosophical commitments might bring one to prefer one sense of an arrow over the other; however, I will argue that consideration of both provides insight into the structure of time. In particular, I will argue for a perspective on matter and antimatter according to which their exchange is just a representation of a particular symmetry of a spacetime structure, the universal covering group of the Poincaré group. This provides a clearer physical sense of how CPT can be interpreted as essentially a spacetime transformation. I begin Section 6.1 with a simple pictorial discussion of these arrows and their philosophical underpinnings. I then give a formal argument that a strong arrow is
2A classic reference is Streater and Wightman (1964); see Bain (2016) for a philosophical review.
152 incompatible with the ordinary unitary dynamics of quantum theory: some ‘time reversing’ operator is always a symmetry. This argument stops short of giving a complete physical interpretation of the operator. Thus, I next turn to considering which time reversing symmetries might have physical significance. Section 6.2 reviews Feynman’s proposal, that time reversal also exchanges matter and antimatter, and its defence by Arntzenius and Greaves. Unfortunately, experimental evidence for discrete symmetry violation makes trouble for this view, but that it can be saved through consideration of the CPT theorem. Section 6.3 then gives an account of the physical significance of CPT symmetry, as an automorphism of a spacetime structure on the representation view. I claim this proposal solves some of the puzzles regarding the CPT theorem highlighted by Greaves (2010). Section 6.4 then briefly reviews the statement of the CPT theorem, correcting in particular the great misnomer that it is a ‘composition’ of three operators, and arguing that it prohibits a strong arrow. Section 6.5 finally discusses the possibility of how one might yet establish a strong arrow of time, in theories that go beyond the standard model.
6.1 Weak and strong arrows
6.1.1 Philosophical underpinning
Our main aim in this chapter is to clarify status of the following two notions of an arrow of time:
1. a weak arrow occurs when no time reversing symmetry exists that can be inter- preted as reversing ‘time alone’;
2. a strong arrow occurs when no time reversing symmetry exists at all.
To illustrate, consider the 2 2 chequered grid of Figure 6.1, with a vertical t-axis ⇥ representing time, and a horizontal x-axis representing space. This description is invariant under PT symmetry, but not T symmetry. In particular, the reversal of ‘time alone’ t t is not a symmetry: so, there is a weak arrow. However, the combined 7!
153 reversal of both time and space t, x t, x is a symmetry: so, there is no strong ( )7!( ) arrow.
t t t
x x
Figure 6.1: Time reversal is not a symmetry, but space-time reversal is.
What could motivate interest in the weak arrow of time, as opposed to the strong? The answer depends on one’s philosophical commitments regarding the role of time within the broader spacetime structure of the world. Here are two perspectives that arrive at different results.
• Temporalism: any symmetry that transforms time translations as t t establishes 7! a symmetry of time. The temporalist holds that in order to understand the sym- metries of time, only time is needed, without regard for any other degrees of freedom. So, since PT symmetry reverses time translations in the sense that
PT : t t, it establishes a time symmetry, nevermind that it transforms space 7! as well.
• Spacetime Contextualism: only a symmetry of time translations t t that does 7! not affect any time-independent degrees of freedom establishes a symmetry of time. The spacetime contextualist is concerned with time translations in their broader con- text as a spacetime symmetries. So, although PT symmetry may be an important spacetime symmetry, it does not really belong to the class of temporal symmetries, because it also transforms space.
On first glance, one might be tempted by temporalism: we are after all ultimately interested in the symmetries of time translations. What does it matter if this symmetry is established by T,PT or CPT? Once we know t t is a symmetry of time translations, 7!
154 we may conclude that time does not have an arrow. For the temporalist, only the failure of all time reversing symmetries would be good evidence for an arrow of time: the only interesting arrow is a strong arrow.
However, the spacetime contextualist has a response that I find convincing. The temporalist view is motivated by a ‘purist’ attitude about time, that only the time translations matter. However, transformations like PT and CPT cannot be defined at all without appeal to non-temporal degrees of freedom. So, by helping one’s self to the availability of PT, CT, or CPT, one has already accepted some amount of spacetime contextualism, by accepting that other degrees of freedom are relevant to time’s struc- ture. In that case, it seems disingenuous to ignore that other degrees of freedom are transformed. The very availability of transformations like PT, CT, and CPT is reason to think that neither transformation can be appropriately described as ‘time reversal’. As a result, existing evidence of T-violation establishes an interesting arrow, though a weak one.
Although temporalism and spacetime contextualism are helpful for framing this discussion, I do not see that a great deal rides on the distinction. A middle ground is simply to accept that there are two interesting notions of an arrow, weak and strong. I will soon show a sense in which a strong arrow is incompatible with the dynamics of quantum theory. For this we will need to introduce language that is a little more precise.
