0

I feel like I know the answer to this question, but there are so many misused terms thrown around that its difficult to find good resources on this topic.

Say we have 1 type of particle. Then the Fock space is the sum of n-particle Hilbert spaces obeying some symmetry. Specifically, enter image description here

(Taken from https://en.wikipedia.org/wiki/Fock_space)

As I understand it, Hilbert spaces contain n-particle states. But what exactly an 'n-particle state' is seems to have varying opinions/definitions.

Some resources say they are n-particle states, but then go on to define an n particle state as n creation operators acting on the vacuum;

$a^{\dagger}(p_1)a^{\dagger}(p_2)...a^{\dagger}(p_n) |{0}\rangle$

for 3-momenta $p_i$

Other resources say that this cannot be in the Fock space because the creation operators in QFT create a particle in a definite momentum state. Instead these operators need to be multiplied with a momentum distribution function and then integrated over, creating a 'general n-particle state';

$\int dp^3 f(p)a^{\dagger}(p) |{0}\rangle$

Lastly, I recall that a Hilbert space is a space of functions, but if we are integrating over all momentum space, it isn't obvious to me that the result is indeed a function.

TLDR:

Is $a^{\dagger}(p) |{0}\rangle$ an element of the Fock space? Why or why not? If not, why do so many people say it is?

Qmechanic
  • 220,844

2 Answers2

4

It depends on how rigorous we're being.

At the typical physical level of rigor and for simplicity for a scalar field/particle, $H$ has a "basis" of states given by $a^\dagger(p)\lvert 0\rangle$, $p\in \Sigma$, where $\Sigma\subset \mathbb{R}^{3,1}$ is the mass shell, i.e. $p^2 = m^2$ for $m$ the mass. $\lvert 0\rangle$, the vacuum, is the unique state in the $\mathbb{C}$ ("0-particle space") of the definition.

At the mathematical rigor of level, the "states" $a^\dagger(p)\lvert 0\rangle$ are not inside $H$ because they are not normalizable (their inner product with each other is a $\delta$-distribution, not a finite value). This is exactly analogous to how standard intros to QM treat the Hilbert space as containing "eigenvectors" of position $\lvert x\rangle$ or momentum $\lvert p\rangle$, but they aren't rigorously elements of the space. See this answer of mine or this answer of mine for more on this topic.

Rigorously, the $a(p)^\dagger$ are operator-valued distributions and you only get an actual operator by applying it to some nice test function $f(\vec p)$ (what these are drawn from depends on the exact way you set up the theory, it's common to choose the Schwartz functions), i.e. $a(f)^\dagger := \int f(\vec p)a(p)^\dagger \mathrm{D}\vec p$ is an actual operator that creates a state with momentum wavefunction $f(\vec p)$. There are some subtleties here related to the measure $\mathrm{D}\vec p$ (you should usually pick the Lorentz-invariant measure on $\vec p\in\mathbb{R}^3$), but this is the general idea.

The set of all states $a(f)\lvert 0\rangle$ is overcomplete, but you can turn it into a basis by picking a set of $f$ that form a basis for the space of test functions.

ACuriousMind
  • 132,081
1

You should be familiar with this problem already from elementary quantum mechanics, where the momentum eigen-"state" (better: eigendistribution) $| \vec{p}\rangle$ with $\langle \vec{p} | \vec{p}^{\, \prime }\rangle=\delta^{(3)}(\vec{p}-\vec{p}^{\, \prime})$ is not an element of the Hilbert space $\mathcal{H}$, as $|\psi\rangle \in \mathcal{H}$ requires normalizability $\langle \psi | \psi \rangle < \infty$ with respect to the scalar product defined on $\mathcal{H}$. However, elements of $\mathcal{H}$ can be obtained as linear combinations of the momentum eigendistributions $| \vec{p} \rangle$ of the form $$ |\psi \rangle = \int\limits_{\mathbb{R}^3} \! d^3p \, f(\vec{p}) \, |\vec{p}\rangle, \tag{1} \label{1} $$ if the condition $$ \langle \psi | \psi \rangle = \int\limits_{\mathbb{R}^3} \! d^3p \, |f(\vec{p}) |^2 < \infty \tag{2} \label{2} $$ is fulfilled. Furthermore, if the normalization condition $\langle \psi | \psi \rangle=1$ holds, the vector given in \eqref{1} represents a quantum mechanical state with momentum space wave function $f(\vec{p})$.

