Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

According to Ballentine, an irreducible tensor of degree k can be defined as a set of $2k + 1$ operators $\{T_q^{\;\;(k)}:(-k \le q \le k)\}$ satisfying the following commutation relations: $$ [J_\pm, T_q^{\;\;(k)}] = \hbar \left( (k \mp q)(k \pm q + 1) \right)^\frac{1}{2} T_{q \pm 1}^{\;\;(k)} $$ $$ [J_z, T_q^{\;\;(k)}] = \hbar q T_{q}^{\;\;(k)} $$ The usual examples employ scalars, vector operators and the irreducible subspaces of cartesian tensor operators. The irreducible tensors that arise from these constructs all have integral degree.

Is it possible to have irreducible tensors of half integral degree, for example $k=\frac{1}{2}$? If so is there a simple example?

share|cite|improve this question
up vote 2 down vote accepted

1) Yes, I assume that OP is already familiar with the irreducible finite-dimensional representations $V_j$ of the Lie algebra $so(3)\cong su(2)$; that they are classified by a non-negative half-integer $j\in \frac{1}{2}\mathbb{N}_{0}$; and that the eigenstates $|j,m\rangle \in V_j$ for $\vec{J}^2$ and $J_z$ satisfy

$$ \vec{J}^2 |j,m\rangle ~=~ \hbar^2 j(j+1)|j,m\rangle , \qquad J_z |j,m\rangle ~=~ \hbar m|j,m\rangle ,$$ $$J_{\pm}|j,m\rangle ~=~ \hbar \sqrt{ (j \mp m)(j \pm m + 1)}|j,m\pm 1\rangle ,$$

where $-j\leq m\leq j$ such that $j-m\in\mathbb{Z}$.

2) Instead, it seems that OP is really asking about the tensor operators $T_q^{\;\;(k)}$ described in eq. (7.112) of the book Quantum mechanics: a modern development by Leslie E. Ballentine,

$$[J_{\pm}, T_q^{\;\;(k)}] ~=~ \hbar \sqrt{ (k \mp q)(k \pm q + 1)} T_{q \pm 1}^{\;\;(k)}, \qquad [J_z, T_q^{\;\;(k)}] ~=~ \hbar q T_{q}^{\;\;(k)}. \qquad (7.112) $$

3) A simple example of such operators

$$T_q^{\;\;(k)}:~V~\to~ V, \qquad V~=~ \oplus_{j\in \frac{1}{2}\mathbb{N}_{0}} V_j,$$

can be build as

$$T_q^{\;\;(k)}~:=~ | k,q\rangle \otimes \langle 0,0|,$$

where $k\in \frac{1}{2}\mathbb{N}_{0}$ is a non-negative half-integer, and $-k\leq q\leq k$ such that $k-q\in\mathbb{Z}$. It is straightforward to check that this example satisfies the wanted commutation relations (7.112).

4) Note that any set of $2k+1$ operators $T_q^{\;\;(k)}:V_0\to V_k$ with $-k\leq q\leq k$ that satisfy eq. (7.112) would have to be proportional to $| k,q\rangle \otimes \langle 0,0|$ with the same proportionality factor independent of $q$, due to Wigner–Eckart theorem.

share|cite|improve this answer
That is a very interesting construction, although it seems kind of unphysical to me because it mixes bosonic and fermionic states in the half integral case. – Friedrich Nov 17 '11 at 17:13
Can you think of an example that could be in actual use? – Friedrich Nov 17 '11 at 19:36
Well, if the terms bosonic and fermionic in this context stand for integral and half-integral irreps, respectively, then it is a fact in physics, that any fermionic operator would turn a bosonic state into a fermionic state, and vice-versa. – Qmechanic Nov 17 '11 at 22:56

Yes, these "tensors" are called "spinors" for $k=1/2$ or perhaps "spintensors" for $k\in{\mathbb Z}+1/2$ but they satisfy the very same algebra you wrote. It is somewhat misleading to ask about "a simple example". There's just one example – the example or the solution – for each value of $k$, up to the choice of a basis and up to the freedom to pick a greater degeneracy. And you have actually written it down.

The collection of $k=1/2$ operators is $2k+1=2$-dimensional and it's known as the spinor. The allowed values are $q=+1/2$ and $q=-1/2$ and your formulae actually describe the exact form of all the commutators for $k=1/2$ as well as any other allowed value of $k$ (integer or half-integer-valued).

Spinors are essential to describe elementary fermions, including electrons, neutrinos, and quarks (and even protons, neutrons, and all other composite spin-1/2 particles). These may be viewed as "many examples" but the maths how the $\vec J$ generators commute with the spinorial objects is exactly the same in all cases.

The reason why one can construct tensors with a "half-integer number of indices" is that the group $SO(3)$ of rotations, generated by the angular momentum, is isomorphic to $SU(2)/{\mathbb Z}_2$. The quotient by ${\mathbb Z}_2$ means that one only allows integer spins. However, if one allows representations of the whole $SU(2)\sim Spin(3)$, one also allows representations (and tensor operators) that change their sign after rotations by 360 degrees and they correspond to $k$ which differs by $1/2$ from an integer.

share|cite|improve this answer
I am not sure if I understand correctly. You are saying that there is just one solution for each $k$? Let's take $k=1$ then there are quite some cases, for example the operators $\{J_+, J_-, J_z\}$ built from linear combinations of the angular momentum operators $\vec{J}$ and also the same combinations of the position operators $\vec{X}$ or the momentum operators $\vec{P}$. – Friedrich Nov 17 '11 at 17:43
I forgot some factors, it's actually $\{-J_+/\sqrt{2},J_-/\sqrt{2},J_z\}$, where $J_+$ and $J_-$ are the usual angular moment raising and lowering operators. – Friedrich Nov 17 '11 at 19:32
Dear @Friedrich, $J_\pm$ is the same thing as $J_x\pm i J_y$, up to the sign of $i$ and the overall normalization, so the difference between $\{J_x,J_y,J_z\}$ and $\{J_+,J_-, J_z\}$ is just a convention about the basis - which linear combinations of the three quantities are being written as the standardized ones. The 3D space of linear combinations of these 3 tensor operators is exactly the same in both cases. That's why we say that the $j=1$ irreducible representation of the group is unique. The same hold for any allowed $j$, too. – Luboš Motl Dec 12 '11 at 11:10

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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