6.1.2 Representations of weak and strong arrows
Let me begin by taking stock of the previous chapters. One of the main themes of this book is that time reversal is not a transformation of spacetime coordinates, but of time translations t t. These can in general be implemented by a time reversal 7! transformation ⌧ with the property that ⌧t⌧ 1 ⇤ t. Now, on what I have called the Representation View, a dynamical theory is just a representation of spacetime transformations amongst the automorphisms of a state space. And, when we look at the representatives of time translations in quantum theory, we find a set of unitary
155 operators generating solutions to the Schrödinger equation, while time reversal Ut ⌧ is uniquely represented by Wigner’s (antiunitary) time reversal operator T. Time reversal symmetry violation, which says that a representation of T does not exist, can then be viewed as evidence for an arrow of time.3 It is in particular a weak arrow, in the language just introduced. Let me first illustrate the weak and strong arrows using a simple physical exam- ple, before diving into relativistic quantum field theory. Consider a bead constrained to a string, which we find always has positive velocity to the East (Figure 6.2). A representation of time translations maps each position-velocity pair x, x , always sat- ( €) isfying x > 0, to the pair ' x, x ⇤ x + xt, x . Such a description is obviously time € t( €) ( € €) asymmetric, and one can confirm this formally: since time reversal transforms states as T x, x ⇤ x, x , it follows that no time reversal operator exists in a state space with ( €) ( €) only positive velocities x > 0. Parity symmetry fails as well, since for the same reason, € the transformation P x, x ⇤ x, x does not exist either. ( €) ( €)
Figure 6.2: Time translations for the asymmetric bead-string system.
However, there is a non-standard representation of time reversal given by the combined PT reversal transformation, which I write, ⇡ x, x ⇤ x, x . This ⇡ cannot ( €) ( €) be written as a composition of P and T, neither of which exist. But it does define
1 a representation of time reversal symmetry, in that ⇡'t ⇡ ⇤ ' t, as can be easily checked.4 Therefore, PT is a dynamical symmetry of the system. This system has a weak arrow, but no strong arrow. But, since PT symmetry implies that t t is a symmetry of time translations, the advocate of temporalism 7! may conclude that time itself is symmetric. In contrast, the spacetime contextualist will resist this conclusion: this is only superficially a symmetry of time, more properly viewed as a symmetry of spacetime. And after all, if a system that always moves in
3Chapter 2 presented this picture of time reversal, Chapter 3 derived Wigner’s T as the unique representative of t t, Chapter 4 defined time reversal symmetry and T-violation, and Chapter 5 7! argued that its T-violation is associated with an arrow of time. 4 1 For all x, x we have, ⇡'t ⇡ x, x ⇤ ⇡'t x, x ⇤ ⇡ x + xt, x ⇤ x xt, x ⇤ ' t x, x . ( €) ( €) ( €) ( € €) ( € €) ( €)
156 one direction is not time symmetric, then what is? On this view, the PT transformation
⇡ does provide a representation of time reversal symmetry, but it is not the right one. Again, although I am sympathetic to spacetime contextualism, I do not have a strong view on this debate. What I would like to point out is that if this system fails to have a strong arrow of time, then it is hard to imagine one that cannot. This turns out to be because it is impossible — at least in quantum theory as we know it.
6.1.3 Generic reversibility and physical significance
In quantum theory, one standardly assumes that the time translations have the structure of R, + , the group of real numbers under addition. The state space as a projective ( ) Hilbert space, consisting of the equivalence classes of Hilbert space states related by a complex unit. A representation of time translations amongst the automorphisms of state space is then a strongly continuousf one parameter unitary representation t . 7! Ut A unitary representation by no means assures that the time reversal operator T is a symmetry, just as it is not assured in classical mechanics. For example, consider a unitary representation ⇤ e itH with, say, H ⇤ 1 P2 + PQ, interpreted in terms of Ut 2m position Q and momentum P. Then since time reversal changes the sign of P but not
1 1 Q, we find that THT , H, and consequently T tT , U t. But the time reversal U group element ⌧ by definition has the property that ⌧t⌧ 1 ⇤ t, and so ' ⇤ T is only a ⌧ 1 representation of it if T tT ⇤ t. This property is not satisfied here: thus, a system U U with this dynamics would be T-violating. However, here again the combination of parity and time reversal does turn out to be a symmetry.
