Consider a set of boson/fermion creation and annihilation operators satisfying the canonical (anti-)commutation rules (CCR/CAR):

$$ [a_i, a_j]_\eta = [a^\dagger_i, a^\dagger_j]_\eta = 0, \quad [a_i, a^\dagger_j]_\eta = \delta_{ij} \quad (i,j = 1,...,N) $$

where $\eta$ is the statistical sign ($+1$ for boson, and $-1$ for fermion), $[A,B]_\eta = AB - \eta BA$, and $N$ is the dimension of one-particle Hilbert space. A general canonical transformation mixes the creation and annihilation operators to produce a new set of bosons/fermions $\{b_j\}_{j=1}^N$, which I call the Bogoliubov quasi-particles:

$$ b_i = \sum_{j=1}^N (u_{ij} a_j + v_{ij} a^\dagger_j) , \quad b^\dagger_i = \sum_{j=1}^N (v^*_{ij} a_j + u^*_{ij} a^\dagger_j) $$

and the $\{b_i\}$ operators should also satisfy the CCR/CAR:

$$ [b_i, b_j]_\eta = [b^\dagger_i, b^\dagger_j]_\eta = 0, \quad [b_i, b^\dagger_j]_\eta = \delta_{ij} \quad (i,j = 1,...,N) $$

One can show that to preserve the CCR/CAR, the matrices $U = \{u_{ij}\}$, $V = \{v_{ij}\}$ should satisfy the requirement

$$ U V^{\mathsf{T}} = \eta V U^{\mathsf{T}} \quad \text{and} \quad U U^\dagger - \eta V V^\dagger = 1 $$

Question: Let $|0_a\rangle, |0_b\rangle$ be the vacuum states of the $a, b$ particles respectively, i.e.

$$ \forall i = 1,...,N: \quad a_i |0_a\rangle = 0, \quad b_i |0_b\rangle = 0 $$

How to express $|0_b\rangle$ in terms of states and operators of the $a$ particles?

  • 1
    $\begingroup$ It's a bit more than an "educated guess," but it it does require that the $U$ matrix be invertible, which is not globally true. For the Bosonic case look at arxiv.org/abs/1608.03289 eq 2.17 2.18 $\endgroup$
    – mike stone
    Nov 10, 2022 at 13:39
  • $\begingroup$ I found this paper useful at one point. Not sure that it really gets at what you're looking for, but it might help. $\endgroup$
    – march
    Nov 10, 2022 at 17:40
  • $\begingroup$ You can check this paper by Kita journals.jps.jp/doi/abs/10.1143/JPSJ.65.908. You can find the same calculation also in his book: link.springer.com/10.1007/978-4-431-55405-9 The result is similar to the one in Mike Stone's note. $\endgroup$
    – skdys
    Jan 6, 2023 at 21:19
  • $\begingroup$ @skdys Could you please specify at which chapter the derivation of the ground state of quasi-particles is given? I had a brief look but did not find it. $\endgroup$ Jan 7, 2023 at 0:42
  • $\begingroup$ In the book, Eq. 8.4 is the definition of a many-body wavefunction $\vert {\Phi} \rangle$ as a function of a matrix $\phi$. In Sec. 8.3, the standard BdG equations are derived, and finally, after Eq. 8.46, the connection between this matrix $\phi$ and the usual coherence factor u and v is shown. I believe that if you reorder the steps (maybe with the help of the paper), you can get to a formal mathematical derivation if that is your objective. To me, this looks pretty formal already from a physicist's point of view. $\endgroup$
    – skdys
    Jan 8, 2023 at 19:03

1 Answer 1


The following is summarized from two notes (both can be found here) by Prof. Michael Stone (@mikestone). Confusing typos in these notes are corrected.

From a general perspective, the two set of operators $\{a\}, \{b\}$ are two realizations of the CCR/CAR. By the Stone-von Neumann theorem, there is a unitary operator $\mathcal{U}$ that acts on the many-body Hilbert space such that

$$ b_i = \mathcal{U} a_i \mathcal{U}^\dagger \ \Rightarrow \ b^\dagger_i = \mathcal{U} a^\dagger_i \mathcal{U}^\dagger $$

Then we find $|0_b\rangle \propto \mathcal{U}|0_a\rangle$, since

$$ a_i |0_a\rangle = \mathcal{U}^\dagger b_j \mathcal{U} |0_a\rangle = 0 $$

However, as noted by Prof. Stone, the general form of $\mathcal{U}$ is very complicated. Thus we take another approach.

Assume that $U$ is invertible (this is not an "innocent assumption"). Then $b_i |0_b \rangle = 0$ is equivalent to

$$ 0 = U^{-1}_{ij} b_j |0_b\rangle = (a_i - S_{ij} a^\dagger_j) |0_b \rangle, \quad S \equiv - U^{-1} V $$

Using the constraint $U V^\mathsf{T} = \eta V U^\mathsf{T}$, one can show that

$$ S^\mathsf{T} = \eta S $$

i.e. the matrix $S$ is symmetric for bosons, and anti-symmetric for fermions. Then we construct the bilinear

$$ Q \equiv \frac{1}{2} \sum_{i,j} S_{ij} a^\dagger_i a^\dagger_j $$

and use the BCH formula to calculate

$$ \begin{equation*} e^Q a_i e^{-Q} = a_i + [Q,a_i] + \frac{1}{2}[Q,[Q,a_i]] + \cdots \end{equation*} $$

Fortunately, the commutator

$$ [Q, a_i] = - S_{ij} a^\dagger_j $$

commutes with $Q$ (for both bosons and fermions). Therefore we simply get

$$ \begin{equation*} e^Q a_i e^{-Q} = a_i + [Q,a_i] = a_i - S_{ij} a^\dagger_j \end{equation*} $$

and $b_i |0_b \rangle = 0$ is equivalent to

$$ e^Q a_i e^{-Q} |0_b\rangle = 0 $$

This can be satisfied by all $a_i$ if and only if

$$ e^{-Q} |0_b \rangle \propto |0_a \rangle $$

The "if" is obvious. To show "only if", suppose that $|\psi\rangle \equiv e^{-Q} |0_b \rangle \ne |0_a \rangle$; then $e^Q$, which consists of only $\{a^\dagger_i\}$, cannot annihilate $|\psi\rangle$. Finally, up to a normalization constant $\mathcal{N}$, the vacuum $|0_b\rangle$ is

$$ |0_b\rangle = \mathcal{N} e^Q |0_a \rangle $$

(When I have time I will update on finding $\mathcal{N}$.)


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.