The virial theorem

The virial theorem relates the expectation kinetic energy of a quantum system to the potential. That is
of theoretical interest, as well as important for computational methods like “density functional theory.”

Consider a quantum system in a state of definite energy  . In other words, consider a quantum system
in a stationary state. It does not have to be the ground state. The quantum system will be assumed to be
in infinite space.

To keep it simple, for now assume that there is a single particle with position vector   in a poten-
tial  . That covers our previous examples of the harmonic oscillator and the hydrogen atom.

Then the virial theorem relates the expectation kinetic energy   to the potential   as follows:

(A.75)

(Recall that nabla,  , is just the multi-dimensional derivative  .) The above formula can be very
useful.

For example, consider the harmonic oscillator. There

so in Cartesian coordinates

Then according to the virial theorem      . So the expectation kinetic energy and the ex-
pectation potential energy are the same. Compute whichever is easiest, or just take half of the total en-
ergy   if you know it.

Also consider the hydrogen atom. There

so in polar coordinates

Then according to the virial theorem the expectation potential energy is minus twice the expectation ki-
netic energy. And their sum, the total energy  , is then minus the expectation kinetic energy. In
short,   and   with   negative.

The virial theorem does not apply to the particle in a pipe, as that particle is in a bounded space. (You
can assume infinite space if you take the potential infinite outside the pipe, but obviously by itself that
does not help much. You could assume infinite space with a potential

if you then take the limit   to get infinite potential outside the pipe and zero inside. That gives
the correct but trivial result that all the energy is kinetic.)

But the virial theorem does apply to any number of particles, not just to one. Just sum over all the parti-
cles:

where i is the particle number.

For example, consider the hydrogen molecule, where there are four particles, two protons and two elec-
trons. Here
where   is   if particle   is a proton and   if it is an electron. Like for the simple hydrogen atom,

so the total expectation potential energy of the system is still twice the total energy   and the total ki-
netic energy is still minus  . And this continues to hold for much bigger systems of nuclei and elec-
trons, which is why it is of interest for computational methods.

In some computations you might need to assume that the electrons are in a state of definite energy, like
in the ground state, but the nuclei are not. In such computations the nuclei are at an assumed position
and you will only compute the state of the electrons. So the summation in

now extends only over the electrons. But this summation does includes potentials of the electrons due
to the attraction by the nuclei, and those terms are no longer equal to minus the corresponding poten-
tials. You may need to evaluate these terms explicitly. But that is not too bad, as these potentials are
now known functions of the individual electron positions only. The difficult term, due to the electron-
electron interaction, is still given by minus the corresponding potential.

Finally, you might wonder where the virial theorem comes from. Well, one way to prove the virial theo-
rem, as found in quantum textbooks and on Wikipedia, is to work out the commutator in

using the formulae in chapter 4.5.4, to give

and then note that the left hand side above is zero for stationary states, (in other words, for states with
a precise total energy). This follows the classical way of deriving the classical virial theorem, but requires
a messy purely mathematical derivation. The theorem then pops up out of the complex mathematics
without any plausible physical reason why there would be such a theorem in the first place.

The original derivation by Fock in 1930 is much more physically appealing and more instructive. The idea
is to slightly stretch the given quantum system: replace every position coordinate coordinate   by a
slightly larger one      . Here   is assumed to be a vanishingly small number. We are in-
terested in what the expectation potential and kinetic energy are in this slightly stretched system.

First however, recall that the square magnitude of the wave function gives the probability of that state,
and that all probabilities must integrate together to 1, certainty. Phrased differently, the expectation
value of one must be one;  , what else? But clearly, if you integrate the same square wave
function magnitude over a slightly larger domain, you will get a value slightly greater than one. This
problem is easily fixed, however, by multiplying the wave function in the stretched system by a suitable
constant slightly less than one. Then   too. (The precise value of the constant depends on the
number of particles and is not important.)

Next, the expectation kinetic energy consists of terms like   times   because
of the form of the kinetic energy operator. Because of the stretching of the coordinate in the bottom of
the derivative, each of these terms changes by a factor  , so

For the potential energy we can use a linear Taylor series to figure out how it changes:

where in the right hand side   and its derivatives are evaluated at  . From that

From the above expressions, it is seen that compared to the unstretched system, in the stretched sys-
tem the sum of expectation kinetic and potential energies is different by an amount
But, as described in {A.7}, if you mess up an energy wave function by an amount of order  , the expec-
tation energy should only be messed up by an amount proportional to  , not  . (In brief, the amounts
of the other energy eigenfunctions found in the messed-up wave function are proportional to  . How-
ever, the probabilities of their energies are proportional to the squares of the amounts.) So the factor
between parentheses in the expression above must be zero, and that is the virial theorem.