## Valeri P. Frolov and Andrei Zelnikov

Print publication date: 2011

Print ISBN-13: 9780199692293

Published to Oxford Scholarship Online: January 2012

DOI: 10.1093/acprof:oso/9780199692293.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2016. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 04 December 2016

# (p.435) Appendix F Quantum Fields

Source:
Introduction to Black Hole Physics
Publisher:
Oxford University Press

(p.435) Appendix F

Quantum Fields

# F.1 Classical Oscillator

## F.1.1 Lagrangian formalism

The action for a classical oscillator of mass m and frequency ω is

$Display mathematics$
(F.1.1)

The variation of the action with the fixed end points gives the equation of motion

$Display mathematics$
(F.1.2)

A solution of this equation for the initial conditions q(0) = q 0, q̇(0) = q̇0 is

$Display mathematics$
(F.1.3)

In the comparison of the quantum and classical theory an important role is played by the action as a function of its end points. Suppose the initial point at the time t 1 is q 1 and the final point at the time t 2 is q 2. The value of the action calculated for the solution Eq. (F.1.3) obeying these boundary conditions is

$Display mathematics$
(F.1.4)

where Δt = t 2t 1.

## F.1.2 Hamiltonian formalism

Denote

$Display mathematics$
(F.1.5)

then the Hamiltonian of the system is

$Display mathematics$
(F.1.6)

(p.436) In this formalism {p, q} are considered as coordinates in a 2D phase space. For any two functions on the phase space, A = A(p, q) and B = B(p, q), one defines the Poisson brackets

$Display mathematics$
(F.1.7)

It is easy to check that

$Display mathematics$
(F.1.8)

The oscillator equations in the Hamiltonian form are

$Display mathematics$
(F.1.9)

This system is equivalent to Eq. (F.1.2) and its solution, with the initial data (p 0, q 0), is

$Display mathematics$
(F.1.10)

## F.1.3 Euclidean oscillator

In the quantum theory the Euclidean version of quantum oscillators is quite often used. In particular, this allows one to establish general relations between quantum observables in the thermal state and the Euclidean expectation values. To perform a transformation to the Euclidean space we use the following Wick's rotation of time t

$Display mathematics$
(F.1.11)

After making this analytical continuation one considers τ as a real parameter, or the Euclidean time. We define the action for the Euclidean oscillator as follows:

$Display mathematics$
(F.1.12)

The variation of this action with the fixed end points gives the equation

$Display mathematics$
(F.1.13)

A solution for the ‘initial’ data $q ( τ 0 ) = q 0 , d q d τ ( τ 0 ) = q ˙ 0$ is

$Display mathematics$
(F.1.14)

The Euclidean action as a function of end points is

$Display mathematics$
(F.1.15)

Here, Δτ = τ 2τ 1.

# (p.437) F.2 Quantum Oscillator

## F.2.1 Heisenberg picture

For a quantum oscillator, q and p are operators obeying the following canonical commutation relations

$Display mathematics$
(F.2.1)

The quantum Hamiltonian Ĥ is defined as

$Display mathematics$
(F.2.2)

In the Heisenberg picture q̂ and p̂ are time‐dependent operators and their evolution is determined by the following equations of motion:

$Display mathematics$
(F.2.3)

These relations imply that any operator Â that is a function of and obeys a similar equation

$Display mathematics$
(F.2.4)

Since Ĥ does not depend on time one can write a solution of this equation in the form

$Display mathematics$
(F.2.5)

It is convenient to rescale q and p

$Display mathematics$
(F.2.6)

Working with the new variables and is equivalent to taking the mass parameter m equal to 1. From now on we shall use the rescaled variables and omit the tilde. In what follows we shall also put the Planck constant ℏ equal to 1.

## F.2.2 Operators of creation and annihilation

Let us write the operators q̂(t) and p̂(t) in the form

$Display mathematics$
(F.2.7)

Since q̂ and p̂ are Hermitian operators the operator â† is the Hermitian conjugated operator to â. Using Eq. (F.1.10) and Eq. (F.2.5) one gets

$Display mathematics$
(F.2.8)

Here, constant operators â and â† are the corresponding initial data. These operators obey the following commutation relations:

$Display mathematics$
(F.2.9)

(p.438) They are known as the operator of creation, â †, and the operator of annihilation, â. The Hamiltonian Ĥ of the quantum oscillator, written in terms of the operators of creation and annihilation, takes the form

$Display mathematics$
(F.2.10)

In the occupation number representation the basis in the Hilbert space of states is

$Display mathematics$
(F.2.11)

This basis is orthonormal

$Display mathematics$
(F.2.12)

One also has

$Display mathematics$
(F.2.13)

The state ǀ0⟩ has the lowest possible energy ω/2. It is called a ground state or a vacuum.

