You are on page 1of 12

Probability Density in Relativistic Quantum Mechanics

and Quantum Field Theory


Subtitled: Whence the Relativistic Wave Solution Normalization Factors?

1 Relativistic Free Scalar Field Equation


For a free scalar field governed by the (relativistic) Klein-Gordon equation

(1)

and its complex conjugate (transpose)

(2)

,
we have solutions

(3)

and

(4)

.
The summation is from infinite k in the negative x direction to infinite k in the positive x direction plus similar

summations for the y and z directions. The reasons for the normalization factors

in the solutions will be explained herein later on.


Solutions (3) and (4) arise for problems with boundary conditions, as such problems entail discrete values for wave
number (equivalently, wavelength or 3-momentum k.) Each term in the summation has a discrete value of 3-
momentum k.
There are also continuous solutions, comparable to (3) and (4), but containing integrals over 3-momentum, rather
than summations. These are Fourier integrals (representing wave packets) rather than Fourier series. They are
generally used for cases without boundary conditions.
Note that only one term in the discrete summation solution can be used in problems where the particle in question
can be approximated by a single value for 3 momentum (and thus a pure complex sinusoid extending to infinity.)
While we shall only address the discrete solutions here, the reader should keep in mind that a parallel development
exists for continuous solutions.

2 Relativistic Quantum Mechanics vs Quantum Field Theory


2.1 Difference Between Solutions in RQM and QFT

In relativistic quantum mechanics (RQM), (x) of (3) [with b†(k) = 0 for all k] represents a single particle general
state that is a sum of discrete momentum eigenstates of that single particle. The coefficients a(k) are numbers,
amplitudes which as we will see, when squared, equal the probability of finding the single particle in that discrete
eigenstate.
In quantum field theory (QFT), on the other hand, the coefficients a(k) and a†(k) in (3) and (4) are not numbers but
operators that each destroy or create single particle eigenstates.
Until we note otherwise near the end, when we wish to extrapolate our results to QFT, we shall restrict the
development herein to RQM (and thus make it easier to understand).

2.2 Definition of Eigensolutions


As noted, the solution (3) in RQM is for single particle general (sum of eigenstates) state, and does not contain
operators. Thus, from henceforth while discussing RQM, we substitute the numerical symbol Ak for a(k), as the
latter symbol is normally used in QFT to represent an operator. Hence, in RQM, the K-G solutions are of form

(5)

where k has unit norm. That is, we define the eigenstate with symbol k as

(6)
,
so that

(7)

,
or more generally,

(8)

.
3 Probability Density
3.1 Review of Non-Relativistic QM Probability Density
In non-relativistic quantum mechanics (NRQM), we encountered 1) the wave function solution to the Schroedinger
equation Ψ, and 2) the particle probability density ρ Ψ*Ψ (or equivalently for a scalar quantity, Ψ†Ψ) We review
here the derivation of that relation for probability density.
The general procedure:
Use the governing wave equation to deduce another equation having the form of the continuity equation

(9)
,
and we will then know that ρ, whatever it turns out to be, must represent a conserved quantity. Its integral over all
space is constant in time. If we normalize ρ such that when integrated over all space, the result equals one, we can
conjecture that ρ is the particle probability density (which when integrated over all space equals the probability that
we will find the particle somewhere in all space, i.e., one.) Then throughout time, as our particle evolves, moves,
and rearranges its probability density distribution, the total probability of finding it somewhere in space is always
one. It turns out, from experiment, that our conjecture that this quantity ρ in NRQM equals probability density is
true.
Using the Schroedinger Equation:
First, pre-multiply the Shroedinger equation by the complex conjugate of the wave function

(10)

.
Then, post-multiply the complex conjugate of the Schroedinger equation by the wave function

(11)

,

where V is real so V=V . Adding (10) to (11), we get

(12)
or

(13)

.
This is the same as the continuity equation (9) if we take

(14)

as our probability density and

(15)

as our probability current. This is how the commonly used relation (14) is found.
3.2 Probabilty Density in RQM
Using the Klein-Gordon Equation
For RQM, we start with the Klein-Gordon field equation (1) rather than Schroedinger equation. First pre-multiply

it by , i.e.,

(16)

