Irreps of space group#

This page formulates little group, small representation, and irreducible representations (irreps) of space groups.

Reciprocal lattice and transformation#

Let \(\mathbf{A} = (\mathbf{a}_{1}, \mathbf{a}_{2}, \mathbf{a}_{3})\) be basis of \(L_{\mathcal{T}}\), which is lattice of translational group \({\mathcal{T}}\). We choose basis of the (crystallographic) reciprocal vectors as

\[\begin{split} \mathbf{B} &= (\mathbf{b}_{1}, \mathbf{b}_{2}, \mathbf{b}_{3}) = A^{-\top} \\ \mathbf{b}_{i} \cdot \mathbf{a}_{j} &= \delta_{ij}. \end{split}\]

When basis vectors \(\mathbf{A}\) is transformed to \(\mathbf{A}' := \mathbf{AP}\),

  • its dual basis vectors in reciprocal space is transformed from \(\mathbf{B}=\mathbf{A}^{-\top}\) to \(\mathbf{B}' := \mathbf{B}\mathbf{P}^{-\top}\).

  • symmetry operation \((\mathbf{R}, \mathbf{v})\) is transformed to

    \[ (\mathbf{P}, \mathbf{0})^{-1} (\mathbf{R}, \mathbf{v}) (\mathbf{P}, \mathbf{0}) = (\mathbf{P}^{-1}\mathbf{RP}, \mathbf{P}^{-1}\mathbf{v}).\]
  • Coefficients of reciprocal vector \(\mathbf{k} = 2 \pi \mathbf{B} \mathbf{k}_{f}\) is transformed to \(\mathbf{k}_{f}' := \mathbf{P}^{\top} \mathbf{k}_{f}\).

Irreps of translation subgroup#

Consider the following irreps of translation subgroup \(\mathcal{T}\) of space group \(\mathcal{G}\),

\[ \Gamma^{(\mathbf{k})} ((\mathbf{E}, \mathbf{t})) = \exp \left( -i \mathbf{k} \cdot \mathbf{t} \right) \quad ( \mathbf{t} \in L_{\mathcal{T}} ), \]
where \(L_{\mathcal{T}}\) is a lattice formed by \(\mathcal{T}\). Let \(L_{\mathcal{T}}^{\ast}\) be a reciprocal lattice of \(L_{\mathcal{T}}\). We can confine \(\mathbf{k}\) within or the surface of a primitive cell of the reciprocal space because, for \(\mathbf{g} \in L_{\mathcal{T}}^{\ast}\), \(\Gamma^{(\mathbf{k})} = \Gamma^{(\mathbf{k} + \mathbf{g})}\). Any Bloch function with \(\mathbf{k}\)
\[ \Psi_{\mathbf{k}}(\mathbf{r}) = e^{ i \mathbf{k} \cdot \mathbf{r} } u_{\mathbf{k}}(\mathbf{r}), \]
where \(u_{\mathbf{k}}(\mathbf{r})\) is periodic with \(L_{\mathcal{T}}\), can be taken as (one-dimensional) basis of \(\Gamma^{\mathbf{k}}\):
\[\begin{split} (\mathbf{E}, \mathbf{t}) \Psi_{\mathbf{k}}(\mathbf{r}) &= \Psi_{\mathbf{k}}( (\mathbf{E}, \mathbf{t})^{-1} \mathbf{r}) \\ &= e^{-i\mathbf{k}\cdot\mathbf{t}} \Psi_{\mathbf{k}}(\mathbf{r}). \end{split}\]

We write point group of space group \(\mathcal{G}\) as \(\mathcal{P}\). For \((\mathbf{R}, \mathbf{v}) \in \mathcal{G}\), \((\mathbf{R}, \mathbf{v}) \Psi_{\mathbf{k}}(\mathbf{r})\) is Bloch function with \(\mathbf{Rk}\):

(1)#\[\begin{split} (\mathbf{E}, \mathbf{t}) (\mathbf{R}, \mathbf{v}) \Psi_{\mathbf{k}}(\mathbf{r}) &= (\mathbf{R}, \mathbf{v}) (\mathbf{E}, \mathbf{R}^{-1}\mathbf{t}) \Psi_{\mathbf{k}}(\mathbf{r}) \nonumber \\ &= \exp \left( -i \mathbf{k} \cdot \mathbf{R}^{-1} \mathbf{t} \right) (\mathbf{R}, \mathbf{v}) \Psi_{\mathbf{k}}(\mathbf{r}) \nonumber \\ &= \exp \left( -i \mathbf{Rk} \cdot \mathbf{t} \right) (\mathbf{R}, \mathbf{v}) \Psi_{\mathbf{k}}(\mathbf{r}) \quad (\because \mathbf{R} \in O(3) ). \end{split}\]