## F.2.3 Thermal quantum oscillator

Let Θ be a temperature. We denote by Θ = 1/Θ the inverse temperature of the system. A thermal state in quantum mechanics is described by a thermal density matrix

$Display mathematics$
(F.2.14)

The partition function Z(β) is determined by the normalization condition

$Display mathematics$
(F.2.15)

that is by the following expression

$Display mathematics$
(F.2.16)

The partition function for the quantum oscillator can be easily calculated using the fact that ǀn⟩ are eigenvectors of the Hamiltonian Ĥ with the eigenvalue equal to (n + 1/2)ω. One has

$Display mathematics$
(F.2.17)

The partition function determines the free energy F defined by the relation

$Display mathematics$
(F.2.18)

All the thermodynamical characteristics of the canonical ensemble can be obtained from its free energy. For a thermal quantum oscillator the free energy is

$Display mathematics$
(F.2.19)

(p.439) The temperature ‐independent term ω/2 in the free energy is a result of the presence of the vacuum energy ω/2 term in the expression for the Hamiltonian Eq. (F.2.10). We can rewrite Ĥ in the form

$Display mathematics$
(F.2.20)

We use the following standard notation for the thermal average of an operator Â

$Display mathematics$
(F.2.21)

The expression for the thermal average can also be written in the form

$Display mathematics$
(F.2.22)

## F.2.4 Green functions

There is a variety of different Green functions or two‐point correlators, which proved to be useful in quantum physics. Usually, they have a form of the average value of bilinear combinations of q̂(t) and q̂(t′). They differ by the choice of the state and the concrete form of the bilinear form, which may involve, for example, special ordering of the operators. We focus now our attention on the following thermal correlators

$Display mathematics$
(F.2.23)

Let us calculate $G β + ( t , t ′ )$ One has

$Display mathematics$
(F.2.24)

Using Eq. (F.2.7), Eqs. (F.2.8), and (F.2.13) one obtains

$Display mathematics$
(F.2.25)

It is easy to check that

$Display mathematics$
(F.2.26)

Using Eq. (F.2.25) and Eq. (F.2.26) one obtains

$Display mathematics$
(F.2.27)

(p.440) In the zero‐temperature limit, when β → ∞, the second term in the right‐hand side vanishes. Thus1

$Display mathematics$
(F.2.28)

From the representation Eq. (F.2.27) it is evident that $G β + ( t , t ′ )$depends only on the time difference tt′ and, hence, it is a function of only one variable. We denote this function by the same symbol $G β + ( t )$ so that one has

$Display mathematics$
(F.2.29)

This property is generic for any two‐time thermal correlator ⟨Â(t)B̂(t′)⟩β. In order to prove this we use the property that for any bounded operators Â and B̂ one has

$Display mathematics$
(F.2.30)

Let us substitute the following expressions

$Display mathematics$
(F.2.31)

into the average ⟨Â(t)B̂(t′)⟩β. Then, we apply Eq. (F.2.30) and the property that e βĤ commutes with Ût and $U ^ t − 1$ As a result of simple manipulations we obtain

$Display mathematics$
(F.2.32)

One can use the expression Eq. (F.2.29) for the functions $G β + ( t )$ and $G β − ( t )$ to define them in the complex plane of the variable t. Then, it is easy to check that the following relation is valid

$Display mathematics$
(F.2.33)

This result is known as the Kubo—Martin—Schwinger relation, or briefly the KMS relation. Similar KMS relations are valid for any thermal two‐point correlator

$Display mathematics$
(F.2.34)

To prove this result we use relations Eq. (F.2.31) for a complex time t + to obtain

$Display mathematics$
(F.2.35)

(p.441) so that

$Display mathematics$
(F.2.36)

Starting with $G β ± ( t )$ functions one can construct other Green functions for the quantum oscillator in the thermal state. In particular, for the commutator [q̂(t), q̂(t′)] one has

$Display mathematics$
(F.2.37)

This correlator in fact does not depend on the temperature, and it coincides with the vacuum expectation value. The reason is evident: The commutator [q̂(t), q̂(t′)] is a c‐number. A similar thermal Green function with the anticommutator {…,…} being substituted for the commutator is known as the Hadamard function2

$Display mathematics$
(F.2.38)

Another Green function that plays an important role in the quantum theory is the causal Green function or the Feynman propagator. It is defined as follows

$Display mathematics$
(F.2.39)

where the T‐product of the operators q̂(t) and q ̂(0) is defined as

$Display mathematics$
(F.2.40)

Here, θ(t) is a Heaviside step function. The definition Eq. (F.2.39) implies that

$Display mathematics$
(F.2.41)

This function is continuous at t = 0, but its derivative has a jump. 𝔖f,β can be defined as a solution of the following equation

$Display mathematics$
(F.2.42)

It obeys the following boundary conditions: At zero temperature this function propagates positive frequency modes to the future and negative frequency modes to the past.

