nLab Mackey theory

Contents

Idea

In representation theory, Mackey theory refers to a set of tools for constructing and classifying irreducible representations of a group GG from that of a normal subgroup NGN \subset G, by using induction from the stabilizer groups of the adjoint action of GG on the set of irreps of NN.

(Also referred to as little group-methods, cf. Wigner classification.)

Details

For finite semidirect products with abelian groups

Consider

and denote:

  • A˜Hom(A, ×)\widetilde A \,\coloneqq\, Hom(A, \mathbb{C}^\times) the dual group of group characters, hence, since AA is abelian, of irreducible representations of AA,

  • the adjoint action of GG on AA and the induced action on A˜\widetilde A:

    G×A˜ A˜ (g,Aχ ×) χ(g 1(-)g) \begin{array}{ccc} G \times \widetilde A &\longrightarrow& \widetilde A \\ \big(g, A \xrightarrow{\chi} \mathbb{C}^\times \big) &\mapsto& \chi\big(g^{-1} (\text{-}) g\big) \end{array}
  • (χ i) iA˜/H(\chi_i)_{i \in \widetilde{A}/H} a set of representatives of the resulting orbits of HH in A˜\widetilde A,

  • H iStab H(χ i)H_i \,\coloneqq\, Stab_H(\chi_i) the corresponding stabilizer subgroups of HH,

  • G iH iAG_i \,\coloneqq\, H_i \rtimes A the corresponding subgroups of GG

Finally, denote

χ^ i:G iAχ i ×, \widehat{\chi}_i \,\colon\, G_i \twoheadrightarrow A \xrightarrow{\chi_i} \mathbb{C}^{\times} \,,

which is a group character on G iG_i, by the definition of H iH_i.

Similarly, for ρIrr(H i)\rho \in Irr(H_i) an irrep of H iH_i, write

ρ^:G iH iρGL(V) \widehat{\rho} \,\colon\, G_i \twoheadrightarrow H_i \xrightarrow{\rho} GL(V)

and consider the induced representations of the tensor products:

(1)θ i,ρInd G i G(χ^ iρ^)Rep (G). \theta_{i,\rho} \;\coloneqq\; Ind_{G_i}^G \big( \widehat{\chi}_i \otimes \widehat{\rho} \big) \;\in\; Rep_{\mathbb{C}}(G) \,.

Proposition

  1. The representations θ i,ρ\theta_{i,\rho} (1) are irreducible.

  2. If θ i,ρθ i,ρ\theta_{i,\rho} \simeq \theta_{i',\rho'} then i=ii = i' and ρρ\rho \simeq \rho'.

  3. Every irrep of GG is of the form θ i,ρ\theta_{i, \rho}, up to isomorphism.

(Serre 1977 Prop. 25 p 62)

Examples

Example

Consider the semidirect product group G n 2G \equiv \mathbb{Z}_n \rtimes \mathbb{Z}_2, with

In order find all the (complex) irreps of n 2\mathbb{Z}_n \rtimes \mathbb{Z}_2 we unwind the statement of Prop. :

(We denote congruence classes by square brackets: []: n[-] \colon \mathbb{Z} \twoheadrightarrow \mathbb{Z}_n.)

First, the complex irreps of the normal subgroup n\mathbb{Z}_n are themselves labeled by [k] n[k] \in \mathbb{Z}_n and given by

n χ [k] × [r] e 2πirkn. \begin{array}{ccc} \mathbb{Z}_n &\xrightarrow{\;\; \chi_{[k]} \;\;}& \mathbb{C}^\times \\ [r] &\mapsto& e^{2 \pi \mathrm{i} \tfrac{r k}{n}} \mathrlap{\,.} \end{array}

On these, the nontrivial element σ 2\sigma \in \mathbb{Z}_2 acts by σ:[k][k]\sigma \colon [k] \mapsto [-k]:

(σχ [k])([r]) χ [k](σ 1[r]σ) χ [k](σ 1σ[r]) =χ [k]([r]) e 2πi(r)kn =e 2πir(k)n χ [k]([r]). \begin{array}{l} (\sigma \chi_{[k]})\big([r]\big) \\ \;\equiv\; \chi_{[k]}\big(\sigma^{-1}\cdot [r] \cdot \sigma\big) \\ \;\equiv\; \chi_{[k]}\big(\sigma^{-1}\cdot \sigma\cdot [-r]\big) \\ \;=\; \chi_{[k]}\big([-r]\big) \\ \;\equiv\; e^{2 \pi \mathrm{i} \tfrac{(-r)k}{n}} \\ \;=\; e^{2 \pi \mathrm{i} \tfrac{r (-k)}{n}} \\ \;\equiv\; \chi_{[-k]}\big([r]\big) \mathrlap{\,.} \end{array}

From this we have two kinds of orbits:

  1. the singleton orbits with stabilizer 2\mathbb{Z}_2:

    {[0]}\big\{ [0] \big\}, which always exists

    and {[n/2]}\big\{ [n/2] \big\}, which exists when nn is an even number,

  2. the pairs with stabilizer the trivial group 11:

    {[k],[k]}\big\{[k],[-k]\big\}, for k{1,,(n1)/2}k \in \{1, \cdots, \lfloor(n-1)/2\rfloor\} (with \lfloor-\rfloor the floor).

