Lecture 12

Problem 72. Let (π,V)(\pi,V) be a finite-dimensional representation of a finite group GG. Define U1,U2VVU_{1},U_{2}\subset V\otimes V by

U1=span{𝐯1𝐯2𝐯2𝐯1|𝐯1,𝐯2V},U2=span{𝐯1𝐯2+𝐯2𝐯1|𝐯1,𝐯2V}U_{1}=\mathrm{span}\{{\bf v}_{1}\otimes{\bf v}_{2}-{\bf v}_{2}\otimes{\bf v}_{% 1}\,|\,{\bf v}_{1},{\bf v}_{2}\in V\},\quad U_{2}=\mathrm{span}\{{\bf v}_{1}% \otimes{\bf v}_{2}+{\bf v}_{2}\otimes{\bf v}_{1}\,|\,{\bf v}_{1},{\bf v}_{2}% \in V\}

Show that (ππ,U1)(\pi\otimes\pi,U_{1}) and (ππ,U2)(\pi\otimes\pi,U_{2}) are subrepresentations of (ππ,VV)(\pi\otimes\pi,V\otimes V) and thus show that

(ππ,VV)(Sym2π,Sym2(V))(ππ,2V).(\pi\otimes\pi,V\otimes V)\cong\big{(}\mathrm{Sym}^{2}\pi,\mathrm{Sym}^{2}(V)% \big{)}\oplus\big{(}\pi\wedge\pi,\textstyle\bigwedge^{2}V\big{)}.

.

Solution: Let aia_{i}\in{\mathbb{C}}, 𝐯i(1),𝐯i(2)V{\bf v}_{i}^{(1)},{\bf v}_{i}^{(2)}\in V. For gGg\in G and iai(𝐯i(1)𝐯i(2)𝐯i(1)𝐯i(2))U1\sum_{i}a_{i}\big{(}{\bf v}_{i}^{(1)}\otimes{\bf v}_{i}^{(2)}-{\bf v}_{i}^{(1)% }\otimes{\bf v}_{i}^{(2)}\big{)}\in U_{1}, we have

ππ(g)(iai(𝐯i(1)𝐯i(2)𝐯i(1)𝐯i(2)))\displaystyle\pi\otimes\pi(g)\left(\sum_{i}a_{i}\big{(}{\bf v}_{i}^{(1)}% \otimes{\bf v}_{i}^{(2)}-{\bf v}_{i}^{(1)}\otimes{\bf v}_{i}^{(2)}\big{)}\right) =iaiππ(g)(𝐯i(1)𝐯i(2)𝐯i(1)𝐯i(2))\displaystyle=\sum_{i}a_{i}\pi\otimes\pi(g)\big{(}{\bf v}_{i}^{(1)}\otimes{\bf v% }_{i}^{(2)}-{\bf v}_{i}^{(1)}\otimes{\bf v}_{i}^{(2)}\big{)}
=iai(π(g)𝐯i(1)π(g)𝐯i(2)π(g)𝐯i(1)π(g)𝐯i(2))U1,\displaystyle=\sum_{i}a_{i}\big{(}\pi(g){\bf v}_{i}^{(1)}\otimes\pi(g){\bf v}_% {i}^{(2)}-\pi(g){\bf v}_{i}^{(1)}\otimes\pi(g){\bf v}_{i}^{(2)}\big{)}\in U_{1},

hence (π,π,U1)(\pi,\otimes\pi,U_{1}) is a subrepresentation. Similarly, given iai(𝐯i(1)𝐯i(2)+𝐯i(1)𝐯i(2))U2\sum_{i}a_{i}\big{(}{\bf v}_{i}^{(1)}\otimes{\bf v}_{i}^{(2)}+{\bf v}_{i}^{(1)% }\otimes{\bf v}_{i}^{(2)}\big{)}\in U_{2}, we have