Equation (F.2.35) can be used to determine the evolution of the operators in the purely imaginary or Euclidean time

$Display mathematics$
(F.2.43)

(p.442) The following object is known as the Matsubara propagator

$Display mathematics$
(F.2.44)

The operator T τ is the Euclidean time ordering

$Display mathematics$
(F.2.45)

One has

$Display mathematics$
(F.2.46)

For τ ϵ [0,β] the Matsubara propagator Δ(τ) coincides with $G β + ( − i τ )$ and is of the form

$Display mathematics$
(F.2.47)

One can easily check that

$Display mathematics$
(F.2.48)

In other words, the Matsubara propagator is periodic in the imaginary time with the period β. It can also be uniquely defined as the Green function for the Euclidean oscillator obeying the differential equation on a circle of length β

$Display mathematics$
(F.2.49)

Here, δ β (τ) is a delta‐function defined on a circle with the period β

$Display mathematics$
(F.2.50)

# F.3 Quantum Field in Flat Spacetime

## F.3.1 Classical scalar field

As an example of the field theory we consider a scalar massive field ϕ in a flat (3 + 1)‐ dimensional spacetime. The action is

$Display mathematics$
(F.3.1)

Here, t = T and X are Cartesian coordinates, and m is the mass of the field. The variation of the action gives the field equation. For the scalar field it is the Klein‐Gordon equation

$Display mathematics$
(F.3.2)

(p.443) It is convenient to consider a system in a box with the sizes {L 1, L 2,L 3} and to impose the following boundary conditions

$Display mathematics$
(F.3.3)

Momentum π conjugated to the field ϕ is

$Display mathematics$
(F.3.4)

The Hamiltonian of the system is

$Display mathematics$
(F.3.5)

Denote

$Display mathematics$
(F.3.6)

where n i are positive integer numbers. Then, a complete set of eigenfunctions of the Laplace operator Δ with the imposed boundary conditions is formed by functions

$Display mathematics$
(F.3.7)

where V = L 1 L 2 L 3 is the volume of the box. These functions obey the equation

$Display mathematics$
(F.3.8)

and the normalization condition

$Display mathematics$
(F.3.9)

A general solution of Eq. (F.3.2) can be written as

$Display mathematics$
(F.3.10)

where the amplitudes of the field q k (t) satisfy the equation

$Display mathematics$
(F.3.11)

Thus, the amplitude of the field for a given wave number k obeys the oscillator equation. The reduction of the field theory Eq. (F.3.1) to the infinite set of oscillators Eq. (F.3.11) enumerated by k can be made more transparent by substituting the field decomposition in harmonics Eq. (F.3.10) into the Lagrangian Eq. (F.3.1). This substitution gives the following result

$Display mathematics$
(F.3.12)

## (p.444) F.3.2 Quantum field

To quantize the field ϕ it is sufficient to quantize a decoupled set of oscillators described by the Lagrangian Eq. (F.3.12). To do this one considers the amplitude q k and a conjugated momenta p k = q̇k as Hermitian operators obeying the canonical commutation relations

$Display mathematics$

Using the completeness of the orthonormal basis Eq. (F.3.7) it is possible to show that these relations are equivalent to following canonical commutation relations for the field variables

$Display mathematics$
(F.3.13)

One may describe the quantum field as an infinite set of oscillators. But now each oscillator is ‘living’ not in the ‘physical space’, but in a ‘space of amplitudes’. Spatial modes Φ k describe an amplitude of the probability for a given mode to be in the vicinity of the space point X. The quantum nature of the field is connected with discrete levels of its quantum amplitudes. Since the free quantum field is equivalent to a set of decoupled oscillators, one can easily apply all previous results concerning the quantum oscillator to such a set. In particular, the operators of creation and annihilation for each mode k are related to the amplitudes q̂k and their momenta q̂k = $p ^ k = q ^ ˙ k$as follows

$Display mathematics$
(F.3.14)

where

$Display mathematics$
(F.3.15)

The operators a ̂k and $a ^ k †$obey the commutation relations

$Display mathematics$
(F.3.16)

The Hamiltonian Eq. (F.3.5) written in terms of the operators of creation and annihilation takes the form

$Display mathematics$
(F.3.17)

A vacuum is a state ǀ0⟩ with the lowest possible energy. It is defined by the condition

$Display mathematics$
(F.3.18)

The quantity E 0 is the energy of the vacuum zero‐point fluctuations. The vacuum energy E 0 is unobservable. However, in the presence of an external field and/or for non‐trivial boundary conditions the vacuum energy $E 0 ′$ differs from its value E 0 in the empty Minkowski spacetime. The difference $E 0 ′ − E 0$ can be measured and hence it has a well‐defined physical meaning. This phenomenon is known as the Casimir effect.