The first singleton in the first case gives rise to two irreps, corresponding to the two irreps of 2\mathbb{Z}_2 (the trivial representation and the sign representation):

Ind p 2 p 2Id((([r],[s])1)(([r],[s])1))=triv \underset{Id}{ \underbrace{ Ind ^{\mathbb{Z}_p \rtimes \mathbb{Z}_2} _{\mathbb{Z}_p \rtimes \mathbb{Z}_2} } } \Big( \big( ([r], [s]) \mapsto 1 \big) \otimes \big( ([r], [s]) \mapsto 1 \big) \Big) \;=\; triv

and

Ind p 2 p 2((([r],[s])1)(([r],[s])e πis))=(([r],[s])e πis), Ind ^{\mathbb{Z}_p \rtimes \mathbb{Z}_2} _{\mathbb{Z}_p \rtimes \mathbb{Z}_2} \Big( \big( ([r], [s]) \mapsto 1 \big) \otimes \big( ([r], [s]) \mapsto e^{\pi \mathrm{i} s} \big) \Big) \;=\; \big( ([r], [s]) \mapsto e^{\pi \mathrm{i} s} \big) \,,

while the second singleton in the first case similarly gives

Ind p 2 p 2((([r],[s])e πir)(([r],[s])1))=(([r],[s])e πir) Ind ^{\mathbb{Z}_p \rtimes \mathbb{Z}_2} _{\mathbb{Z}_p \rtimes \mathbb{Z}_2} \Big( \big( ([r], [s]) \mapsto e^{\pi \mathrm{i} r} \big) \otimes \big( ([r], [s]) \mapsto 1 \big) \Big) \;=\; \big( ([r], [s]) \mapsto e^{\pi \mathrm{i} r} \big)

and

Ind p 2 p 2((([r],[s])e πir)(([r],[s])e πis))=(([r],[s])e πi(r+s)). Ind ^{\mathbb{Z}_p \rtimes \mathbb{Z}_2} _{\mathbb{Z}_p \rtimes \mathbb{Z}_2} \Big( \big( ([r], [s]) \mapsto e^{\pi \mathrm{i} r} \big) \otimes \big( ([r], [s]) \mapsto e^{\pi \mathrm{i} s} \big) \Big) \;=\; \big( ([r], [s]) \mapsto e^{\pi \mathrm{i} (r + s)} \big) \,.

The second case gives rise to one irrep for each k{1,,(n1)/2}k \in \{1, \cdots, \lfloor(n-1)/2\rfloor\}:

Ind p p 2(([r]e 2πirkn)([r]1)) [ p 2] [ p]([r]e 2πirkn). \begin{array}{l} Ind ^{ \mathbb{Z}_p \rtimes \mathbb{Z}_2 } _{ \mathbb{Z}_p } \Big( \big( [r] \mapsto e^{2 \pi \mathrm{i} \tfrac{r k}{n}} \big) \otimes \big( [r] \mapsto 1 \big) \Big) \\ \;\simeq\; \mathbb{C}\big[ \mathbb{Z}_p \rtimes \mathbb{Z}_2 \big] \otimes_{ \mathbb{C}[\mathbb{Z}_p] } \Big( [r] \mapsto e^{2 \pi \mathrm{i} \tfrac{r k}{n}} \Big) \mathrlap{\,.} \end{array}

These latter irreps are 2-dimensional, with an evident matrix representation ()^\widehat{(-)} given by

([1],[0])^=[e +2πikn 0 0 e 2πikn],([0],[1])^=[0 1 1 0]. \widehat{ \big( [1], [0] \big) } \;=\; \left[ \begin{matrix} e^{+ 2 \pi \mathrm{i} \tfrac{k}{n}} & 0 \\ 0 & e^{- 2 \pi \mathrm{i} \tfrac{k}{n}} \end{matrix} \right] \,,\;\;\;\;\;\; \widehat{ \big( [0], [1] \big) } \;=\; \left[ \begin{matrix} 0 & 1 \\ 1 & 0 \end{matrix} \right] \mathrlap{\,.}

As a consistency check that we found all irreps, we verify that the sum of squares formula holds:

We have found two 1-dimensional irreps when nn is odd and four of them when nn is even, together with (n1)/2\lfloor (n-1)/2\rfloor irreps of dimension 2, whence the sum of their squares of dimensions is,

for odd nn:

21 2+(n1)/22 2=2+n124=2+2(n1)=2n 2 \cdot 1^2 \,+\, \lfloor (n-1)/2\rfloor \cdot 2^2 \;=\; 2 + \frac{n-1}{2} 4 \;=\; 2 + 2 (n-1) \;=\; 2 n

and for even nn:

41 2+(n1)/22 2=22+n224=22+2(n2)=2n, 4 \cdot 1^2 \,+\, \lfloor (n-1)/2\rfloor \cdot 2^2 \;=\; 2 \cdot 2 \,+\, \frac{n-2}{2} 4 \;=\; 2 \cdot 2 + 2 \cdot (n-2) \;=\; 2 n \,,

correctly coinciding with the order of our group :

| n 2|=| 2|| n|=2n. {\vert \mathbb{Z}_n \rtimes \mathbb{Z}_2 \vert} \;=\; {\vert \mathbb{Z}_2 \vert} \cdot {\vert \mathbb{Z}_n \vert} \;=\; 2 n \,.

Example

For construction of irreps of wreath products of groups see there.

References

The original article:

General discussion:

  • J. M. G. Fell, R. S. Doran: Representations of *\ast-Algebras, Locally Compact Groups, and Banach *\ast-Algebraic Bundles, volume 2: Banach *\ast-Algebraic Bundles, Induced Representations, and the Generalized Mackey Analysis, Academic Press (1988) [ISBN:9780122527227]

Review for the case of finite groups:

Further developments:

Review for the case of Lie groups (cf. Wigner classification):

Last revised on May 11, 2025 at 17:29:19. See the history of this page for a list of all contributions to it.