ππ(g)(iai(𝐯i(1)𝐯i(2)+𝐯i(1)𝐯i(2)))\displaystyle\pi\otimes\pi(g)\left(\sum_{i}a_{i}\big{(}{\bf v}_{i}^{(1)}% \otimes{\bf v}_{i}^{(2)}+{\bf v}_{i}^{(1)}\otimes{\bf v}_{i}^{(2)}\big{)}\right) =iaiππ(g)(𝐯i(1)𝐯i(2)+𝐯i(1)𝐯i(2))\displaystyle=\sum_{i}a_{i}\pi\otimes\pi(g)\big{(}{\bf v}_{i}^{(1)}\otimes{\bf v% }_{i}^{(2)}+{\bf v}_{i}^{(1)}\otimes{\bf v}_{i}^{(2)}\big{)}
=iai(π(g)𝐯i(1)π(g)𝐯i(2)+π(g)𝐯i(1)π(g)𝐯i(2))U2,\displaystyle=\sum_{i}a_{i}\big{(}\pi(g){\bf v}_{i}^{(1)}\otimes\pi(g){\bf v}_% {i}^{(2)}+\pi(g){\bf v}_{i}^{(1)}\otimes\pi(g){\bf v}_{i}^{(2)}\big{)}\in U_{2},

hence (π,π,U2)(\pi,\otimes\pi,U_{2}) is a subrepresentation.

Note that U1U2=VVU_{1}\cup U_{2}=V\otimes V as

𝐯1𝐯2=12(𝐯1𝐯2𝐯2𝐯1+𝐯1𝐯2+𝐯2𝐯1){\bf v}_{1}\otimes{\bf v}_{2}=\frac{1}{2}({\bf v}_{1}\otimes{\bf v}_{2}-{\bf v% }_{2}\otimes{\bf v}_{1}+{\bf v}_{1}\otimes{\bf v}_{2}+{\bf v}_{2}\otimes{\bf v% }_{1})

for all 𝐯1,𝐯2V{\bf v}_{1},{\bf v}_{2}\in V. If 𝐞1,𝐞n{\bf e}_{1},\ldots{\bf e}_{n} is a basis of VV it is clear {𝐞i𝐞j𝐞j𝐞i| 1in,1j<i}\{{\bf e}_{i}\otimes{\bf e}_{j}-{\bf e}_{j}\otimes{\bf e}_{i}\,|\,1\leq i\leq n% ,1\leq j<i\} and {𝐞i𝐞j+𝐞j𝐞i| 1in,1ji}\{{\bf e}_{i}\otimes{\bf e}_{j}+{\bf e}_{j}\otimes{\bf e}_{i}\,|\,1\leq i\leq n% ,1\leq j\leq i\} are bases of U1U_{1} and U2U_{2} respectively. Thus dim(vV)=n2=dim(U1)+dim(U2)\operatorname{dim}(v\otimes V)=n^{2}=\operatorname{dim}(U_{1})+\operatorname{% dim}(U_{2}) and VV=U1U2V\otimes V=U_{1}\oplus U_{2}.

Let ϕ\phi be the map

ϕ:Sym2(V)U2;𝐞i𝐞j𝐞i𝐞j+𝐞i𝐞j,\phi:\mathrm{Sym}^{2}(V)\longrightarrow U_{2}\ ;\ {\bf e}_{i}{\bf e}_{j}% \longmapsto{\bf e}_{i}\otimes{\bf e}_{j}+{\bf e}_{i}\otimes{\bf e}_{j},

which we extend linearly. This map is clearly a bijection as it maps a basis to a basis thus these vector spaces are isomorphic. For any gGg\in G and v1,v2Vv_{1},v_{2}\in V we have

