2
$\begingroup$

Let us say I have a classical field theory, with a field $\phi(\vec r,t)$ which satisfies the relevant Euler-Lagrange equation for the Lagrangian $\mathscr{L}$. Is the general procedure (i.e. one that will work for any $\mathscr{L}$) for going from this classical field to the quantum field operators in QFT?

I know that one can look at commutators, but is this really enough to uniquely specify the field operators?

$\endgroup$
1
  • $\begingroup$ No, quantization is generally not unique. Take e.g. the harmonic oscillator. $\endgroup$
    – Qmechanic
    Commented Jul 31, 2017 at 13:30

1 Answer 1

2
$\begingroup$

The Euler-Lagrange equation has a set of solutions. In classical physics, we consider only solutions which are c-number valued (or Grassmann-number valued if you include "classical" fermion fields); to quantise, we use solutions whose value is a linear operator from a Hilbert space to a Hilbert space (usually the same one). The trick is to ensure they satisfy canonical commutation relations, CCRs, (or canonical anticommutation relations, CARs, for fields satisfying Fermi-Bose stats) analogous to canonical relations. Just as Poisson brackets satisfy $\{q_i,\,p_j\}=\delta_{ij},\,\{q_i,\,q_j\}=\{p_i,\,p_j\}=0$ in discrete classical mechanics, and the first of these generalises in classical field theory to $\{\phi(t,\,\mathbf{x}),\,\pi(t,\,\mathbf{y})\}=\delta(\mathbf{x}-\mathbf{y})$, for a QFT we take $[\hat{\phi}(t,\,\mathbf{x}),\,\hat{\pi}(t,\,\mathbf{y})]_\pm=\text{i}\hbar\delta(\mathbf{x}-\mathbf{y})$ where we use a commutator or anticommutator accordingly (and of course also the $\hat{\phi}$-$\hat{\phi}$ and $\hat{\pi}$-$\hat{\pi}$ (anti)commutators vanish).

Linear ELEs lead to an especially simple result. Suppose a real classical $\phi$ satisfies a linear ELE with solutions $\phi=\sum_\sigma(\phi_\sigma a_\sigma +\text{c.c.})$ with complex coefficients $a_\sigma$ and complex functions $\phi_\sigma$; then the quantised counterpart is the Hermitian field $\hat{\phi}=\sum_\sigma(\phi_\sigma \hat{a}_\sigma +\text{h.c.})$. (You can obtain a similar expression for $\hat{\pi}$ from this.) The spacetime-constant coefficients are promoted to operators, the $\hat{a}_\sigma$ being annihilation operators so $\hat{\phi}| 0\rangle=\sum_\sigma\phi_\sigma^\ast\hat{a}_\sigma^\dagger| 0\rangle$ is a linear combination of $1$-particle states, with $| 0\rangle$ the vacuum ket.

$\endgroup$
6
  • $\begingroup$ Hi, thanks for your answer. Please could you go into a little more detail of how we use the CCRs and CARs to find the field operators? i.e. how do we know that $\hat{\phi}=\sum_\sigma(\phi_\sigma \hat{a}_\sigma +\text{h.c.})$ is the correct field operator? $\endgroup$ Commented Jul 31, 2017 at 12:38
  • $\begingroup$ @Quantumspaghettification That's the most general Hermitian solution of a linear ELE. We don't deduce it from canonical relations, but we can use this solution format to restate those relations equivalently as (anti)commutator relations on the ladder operators. $\endgroup$
    – J.G.
    Commented Jul 31, 2017 at 14:02
  • $\begingroup$ Sorry to be a pain, but how do we know that it is the most general? i.e. why does swapping the complex coefficients for creation and annihilation operators make the most general Hermitian field? $\endgroup$ Commented Jul 31, 2017 at 14:38
  • $\begingroup$ @Quantumspaghettification Because the solution basis is still the same set of functions; we're just changing the "scalars" used as coefficients to span a vector space. $\endgroup$
    – J.G.
    Commented Jul 31, 2017 at 14:46
  • 1
    $\begingroup$ @Quantumspaghettification The only thing that's specific is their (anti)commutators, deduced from canonical relations. You can use the formula for $\hat{\phi}$ to obtain a Hamiltonian in which the role of these operators in particle number is obvious; you then conclude all $\hat{a}$s annihilate the vacuum ket. $\endgroup$
    – J.G.
    Commented Jul 31, 2017 at 15:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.