Be careful the basis for calculating inner product in Eq. (1), when you use crystallographic coordinates [1]:

\[\begin{split} \mathbf{v} &=: \mathbf{A} \mathbf{v}_{f} \\ \mathbf{A}^{-1} \mathbf{R} \mathbf{A} &=: \mathbf{R}_{f} \quad \in \mathrm{SL}(3) \\ (\mathbf{k}, \mathbf{R}^{-1} \mathbf{v}) &= (\mathbf{R} \mathbf{k}, \mathbf{v}) \nonumber \\ &= 2 \pi \left( ( \mathbf{A}^{\top} \mathbf{A}^{-1} ) \mathbf{R}_{f} ( \mathbf{A}^{\top} \mathbf{A}^{-1} )^{-1} \mathbf{k}_{f}, \mathbf{v}_{f} \right) \nonumber \\ &= 2 \pi \left( \mathbf{R}_{f}^{\top} \mathbf{k}_{f}, \mathbf{v}_{f} \right) \quad (\because \mathbf{R}_{f}^{\top} = \mathbf{A}^{\top} \mathbf{R}^{-1} \mathbf{A}^{-\top} ). \end{split}\]

Little group#

Let \(\mathcal{P}\) be point group of space group \(\mathcal{G}\). We define little co-group of \(\mathbf{k}\),

\[ \overline{\mathcal{G}}^{\mathbf{k}} = \left\{ \mathbf{R} \in \mathcal{P} \mid \mathbf{Rk} \equiv \mathbf{k} \right\}, \]
where \(\equiv\) is up to reciprocal vectors in \(L_{\mathcal{T}}^{\ast}\). When we use crystallographic coordinates w.r.t. primitive basis vectors, the condition \(\mathbf{Rk} \equiv \mathbf{k}\) is written as
\[ \mathbf{Rk} \equiv \mathbf{k} \Longleftrightarrow \mathbf{R}_{f}^{\top} \mathbf{k}_{f} - \mathbf{k}_{f} \in \mathbb{Z}^{3}. \]

Because lattice \(L_{\mathcal{T}}\) is invariant with \(\mathbf{R} \in \mathcal{P}\), we obtain

\[ \mathbf{R} \mathbf{k}_{1} \equiv \mathbf{k}_{2} \Rightarrow \overline{\mathcal{G}}^{\mathbf{k}_{2}} = \mathbf{R} \overline{\mathcal{G}}^{\mathbf{k}_{1}} \mathbf{R}^{-1}. \]
When we decompose point group \(\mathcal{P}\) as
\[ \mathcal{P} = \coprod_{i=1}^{q} \mathbf{P}_{i} \overline{\mathcal{G}}^{\mathbf{k}_{1}}, \]
\(\mathbf{P}_{i}\mathbf{k}_{1} \equiv \mathbf{k}_{i}\) with star of \(\mathbf{k}_{1}\), \(\left\{ \mathbf{k}_{1}, \dots, \mathbf{k}_{q} \right\}\).

We define little group (of the first kind) of \(\mathbf{k}\) as

\[ \mathcal{G}^{\mathbf{k}} = \coprod_{ \{ i \mid \mathbf{S}_{i} \in \overline{\mathcal{G}}^{\mathbf{k}} \} } (\mathbf{S}_{i}, \mathbf{w}_{i}) \mathcal{T}, \]
where \(\mathcal{G} = \coprod_{i} (\mathbf{S}_{i}, \mathbf{w}_{i}) \mathcal{T}\). We can decompose space group \(\mathcal{G}\) as
\[ \mathcal{G} = \coprod_{i=1}^{q} (\mathbf{P}_{i}, \mathbf{x}_{i}) \mathcal{G}^{\mathbf{k}_{1}}. \]
Then, for Bloch function \(\Psi_{\mathbf{k}_{1}}(\mathbf{r})\), \((\mathbf{P}_{i}, \mathbf{x}_{i}) \mathcal{G}^{\mathbf{k}_{1}} \Psi_{\mathbf{k}_{1}}(\mathbf{r})\) is Bloch function for \(\mathbf{k}_{i}\). Thus, our task to calculate irreps of \(\mathcal{G}\) is boiled down to calculating irreps of little group \(\mathcal{G}^{\mathbf{k}}\), which is called small representation.