ϕ(π(g)𝐯1𝐯2)=ϕ(π(g)𝐯1π(g)𝐯2)\displaystyle\phi(\pi(g){\bf v}_{1}{\bf v}_{2})=\phi(\pi(g){\bf v}_{1}\pi(g){% \bf v}_{2}) =π(g)𝐯1π(g)𝐯2+π(g)𝐯2π(g)𝐯1\displaystyle=\pi(g){\bf v}_{1}\otimes\pi(g){\bf v}_{2}+\pi(g){\bf v}_{2}% \otimes\pi(g){\bf v}_{1}
=π(g)(𝐯1𝐯2+𝐯2𝐯1)=π(g)ϕ(𝐯i𝐯j),\displaystyle=\pi(g)({\bf v}_{1}\otimes{\bf v}_{2}+{\bf v}_{2}\otimes{\bf v}_{% 1})=\pi(g)\phi({\bf v}_{i}{\bf v}_{j}),

so ϕ\phi is a GG-homomorphism and they are isomorphic representations.

In a very similar fashion we define the map

ψ:2(V)U1;𝐞i𝐞j𝐞i𝐞j𝐞i𝐞j,\psi:\textstyle\bigwedge^{2}(V)\longrightarrow U_{1}\ ;\ {\bf e}_{i}\wedge{\bf e% }_{j}\longmapsto{\bf e}_{i}\otimes{\bf e}_{j}-{\bf e}_{i}\otimes{\bf e}_{j},

and show that the representations are isomorphic (omitted). Thus

(ππ,VV)(ππ,U1)(ππ,U2)(Sym2π,Sym2(V))(ππ,2V).(\pi\otimes\pi,V\otimes V)\cong(\pi\otimes\pi,U_{1})\oplus(\pi\otimes\pi,U_{2}% )\cong\big{(}\mathrm{Sym}^{2}\pi,\mathrm{Sym}^{2}(V)\big{)}\oplus\big{(}\pi% \wedge\pi,\textstyle\bigwedge^{2}V\big{)}.

.

Problem 73. Let (π,2)(\pi,{\mathbb{C}}^{2}) be a representation of a group GG. Suppose that the matrix of π(g)\pi(g) with respect to the basis {b1,b2}\{b_{1},b_{2}\} is given by

π(g)=(abcd).\pi(g)=\left(\begin{matrix}a&b\\ c&d\end{matrix}\right).
  1. (a)

    Compute the matrix of Sym2π(g)\mathrm{Sym}^{2}\pi(g) with respect to the basis {b12,b1b2,b22}\{b_{1}^{2},b_{1}b_{2},b_{2}^{2}\}.

  2. (b)

    Compute the matrix of ππ(g)\pi\wedge\pi(g) with respect to the basis {b1b2}\{b_{1}\wedge b_{2}\}.

Solution:

  1. (a)

    We compute

    Sym2π(g)b12=(π(g)b1)(π(g)b1)=(ab1+cb2)(ab1+cb2)=a2b12+2acb1b2+c2b22\displaystyle\mathrm{Sym}^{2}\pi(g)b_{1}^{2}=(\pi(g)b_{1})(\pi(g)b_{1})=(ab_{1% }+cb_{2})(ab_{1}+cb_{2})=a^{2}b_{1}^{2}+2acb_{1}b_{2}+c^{2}b_{2}^{2}
    Sym2π(g)b1b2=(π(g)b1)(π(g)b2)=(ab1+cb2)(bb1+db2)=abb12+(ad+bc)b1b2+cdb22\displaystyle\mathrm{Sym}^{2}\pi(g)b_{1}b_{2}=(\pi(g)b_{1})(\pi(g)b_{2})=(ab_{% 1}+cb_{2})(bb_{1}+db_{2})=abb_{1}^{2}+(ad+bc)b_{1}b_{2}+cdb_{2}^{2}
    Sym2π(g)b22=(π(g)b2)(π(g)b2)=(bb1+db2)(bb1+db2)=b2b12+2bdb1b2+d2b22,\displaystyle\mathrm{Sym}^{2}\pi(g)b_{2}^{2}=(\pi(g)b_{2})(\pi(g)b_{2})=(bb_{1% }+db_{2})(bb_{1}+db_{2})=b^{2}b_{1}^{2}+2bdb_{1}b_{2}+d^{2}b_{2}^{2},