Then take the complex conjugate Klein-Gordon equation post multiplied by , i.e.,

(17)
,
and subtract the former (16) from the latter (17). The result (after multiplying by i) and using the identity shown
under (13) is

(18)

.
This has the form of the continuity equation

(19)

,
where we introduce 4D vector notation. The probability density for a Klein-Gordon particle is then

(20)

,
with the probability current

(21)

.
Importantly, and perhaps surprisingly, the relativistic form of the probability density (20) is not the same as (14), the
non-relativistic probability density.
Probability for Discrete Solutions
For a single particle state in RQM, the probability density (20) in terms of the solution (5) is

(22)
When we integrate over the volume V, all terms with k/ ≠ k go to zero, leaving

(23)

and thus |Ak|2 is the probability of measuring the kth eigenstate.


Reason for Normalization Factors

Note that obtaining (23) is the reason for the normalization factors 1/ used in the solution
. Those factors result in a total probability of one for a single particle and |Ak|2 as the probability for measuring
the kth state. That is, the form of the relativistic field equation gave us the form of the probability density
in (20) and (22). The time derivatives in (20) gave us a factor of ωk, and the two terms a factor of 2. These cancel
in (23)with the 2ωk in the denominators. The V term in the denominator cancels in the integration over volume
in (23) and the result is a total probability of 1.
Difference from NRQM
Note that

(24)

,
because in RQM (unlike in NRQM), the LHS of (24) does not represent the integral of the probability density over
space.
Relativistic Invariance of Probability
This probability value of unity in (23) is a relativistic invariant. If we change our frame, the energy spectrum (i.e.,
the ωk values) will change (K.E. looks different for each energy-momentum eigenstate). But these changes cancel
out in the probability calculation and always result in a probability of one for any frame. Further, the Ak here are
constants that do not vary with frame, so the probability of finding any particular state is also independent of what
frame the measurements are taken in.

3.3 Spin and Spin 1 Probability Densities

For spin (Dirac spinor) and spin 1 (vector) particles, one proceeds in similar fashion to deduce the continuity
equation from the corresponding wave equation.
Dirac particles

Pre-multiply the Dirac equation by

(25)
and add to the complex conjugate transpose Dirac equation post-multiplied by ψ

(26)

to yield

(27)

Thus, the probability density for a relativistic spin particle is

(28)

,
with the probability current readily deduced from (27). The four current is then

(29)

Vector Particles
For spin 1 massless (massive) particles, one uses the Maxwell (Proça) wave equation and proceeds in a manner
similar to that used above for Klein-Gordon particles, except that now our pre and post multiplications must be inner
products with the photon wave solution Aα. It is a good exercise to work this out to find the result

(30)

with 4 current

(31)
.
All this works if we consider the photon solution having the form of (5).
Caveat (Subtler, Advanced Refinement of Understanding)
One finds later, however, that any particle which is its own antiparticle, such as the photon, has the

property (or for photons, ), as the action of taking the complex


conjugate effectively switches charge (if you haven’t studied this yet, don’t worry about how right now). For this to
be true, the wave equation solution must have the form

(32)

where C is a real number, rather than the more general form of (3).
Since the photon is its own anti-particle, that means (30) equals zero. This led early researchers to conclude that
probability density as calculated herein is actually charge density (or at least proportional to charge density), as the
photon has zero charge.
However, as long as we confine ourselves to RQM, where, like NRQM, we can consider there to be only particles
and no antiparticles, we can take the photon to have the form of (5), i.e., (32) without the second term on the RHS.
For this case, (30) can be considered the photon probability density.
Subtleties such as this, as well as a number of others, are best handled with QFT, rather than RQM.