(p.445) Denote by n k a non‐negative integer number for a mode k and by {n k} a set of such integer numbers. We assume that only a finite number of them do not vanish. In the occupation number representation the basis in the Hilbert space of states is

$Display mathematics$
(F.3.19)

This basis is orthonormal

$Display mathematics$
(F.3.20)

## F.3.3 Thermal fields

A thermal state of the quantum field with the inverse temperature β is a state when each of its modes is in a thermal state described by the density matrix Eq. (F.2.14). The corresponding density matrix for the quantum field is

$Display mathematics$
(F.3.21)

Here, Ĥ is a complete Hamiltonian Eq. (F.3.17) for the quantum field and Ĥ0 is its version with zero‐point energy subtracted. The trace operator is the trace over the complete Hilbert space of the field states. Using the representation of the field as a set of decoupled oscillators, one obtains

$Display mathematics$
(F.3.22)

This is a free energy of the thermal quantum field at the temperature Θ = β −1.

## F.3.4 Continuous spectrum

The imposed boundary conditions, that is vanishing of the field at the boundary, are somehow artificial. However, if the quantization box is large and one is studying observables in a domain far away from the boundary, the result does not depend on the particular boundary conditions. They affect the state of the field only close to the boundary. Usually, for large volume the surface effects can be neglected.

In order to exclude effects connected with the spatial location of the boundary it is convenient to use the other boundary conditions, known as the periodicity conditions. Let us again take a cube with boundaries at ǀX iǀ = L i/2 and impose the following periodic boundary conditions

$Display mathematics$
(F.3.23)

In other words, the opposite boundaries of the cube are identified with one another and we have a space that is 3‐torus 𝕋3.

(p.446) The complete set of eigenfunctions of the Laplace operator with given periodicity conditions is well known. For each of three orthogonal directions one can use sin(k i X i) and cos(k i X i) functions. The periodicity conditions imply that

$Display mathematics$
(F.3.24)

where n i are non‐negative integer numbers. The ‘doubling’ of the basic functions is ‘compensated’ by the fact that the corresponding wavelengths k i are twice ‘rare’ than before. Very often it is more convenient instead of real eigenfunctions of the Laplace operator to use the complex eigenfunctions exp(±ik i X i). These functions also obey the periodicity conditions when Eq. (F.3.24) is satisfied. We use the following solutions

$Display mathematics$
(F.3.25)

Here, $k = ( 2 π n 1 L 1 , 2 π n 2 L 2 , 2 π n 3 L 3 )$ where n i are the integer numbers, which can be both positive and negative. Under this condition the real basic functions sin(k i X i) and cos(k i X i) can be obtained as linear combinations of solutions Φ k and their complex conjugated. The basis functions obey the following normalization conditions

$Display mathematics$
(F.3.26)

A mode decomposition of the field operator ϕ̂(t, X) now takes the form

$Display mathematics$
(F.3.27)

Making the size of the quantization box infinitely large one reduces the distance between the nearby wavelength, so that finally in the infinite‐size limit one obtains a continuous spectrum of the Laplace operator.

The continuous spectrum formulas can be obtained from the discrete ones by the following substitution

$Display mathematics$
(F.3.28)

where $A = V / ( 2 π ) 3$ The discrete and continuous δ‐functions are defined as usual to fulfill the relations ∑k δ k,k′= 1 andd k δ(kk′) = 1.

The continuous spectrum basis functions are

$Display mathematics$
(F.3.29)

(p.447) They obey the normalization condition

$Display mathematics$
(F.3.30)

The field representation in the limit of an infinite box is

$Display mathematics$
(F.3.31)

where the operators of creation and annihilation with continuous index k obey the commutation relations

$Display mathematics$
(F.3.32)

## F.3.5 Green functions

A Green function for a quantum field φ̂ is an average value of a bilinear combination of φ̂(t,X) and φ̂(t′,X′). Again, as in the case of a quantum oscillator, there exists a variety of Green functions that differ by the choice of a state used for averaging and by a concrete form of the bilinear form, which may include special operator ordering.

Let us consider the following thermal correlators

$Display mathematics$
(F.3.33)

where

$Display mathematics$
(F.3.34)

These functions are known as thermal Wightman functions. Since the spacetime is static one has

$Display mathematics$
(F.3.35)

This property can be proved by using the relations

$Display mathematics$
(F.3.36)

and the fact that cyclic permutations of operators within the Tr ‐operation do not change the result.

The property Eq. (F.3.35) means that $G β +$ and $G β -$ functions depend only on the difference tt′. Moreover, since the system is invariant under spatial translations and rotations, these fuctions depend only on r = ǀXX′ǀ. Thus, we have

$Display mathematics$
(F.3.37)

(p.448) To calculate these quantities it is sufficient to substitute Eq. (F.3.31) in their definition Eq. (F.3.33), and to use the relations

$Display mathematics$
(F.3.38)

The calculations give

$Display mathematics$
(F.3.39)

Here, $G β , ω ± ( t )$ is the corresponding Green function for a thermal oscillator with the frequency co given by Eq. (F.2.29). Similar representations are valid for the other Green functions of the quantum field.