Whether or not the time reversal operator T gives rise to a symmetry, it is always possible to construct some time reversing (antiunitary) operator that does, in that T 1 t ⇤ t. This might be PT, or it might be something else entirely. That fact is TU T U assured by the following.5
5This simple idea, that conjugation with respect to the unitary dynamics implements a time reversing symmetry, is similar in spirit to many proofs of the CPT theorem; see below, and for Swanson (2019) for a discussion in the context of algebraic proofs of CPT symmetry.
157 Proposition 6.1. Every strongly continuous unitary representation t of R, + extends 7! Ut ( ) to a representation of time reversal symmetry, in that there exists some antiunitary satisfying T 1 t ⇤ t. TU T U
Proof. Let H be the self-adjoint operator such that ⇤ e itH, assured to exist by Ut Stone’s theorem, with spectrum ⇤ R. Let H be its spectral representation on ✓ s L2 ⇤ , in that H x ⇤ x for all in its domain, and VHV 1 ⇤ H for some unitary ( ) s ( ) s V : L2 ⇤ (cf. Blank, Exner, and Havlíček 2008, §5.8). If K is the conjugation H! ( ) operator on L2 ⇤ , in that K ⇤ for all L2 ⇤ , then K, H ⇤ 0, since for all in ( ) ⇤ 2 ( ) [ s] the domain of H we have KH K 1 x ⇤ x x ⇤ H x . Thus, :⇤ V 1KV is the s s ( ) ( ) s ( ) T desired antiunitary operator, since our definitions imply that , H ⇤ 0, and hence [T ] 1 itH 1 it H 1 itH t ⇤ eT( )T ⇤ e T T ⇤ e ⇤ t. ⌅ TU T U
This provides a compelling argument against the possibility of a strong arrow of time: in any quantum theory with a unitary dynamics, including the entire standard model, some representation of time reversal symmetry is always possible, and so a strong arrow does not exist. The temporalist in particular will find there is no interesting arrow of time. The only way to have a truly strong arrow of time is to go beyond standard quantum theory, for example to contexts in which gravitational effects are incorporated. I will discuss this possibility briefly in Section 6.5.
However, if the standard time reversal operator T is known, then this generic antiunitary has the form ⇤ UT for some unitary U, whose physical significance T T is left uninterpreted.6 This leaves open a concern about whether it represents any transformation of physical significance. On the representation view, the meaning of a symmetry U arises as in a representation of some symmetry structure, like a group of spacetime symmetries. No such group is specified in the construction above, which might make this transformation unconvincing as a physical symmetry. At the very least, it would test the resilience of the temporalist’s commitment to accepting any time reversing symmetry whatsoever. A more convincing time reversing symmetry of quantum theory would be
6Just define U :⇤ T 1, and recall that the composition of two antiunitaries is always unitary. T
158 CPT symmetry. To make this case, we will need to say something about where this transformation gets its meaning, and what sort of operator it represents. This is the task of Section 6.3. But let me first review a charming alternative due to Feynman, that time reversal is ‘really’ the CT transformation in disguise.
6.2 The Feynman view on time reversal
6.2.1 Reversing time and charge
Richard Feynman argued that “positrons can be represented as electrons with proper time reversed relative to true time” (Feynman 1949, p.753). More generally, his view is that what we usually call ‘time reversal’ is, surprisingly, best understood as including charge conjugation. Feynman’s view made a scattering of remarkable appearances in the early 1950’s. It was advocated by Satosi Watanabe (1951), and then by Gerhart Lüders, who adopted it in his famous paper giving the first argument for a CPT theorem:
“In the following, two types of time reversal will appear: the time rever- sal ‘of the first kind’ which, loosely speaking, consists in a reversal of the ‘motion’ of all particles, and the time reversal ‘of the second kind’, a simul- taneous performance of a proper time reversal and a particle-antiparticle conjugation.” (Lüders 1954, p.4)
This interpretation was dropped in later sharpenings of the CPT theorem by Pauli (1955), Lüders (1957) and Jost (1957). However, Bell formulated a version of the CPT theorem that adopted a similar perspective:
“The reversal in time referred to above is that which arises naturally when the theory is approached in its field aspect; it is found to involve reversing the signs of electromagnetic charges. It is to be distinguished therefore from the ‘Wigner type’ time reversal, which perhaps arises more naturally
159 when the particle apsect is emphasized and which is defined as leaving the charges unaltered” (Bell 1955, p.479).
Feynman even appears to have smuggled his proposal into the modern theory of Feynman diagrams, which represent contributions to a scattering amplitude in a perturbative approximation. When the CT transformation exists, reflecting a Feynman diagram (Figure 6.3) about the vertical ‘time’ axis produces a CT transformation, which both reverses time and exchanges matter and antimatter!7
+ + µ µ e e