4 Probability Density in Quantum Field Theory


QFT differs from RQM in that the coefficients a(k), b(k), and their complex conjugates in
solutions (3) and (4) turn out to be operators that create and destroy particles and antiparticles, rather than numeric
quantities. The particles and antiparticles are then represented as kets, such as | , while the solutions

and are operator fields (often simply referred to as fields.)
QFT also is far more amenable to multiparticle states than NRQM or RQM, and such states are represented with ket
forms such as |3 k, Ak’α (denoting 3 scalars all with momentum k and one photon with momentumk/.)
QFT also does not typically deal with probability densities, as that is something more suitable for a single particle
state. For two electrons, for example, the total probability of finding an electron somewhere in space would be two.
For this and other reasons, QFT deals primarily with number of particles, rather than probability density for a given
particle.
However, for purposes of edification, clarification, and ready comparison of QFT to RQM and NRQM, we will
herein illustrate the QFT equivalent of probability density.
Probability Density Operator in QFT
One starts with the Klein-Gordon equation, its complex conjugate and the solutions (3) and (4). The same
steps (16) to (20) are then followed, except that 1) we are not using the solution form (5) because we retain the terms
with b(k) and b†(k), and 2) the final result will be a density operator, not a simple numeric density. We will then
need to find the expectation value of that operator for a particular particle (ket) in order to get a numerical
probability density.
Proceeding as before in steps (16) to (20), we end up with a relation similar to (20)
(33)

,
which is now designated as an operator because the fields on the RHS of (33) are themselves operators. When we
use the solution form of (3) and (4) in (33), we end up with (after some algebra and certain subtleties[1])

(34)

.
The operator products on the RHS above are found in the development of QFT to equal number operators. That is,

(35)

This means that the particle number operator operating on a ket (single or multi-particle) yields the number of
particles times the original ket, e.g.,

(36)

.
The antiparticle number operator behaves similarly for antiparticle kets.
Thus (34) becomes

(37)

,
and for (37) acting on a single particle state | k’ , we find all the number operators except the one with

k’
/
momentum k yield zero. Defining < k’| we then find the expectation value for the density is

(38)

Since expectation value is, by its very nature, probabilistic, and so, by its nature, is probability density, (38) is
essentially the probability density itself, i.e.,

(39)

The value of ρ = 1/V in (38) is what we expect (recall from NRQM) the probability density to be for a pure complex
sinusoid, i.e., constant over all space. Integration of (38) over the entire volume in which the particle is contained
yields one, the total probability of finding the particle.
Antiparticle Probability Density
Note from (37) and (38) that antiparticle states would have negative probability! This was another tip to early
researchers that the density they were dealing with was more readily related to charge. Thus, one may be starting to
see that the concept of probability density and the mathematics associated with it, seem to lose some applicability in
QFT. Particle number, however, takes on significance, as now antiparticles would simply be designated as having
negative particle numbers.
Caution in Evaluating Expectation Values of Density Operators
Some care must be taken in the evaluation of expectation values similar to that of (38). The Dirac bracket,
expressed in wave mechanics, is an integration over space. But for operators with a spatial dependence such
asρoper often has (and which is typical of charge, mass or any type of density), the spatial dependence in the operator
is not included in the integration. That is,

(40)
We carry out the integration over the dummy spatial variable of the states, but not over the spatial variable of the
density operator. In the special case of (38), ρoper does not have spatial dependence, so this is not important, though
in other cases to be met later in QFT, it will be good to keep this clearly in mind.
“Norms” and Streamlining Notation in QFT
However, we can streamline our thinking by utilizing a vector space interpretation of the bras and kets, rather than
the wave interpretation. In this, we simply consider the Dirac bracket of a state to be an inner product of that state
with itself and equal to unity. This is the common approach in QFT.
And since, as we have seen, the inner product of a state with itself is no longer the same as the probability (at least in
relativistic theories), QFT calls that inner product the norm of the state, i.e.,

(41)

.
Multiparticle States in QFT
Norms for multiparticle states are taken by definition as equal to unity. Thus,

(42)

.
Any inner product of a state (multi or single particle) with a different state is zero. Thus,

(43)

.
For probability density, we have

(44)
This yields a total probability of measuring any of the particles inside the volume in which they are contained of 4.
Probabilities greater than one make little sense, and so for the multiparticle QFT, one deals instead almost
exclusively with particle numbers.

Return to Contents List (Home page)

[1]
There is a second way to get another ρ and j that satisfy the continuity equation. Do this by taking the Klein-

Gordon equation and post multiply it by , then subtract that from the complex conjugate field
equation pre multiplied by . If you add this expression for ρ to the one found the original way and divide by two
(to get the average), you will have (34).

You might also like