Let G, G f,β, and $G β ( 1 )$ be the thermal average of the commutator, the Feynman propagator and the Hadamard function for the quantum field (Takagi 1986; Birrell and Davies 1982), respectively

$Display mathematics$
(F.3.40)

Here,

$Display mathematics$
(F.3.41)

To obtain any of these functions it is sufficient to substitute in the relation similar to Eq. (F.3.39) a corresponding Green function for the oscillator

$Display mathematics$
(F.3.42)

Here, the • ‐symbol specifies a concrete choice of the Green function.

The thermal oscillatory Green functions consists of two parts. The first, which coincides with the vacuum average, differs for different Green functions, while the second (which vanishes at the zero‐temperature limit and may be absent in some cases) is a universal expression

$Display mathematics$
(F.3.43)

In a special case of a massless field, m = 0, the expressions for the Green functions can be written in terms of the elementary functions. One has ω = k, and the integral in Eq. (F.3.42) can be easily calculated. For B β (t, r) related to ℬ(t) by Eq. (F.3.42) one finds

$Display mathematics$
(F.3.44)
(p.449)

Fig. F.1 Euclidean space for the thermal field. It has the topology of the cylinder S 1 ×ℝ3. The directions of the Euclidean time τ and spatial dimensions are schematically shown.

Using this result and the following relation

$Display mathematics$
(F.3.45)

where ϵ 〉 0 is an infinitesimal quantity and Ƥ is the principal value, one obtains the real‐time thermal Green functions for massless fields

$Display mathematics$
(F.3.46)

For imaginary time t = − the Wightman functions are equal, $G β + ( − i τ , r ) = G β − ( − i τ , r )$ and the Matsubara propagator is

$Display mathematics$
(F.3.47)

It is periodic in the imaginary time with the period β

$Display mathematics$
(F.3.48)

and obeys the following equation

$Display mathematics$
(F.3.49)

For r = 0

$Display mathematics$
(F.3.50)

Thermal Green functions for massive fields and fields of other spins can be obtained in a similar manner. In all cases the general property remains valid: The quantum field at the finite (p.450) temperature α 1 is connected by analytical continuation to the Euclidean field theory on the Euclidean space with the topology of the cylinder S 1 × ℝ3, where the length of S 1 is β.

## F.3.6 Measurement of temperature

Let us discuss now how the temperature of the radiation can be measured. In a general setup, a thermometer is a system with internal degrees of freedom that can interact with thermal radiation. We assume that this interaction is weak and does not disturb the state of the radiation. We expect that the internal degrees of freedom of the thermometer will be excited as a result of an interaction and, after some time, they will be distributed thermally. By ‘measuring’ observables depending on this distribution one can ‘measure’ temperature of the radiation. Usually, it is assumed that the thermometer has small size. This is important when the thermal radiation is affected by an external field. For example, in the presence of a static gravitational field the local temperature depends on a point. In order to measure this local temperature the size of the thermometer must be much smaller than a characteristic scale at which the temperature changes.

For illustration let us consider a simple model. We choose a quantum oscillator as a thermometer and consider its interaction with a scalar massless field ϕ. Let the ‘thermometer’ be located at a point X 0 of a flat spacetime. The system, that is the quantum field ϕ and the oscillator q, is described by the following action

$Display mathematics$
(F.3.51)

Here, S 0[ϕ] is the field action Eq. (F.3.1), S 0[q] is the oscillator action Eq. (F.1.1). The interaction action is

$Display mathematics$
(F.3.52)

where ϕ(t) = ϕ(t, X 0). We denote by Ĥ the Hamiltonian of the complete system. It has the form

$Display mathematics$
(F.3.53)

In the Schrodinger picture the evolution of the complete system is described by the state vector ǀΨ(t)} obeying the Schrödinger equation

$Display mathematics$
(F.3.54)

In this picture the operators do not evolve, so that

$Display mathematics$
(F.3.55)

Let us write

$Display mathematics$
(F.3.56)

(p.451) Substitution of Eq. (F.3.56) into Eq. (F.3.54) gives

$Display mathematics$
(F.3.57)

Denote by ǀA⟩ the vectors of the orthonormal basis in the Hilbert space of the states of the system without interaction, and A is an index enumerating the vectors of this set. We additionally assume that ǀA⟩ are eigenstates of the free Hamiltonian

$Display mathematics$
(F.3.58)

Let us write

$Display mathematics$
(F.3.59)

Substituting this relation into Eq. (F.3.57) and multiplying the obtained relation by ⟨Bǀ we get

$Display mathematics$
(F.3.60)

We assume that the interaction constant λ is small and use the perturbation theory to solve the Eq. (F.3.60). In the first‐order approximation, the amplitude of the probability of the transition from the initial state ǀA⟩ to the final state ǀB⟩ at time t is

$Display mathematics$
(F.3.61)