The situation is exactly the same in Fock space (also being a Hilbert space). The vacuum state $|0\rangle $ is a properly normalized state with $\langle 0 | 0\rangle=1$. On the other hand, the one-particle momentum eigendistribution $$ |\vec{p}\rangle := a^\dagger(\vec{p}) |0 \rangle \tag{3} \label{3} $$ with $$ \langle \vec{p} | \vec{p}^{\, \prime} \rangle= \langle 0 |a(\vec{p}) a^\dagger(\vec{p}^{\, \prime}) |0\rangle = \langle 0 |\underbrace{[a(\vec{p}), a^\dagger(\vec{p}^\prime) ]}_{\mathcal{N}(\vec{p}) \, \delta^{(3)}(\vec{p}-\vec{p}^{\, \prime})}|0\rangle = \mathcal{N}(\vec{p}) \delta^{(3)}(\vec{p}-\vec{p}^{\, \prime}), \tag{4} \label{4} $$ where $\mathcal{N}(\vec{p})$ is a factor associated with your favourite normalization convention. As before, $|\vec{p}\rangle$ defined in \eqref{3} is not an element of the underlying Hilbert space ($=$ Fock space), but (choosing the noncovariant normalization $\mathcal{N}(\vec{p})=1$ for simplicity) $$ |\psi \rangle = \int\limits_{\mathbb{R}^3} \! d^3p \, f(\vec{p}) | \vec{p} \rangle = \int\limits_{\mathbb{R}^3} \! d^3p \, f(\vec{p}) a^\dagger(\vec{p}) |0\rangle \tag{5} \label{5} $$ with the condition \eqref{2} is a Fock space element. Choosing $\langle \psi | \psi \rangle=1$, the vector in \eqref{5} represents the most general form of a (pure) one particle state of your theory, where $f(\vec{p})$ is the corresponding momentum space wave function.

If you do not like the creation operators $a^\dagger(\vec{p})$, you may define $$ a_f^\dagger := \int\limits_{\mathbb{R}^3} \! d^3p \, f(\vec{p}) a^\dagger(\vec{p})\tag{6} \label{6} $$ creating the normalized state $|\psi \rangle = a_f^\dagger |0\rangle$ given in \eqref{5} from the vacuum state, where the commutation relation $$ [a_f, a_g^\dagger ]= \int\limits_{\mathbb{R}^3} \! d^3 p \, f^\ast(\vec{p}) g(\vec{p}) \tag{7} \label{7} $$ holds.

The construction of the two-particle subspace of Fock space should now be clear. Start with the non-normalizable two-particle momentum eigendistributions $$ |\vec{p}_1, \vec{p}_2 \rangle = a^\dagger(\vec{p}_1) a^\dagger(\vec{p}_2) |0 \rangle \tag{8} \label{8} $$ and construct the Fock space element $$ |\chi \rangle = \int\limits_{\mathbb{R}^3} \! d^3p_1 \!\int\limits_{\mathbb{R}^3} \! d^3 p_2 \, f(\vec{p}_1, \vec{p}_2) a^\dagger (\vec{p}_1) a^\dagger(\vec{p}_2) |0\rangle, \tag{9} \label{9} $$ where $f(\vec{p}_1, \vec{p}_2)=f(\vec{p}_2, \vec{p}_1)$ is symmetric under $\vec{p}_1 \leftrightarrow \vec{p}_2$ (remember $[a^\dagger(\vec{p}_1), a^\dagger(\vec{p}_2) ]=0$) with $$ \int\limits_{\mathbb{R}^3} \! d^3p_1 \! \int\limits_{\mathbb{R}^3} \! d^3p_2\, |f(\vec{p}_1, \vec{p}_2) |^2 < \infty. \tag{10} $$

Hyperon
  • 10,007