Projective representation#

A projective representation of group \(G\) is a non-singular matrix function \(\Delta\) on \(G\) if it satisfied the following conditions: For each group product \(G_{k} = G_{i}G_{j}\), there exists scalar function \(\mu\) (factor system) such that

\[ \Delta(G_{i}) \Delta(G_{j}) = \mu(G_{i}, G_{j}) \Delta(G_{k}), \]
and
\[ \mu(G_{i}, G_{j}G_{k}) \mu(G_{j}, G_{k}) = \mu(G_{i}G_{j}, G_{k}) \mu(G_{i}, G_{j}) \quad \mbox{(cocycle condition)}. \]

Small representation#

Let \(\Gamma^{\mathbf{k}}_{p}\) be irrep of \(\mathcal{G}^{\mathbf{k}}\), for \(\mathbf{S}_{i} \mathbf{S}_{j} = \mathbf{S}_{k}\),

\[ \mathbf{\Gamma}^{\mathbf{k}}_{p}((\mathbf{S}_{i}, \mathbf{w}_{i})) \mathbf{\Gamma}^{\mathbf{k}}_{p}((\mathbf{S}_{j}, \mathbf{w}_{j})) = \exp \left( -i \mathbf{k} \cdot ( \mathbf{w}_{i} + \mathbf{S}_{i} \mathbf{w}_{j} - \mathbf{w}_{k} ) \right) \mathbf{\Gamma}^{\mathbf{k}}_{p}((\mathbf{S}_{k}, \mathbf{w}_{k})). \]
Then, Simplifying by
\[\begin{split} \mathbf{\Gamma}^{\mathbf{k}}_{p}((\mathbf{R}, \mathbf{v})) &=: \exp \left( -i \mathbf{k} \cdot \mathbf{v} \right) \mathbf{D}^{\mathbf{k}}_{p}((\mathbf{R}, \mathbf{v})) \\ \mathbf{D}^{\mathbf{k}}_{p}((\mathbf{S}_{i}, \mathbf{w}_{i})) \mathbf{D}^{\mathbf{k}}_{p}((\mathbf{S}_{j}, \mathbf{w}_{j})) &= \exp \left( -i \mathbf{g}_{i} \cdot \mathbf{w}_{j} \right) \mathbf{D}^{\mathbf{k}}_{p}((\mathbf{S}_{k}, \mathbf{w}_{k})), \end{split}\]
where
\[\begin{split} \mathbf{g}_{i} &= \mathbf{S}_{i}^{-1} \mathbf{k} - \mathbf{k} \\ &= 2 \pi \mathbf{B} \left( \mathbf{S}_{i, f}^{\top} \mathbf{k}_{f} - \mathbf{k}_{f} \right) \quad (\mathbf{S}_{i, f} = \mathbf{A} \mathbf{S}_{i} \mathbf{A}^{-1}). \end{split}\]

If we choose origin such that \((\mathbf{E}, \mathbf{t}) \in \mathcal{G}\), \(\mathbf{D}^{\mathbf{k}}_{p}((\mathbf{E}, \mathbf{t})) = \mathbf{1}\) and \(\exp \left( -i \mathbf{g}_{i} \cdot (\mathbf{w}_{j} + \mathbf{t}) \right) =\exp \left( -i \mathbf{g}_{i} \cdot \mathbf{w}_{j} \right) \) hold. Then, it is sufficient to consider on \(\overline{\mathcal{G}}^{\mathbf{k}}\),

\[ \mathbf{D}^{\mathbf{k}}_{p}(\mathbf{S}_{i}) \mathbf{D}^{\mathbf{k}}_{p}(\mathbf{S}_{j}) = \exp \left( -i \mathbf{g}_{i} \cdot \mathbf{w}_{j} \right) \mathbf{D}^{\mathbf{k}}_{p}(\mathbf{S}_{k}) \]
Here, \(\mu_{\mathbf{k}}(S_{i}, S_{j}) := \exp \left( -i \mathbf{g}_{i} \cdot \mathbf{w}_{j} \right)\) holds the cocycle condition and \(\mu_{\mathbf{k}}(\mathbf{E}, \mathbf{E}) = 1\). Now, our task is to enumerate projective irreps of little co-group \(\overline{\mathcal{G}}^{\mathbf{k}}\) with factor system \(\mu_{\mathbf{k}}\).

When we use crystallographic coordinates, the factor system is written as

\[\begin{split} \mu_{\mathbf{k}}(\mathbf{S}_{i}, \mathbf{S}_{j}) &= \exp \left( -i (\mathbf{g}_{i}, \mathbf{w}_{j}) \right) \\ &= \exp \left( -2 \pi i (\mathbf{S}_{i, f}^{\top} \mathbf{k}_{f} - \mathbf{k}_{f}, \mathbf{w}_{j, f}) \right) \quad (\mathbf{w}_{j} =: \mathbf{A} \mathbf{w}_{j, f}). \end{split}\]