The probability of the transition AB is

$Display mathematics$
(F.3.62)

Till now the consideration was quite general and we did not use specific properties of the interaction. Basically, the relation Eq. (F.3.62) reproduces the standard quantum mechanical result of the perturbation theory for the probability of transitions under a general perturbation V̂. In order to apply this result to our system of a quantum oscillator interacting with the thermal radiation we proceed as follows. First, note that a state ǀA⟩ of the complete system can be written as

$Display mathematics$
(F.3.63)

where ǀn⟩ is a complete set of the states of a non-interacting quantum oscillator and ǀN⟩ is a complete set of the field states. Accordingly, we write the index A = {n, N}. For the index B we write B = {m,M}. We choose the states ǀn) and ǀN⟩ to be the eigenstates of the corresponding free Hamiltonian

$Display mathematics$
(F.3.64)

(p.452) In this basis one has

$Display mathematics$
(F.3.65)

We used here that $〈 m | q ^ ( t ) | n 〉 = e i ( ω m − ω n ) t q m n$. These matrix elements can be found by using the expression Eq. (F.2.7)

$Display mathematics$
(F.3.66)

where ω is the frequency of the oscillator. One has

$Display mathematics$
(F.3.67)

The relations Eq. (F.3.65) and Eq. (F.3.67) show that in the lowest order of the perturbation theory the transitions are only of two types:

• either the oscillator absorbs a single ‘photon’ of energy ω and ‘jumps’ from the lower level n to the upper level n + 1;

• or it emits a ‘photon’ of the frequency ω; and ‘jumps’ from the upper level n + 1 to the lower level n.

To obtain the probability for the transition w mn one needs to average over unobserved final states of the field ǀM⟩ and to use the thermal density matrix for the initial state of the field

$Display mathematics$
(F.3.68)

Here, P N is a probability of a state ǀN⟩. As a result of these operations one gets

$Display mathematics$
(F.3.69)

Here,

$Display mathematics$
(F.3.70)

is the positive‐frequency Wightman function

$Display mathematics$
(F.3.71)

It is easy to show that a double integral in Eq. (F.3.69) can be rewritten as follows

$Display mathematics$
(F.3.72)

(p.453) This relation shows that w mn is proportional to time. Thus, for the transision probability per unit time one has

$Display mathematics$
(F.3.73)

where

$Display mathematics$
(F.3.74)

Let us denote ẇn+ = ẇn+1,n and ẇn = ẇn,n+1. Equation (F.3.67) shows that

$Display mathematics$
(F.3.75)

Thus,

$Display mathematics$
(F.3.76)

Simple calculations give

$Display mathematics$
(F.3.77)

for either positive or negative ω. Thus,

$Display mathematics$
(F.3.78)

The obtained relation implies that the ratio of the probability of the transition nn + 1 to the probability of the inverse process n + 1 → n is a universal function. It does not depend on the details of interaction and it is determined only by the temperature of the radiation β −1 and the energy difference of the corresponding levels. This ratio is determined only by the energy spectrum of the Wightman function G̃(ω) for the radiation. This result is quite general and it allows one to prove that the equilibrium state of the quantum oscillator interacting with the thermal bath with temperature β −1 is described by the thermal density matrix with the same temperature.

This can be easily shown as follows. Denote by p n(t) the probability of the oscillator to be at the level n. Then, the change in time of this probability is

$Display mathematics$
(F.3.79)

For n = 0 one has

$Display mathematics$
(F.3.80)

In the equilibrium state ṗn = 0. The relation Eq. (F.3.80) gives

$Display mathematics$
(F.3.81)

(p.454) Starting with this relation and using Eq. (F.3.79) we obtain

$Display mathematics$
(F.3.82)

The normalization condition $∑ n = 0 ∞ p n = 1$determines p 0

$Display mathematics$
(F.3.83)

Thus, the equilibrium distribution of the oscillator over the energy levels has a thermal law. The details of its interaction with the thermal bath, and especially the value of the coupling constant λ determine the rate of transitions ẇn±, and, in particular, how fast the equilibrium is reached. The smaller λ the longer one needs to wait when the oscillator reaches its equilibrium thermal state. In other words, if the interaction is weak the time required to ‘measure’ the temperature of the thermal radiation is large.

# F.4 Quantum Theory in (1+1)‐Spacetime

## F.4.1 Field equations

Till now we have considered the case when a quantum field is freely propagating in a flat spacetime, so that the mode functions are the standard plane waves. Let us consider a slightly more complicated situation when the scalar field moves in a space with a potential. For simplicity we assume that the space has only one dimension and denote a spatial coordinate by x. The action for the field ϕ(t, x) is

$Display mathematics$
(F.4.1)

Here, U(x) is the potential of the form shown in Figure F.2. It is positive and falls rapidly enough at ǀxǀ → ∞. One can also consider this action as a generalization of the (1 + 1)‐ dimensional version of Eq. (F.3.1) to the case when the mass of the field depends on x.

The field equation is

$Display mathematics$
(F.4.2)

Fig. F.2 Potential U(x). At far distances ǀxǀ → 0 it falls rapidly enough.

(p.455) The field can be decomposed into modes

$Display mathematics$
(F.4.3)

where u k are complex solutions of the following eigenvalue problem

$Display mathematics$
(F.4.4)

The latter equation is simply a standard Schrödinger equation of quantum mechanics

$Display mathematics$
(F.4.5)

It describes a one‐dimensional motion of a particle of the energy E = k 2 in the presence of the potential barrier. This problem is discussed in standard books on quantum mechanics. Here, we just collect the main results that are required for our analysis.

## F.4.2 Bases in the solution space

The spectrum of the problem Eq. (F.4.5) is continuous and the energy E is positive. The energy levels are degenerate, namely there exist two linear independent solutions for a given energy E. It is convenient to consider complex solutions of Eq. (F.4.5). The scalar product for any two complex solutions f 1 and f 2 is defined as follows

$Display mathematics$
(F.4.6)

A complex solution is uniquely specified by its asymptotic behavior at x = ±∞. In the general case, for a given k 〉 0 one has

$Display mathematics$
(F.4.7)

For the continious spectrum the scalar product can be written in terms of the scattering coefficients α k, β k,γ k and δ k. Since a solution of the original wave equation Eq. (F.4.2) is obtained by multiplying f k(x) by e ikt, it is evident that a term with e ikx describes a mode propagating to the left, while a mode with e ikx moves to the right. This can be presented by the following conformal diagram (see Figure F.3).

One can interpret the solution Eq. (F.4.7) as follows. An initial wave consists of two parts: One, with the amplitude β k, is going from x = −∞ to the right, and the other, with the amplitude α k, is propagating from x = 0∞ to the left. After the scattering on the potential U(x) the right‐moving mode will have the amplitude δ k at x = ∞, while the left‐moving mode will have the amplitude γ k at x = −∞. Since the mode is uniquely specified by its initial data α k and β k, the ‘future’ asymptotic data, γ k and δ k, are functions of the initial data. These functions depend on the form of the potential U(x). For any two solutions Eq. (F.4.7) one has

$Display mathematics$
(F.4.8)
(p.456)

Fig. F.3 Conformal diagram for a general solution f. A shadowed region schematically shows a domain where the scattering on the potential takes place.

We shall use the normalization condition

$Display mathematics$
(F.4.9)

A standard convenient choice of the complete set of normalized complex solutions consists of functions u in and u up. These functions are specified by the following scattering data

$Display mathematics$
(F.4.10)
$Display mathematics$
(F.4.11)

In these solutions we assume that k 〉 0. The conformal diagrams for these functions are shown in Figure F.4.

The normalization condition Eq. (F.4.9) results in

$Display mathematics$
(F.4.12)

These solutions obey the following relations

$Display mathematics$
(F.4.13)

It is well known (and easy to demonstrate) that the Wronskian

$Display mathematics$
(F.4.14)

of any two linearly independent solutions ν 1 and ν 2 to Eq. (F.4.5) is a constant. By calculating the Wronskian at both infinities, x = ±∞, for solutions Eqs. (F.4.10) and (F.4.11) and their (p.457)

Fig. F.4 Conformal diagram for the solutions u in and u up.

complex conjugate, one obtains relations between the coefficients that enter the solutions. In particular, these conditions reproduce Eq. (F.4.12). Besides this they give two additional relations
$Display mathematics$
(F.4.15)

A solution u up,k describes a stationary wave propagating from x = −∞. It is partially scattered back and partially passes through the potential barrier. The coefficients r k and t k are the reflection and transmission coefficients, respectively. Similarly, u in,k(x) describes a stationary wave propagating from x = ∞, and R k and T k are its reflection and transmission coefficients. Equation (F.4.15) shows that the transmission amplitudes T k and t k are the same, while for the reflection amplitudes one has ǀR kǀ = ǀr kǀ, while their phases can be different.

Since the potential in Eq. (F.4.4) is real, one can always choose basic solutions to be real. For each of k 〉 0 one has two real solutions that we denote ν 1,k and ν 2,k. The conformal diagram for these solutions are shown in Figure F.5. These solutions are special versions of a general solution Eq. (F.4.7). For $v 1 , k α k = δ k * = C + , k$and $γ k = β k * = C − , k$. Similarly, for $v 2 , k α k = δ k * = D + , k$and $γ k = β k * = D − , k$The normalization and orthogonality conditions for ν 1,k and ν 2,k give

$Display mathematics$
(F.4.16)

The coefficients C +,k and D +,k are uniquely determined by the coefficient C −,k and D k. The corresponding relations include the reflection and transmission coefficient. One can use the following asymptotic data, specifying the real solutions ν 1,k and ν 2,k:

$Display mathematics$
(F.4.17)
(p.458)

Fig. F.5 Conformal diagram for a real solutions ν 1 and ν 2.

Here,$z k = 1 + | t k |$

Basic solutions u in,k and u up,k can be written as linear combinations of ν 1,k and ν 2,k

$Display mathematics$
(F.4.18)

where

$Display mathematics$
(F.4.19)

Denote

$Display mathematics$
(F.4.20)

It is evident that this function is real and symmetric with respect to its arguments x and x

$Display mathematics$
(F.4.21)

Using Eqs. (F.4.18) and (F.4.19) one can check that

$Display mathematics$
(F.4.22)

At far distances, where the potential U(x) becomes small, the function V k has the following asymptotic form

$Display mathematics$
(F.4.23)

In this expression … denote omitted terms that either decrease at infinity or contain the fast‐ oscillating factor ∼exp[±ik(x + x′)]. The latter does not contribute to the integral over k.

## (p.459) F.4.3 Quantization

To quantize the scalar field ϕ one can write it in the following two equivalent forms

$Display mathematics$
(F.4.24)

In the first line, the operators q̂1,k(t) and q̂2,k(t) are standard oscillatory‐amplitude operators. In the second and third lines, $a ^ k † , b ^ k † , a ^ k$ and b̂k are the operators of creation and annihilation obeying the relations

$Display mathematics$
(F.4.25)

All other independent commutators vanish. The operator $a ^ k †$ creates left‐moving quanta in the mode u in,k, while $b ^ k †$ creates right‐moving quanta in the mode u up,k. The vacuum state is a state when neither left‐moving nor right‐moving quanta are present. It is defined by the conditions

$Display mathematics$
(F.4.26)

The Hamiltonian of the system (with zero‐point fluctuation contribution excluded) is

$Display mathematics$
(F.4.27)

## F.4.4 Equilibrium thermal state

Thermal equilibrium state with temperature Θ = β −1 is described by the thermal density matrix

$Display mathematics$
(F.4.28)

The thermal Wightman functions are defined as

$Display mathematics$
(F.4.29)

Using Eq. (F.4.24) one obtains

$Display mathematics$
(F.4.30)

(p.460) Similar expressions are valid for all other Green function

$Display mathematics$
(F.4.31)

Here, the •‐symbol specifies the corresponding Green function.

In particular, the Matsubara propagator is

$Display mathematics$
(F.4.32)

This propagator is periodic in τ with the period β and obeys an equation

$Display mathematics$
(F.4.33)

The Matsubara propagator can be obtained as a Green function on the Euclidean cylinder S 1 ×ℝ1 in the presence of the potential U and with the size of the compact dimension equal to β (see Figure F.6).

## F.4.5 Local observables

As an application of the developed formalism, let us calculate the asymptotic value of the thermal averages of the stress‐energy tensor for the massless field ϕ̂. To obtain the temperature-dependent part of T μν we shall use the renormalized Hadamard function

$Display mathematics$

In the asymptotic region, where the potential U(x) vanishes, the stress‐energy tensor can be written as follows

$Display mathematics$
(F.4.34)

Fig. F.6 Euclidean space for the (1 + 1)‐thermal field. It has the topology of the cylinder S1 x R1. The direction of the Euclidean time r and the spatial dimension x are schematically shown. A shadowed region schematically shows a domain where the scattering on the potential takes place.

(p.461) Simple calculations give

$Display mathematics$
(F.4.35)

Thus, outside the potential for the thermal state one has p = ϵ. This result is in agreement with the expected one.

## F.4.6 Non‐equilibrium thermal state

The density matrix Eq. (F.4.28) for the thermal equilibrium state can be written as a product of two independent density matrices

$Display mathematics$
(F.4.36)

Here, Ĥin and Ĥup are the Hamiltonians for in‐ and up‐quanta given by Eq. (F.4.27). In the thermal equilibrium state the in‐ and up‐quanta are emitted with the same temperature β −1 from x = oo and x = − ∞, respectively. Since Ĥin and H ̂up are independent, one can consider a more general state, which is described by the following density matrix

$Display mathematics$
(F.4.37)

For β inβ up this state is stationary, but not inequilibrium. In such a state there exists a net flux of thermal radiation from the hotter source to the colder one. We call such a state a non‐equilibrium thermal state.

The renormalized Hadamard function for this state is

$Display mathematics$
(F.4.38)

and D up is obtained from D in by changing in → up. In these relations {c.c.} means a complex conjugated expression, and

$Display mathematics$
(F.4.39)

To calculate the renormalized stress‐energy tensor at x →∞ we use the following asymp-totics3

$Display mathematics$
(F.4.40)

Here,… denote the terms that either vanish at infinity or rapidly oscillate there, ~ exp[ik(x + x′)]. These terms will not contribute to the asymptotic value of the T μν. Denote ν = t + x and u = tx, then, (p.462)

$Display mathematics$