    so the matrix of Sym2π(g)\mathrm{Sym}^{2}\pi(g) is

    (a2abb22acad+bc2bdc2cdd2).\left(\begin{matrix}a^{2}&ab&b^{2}\\ 2ac&ad+bc&2bd\\ c^{2}&cd&d^{2}\end{matrix}\right).
  2. (b)

    Similarly to (a),

    ππ(g)(b1b2)=(π(g)b1)(π(g)b2)=(ab1+cb2)(bb1+db2)=ab(b1b1)+ad(b1b2)+cb(b2b1)+cd(b2b2).\pi\wedge\pi(g)(b_{1}\wedge b_{2})=(\pi(g)b_{1})\wedge(\pi(g)b_{2})=(ab_{1}+cb% _{2})\wedge(bb_{1}+db_{2})=ab(b_{1}\wedge b_{1})+ad(b_{1}\wedge b_{2})+cb(b_{2% }\wedge b_{1})+cd(b_{2}\wedge b_{2}).

    Now, b1b1=b2b2=0b_{1}\wedge b_{1}=b_{2}\wedge b_{2}=0, and b2b1=b1b2b_{2}\wedge b_{1}=-b_{1}\wedge b_{2}, hence

    ππ(g)(b1b2)=(adbc)(b1b2).\pi\wedge\pi(g)(b_{1}\wedge b_{2})=(ad-bc)(b_{1}\wedge b_{2}).

Problem 74. Let (σ,V)(\sigma,V) be an irreducible five-dimensional representation of S5S_{5}. Decompose (Sym2σ,Sym2V)(\operatorname{Sym}^{2}\sigma,\operatorname{Sym}^{2}V) and (σσ,Λ2V)(\sigma\wedge\sigma,\Lambda^{2}V) into irreducible representations.

Solution: The character table of S5S_{5} reads (see Lecture 14)

size: 1 10 15 20 20 30 24
ee (1 2)(1\;2) (1 2)(3 4)(1\;2)(3\;4) (1 2 3)(1\;2\;3) (1 2 3)(4 5)(1\;2\;3)(4\;5) (1 2 3 4)(1\;2\;3\;4) (1 2 3 4 5)(1\;2\;3\;4\;5)
(Id,)({\,\mathrm{Id}},\mathbb{C}) 1 1 1 1 1 1 1
(sgn,)(\mathrm{sgn},{\mathbb{C}}) 1 -1 1 1 -1 -1 1
(π,W0)(\pi,W_{0}) 4 22 0 1 -1 0 -1
(sgnπ,W0)(\mathrm{sgn}\,\pi,W_{0}) 4 -2 0 1 1 0 -1
(ρ,V)(\rho,V) 5 1 1 -1 1 -1 0
(sgnρ,V)(\mathrm{sgn}\rho,V) 5 -1 1 -1 -1 1 0
(ππ,2W0)(\pi\wedge\pi,\bigwedge^{2}W_{0}) 6 0 -2 0 0 0 1

The formulas for the character of Sym2σ\mathrm{Sym}^{2}\sigma and σσ\sigma\wedge\sigma are given by 12(χσ(g)2±χσ(g2))\frac{1}{2}\big{(}\chi_{\sigma}(g)^{2}\pm\chi_{\sigma}(g^{2})\big{)}, so we need to compute χσ(g2)\chi_{\sigma}(g^{2}) for gg in each of the conjugacy classes of S5S_{5}:

χσ(e2)=5\displaystyle\chi_{\sigma}(e^{2})=5
χσ((1 2)2)=χσ(e)=5\displaystyle\chi_{\sigma}\big{(}(1\;2)^{2}\big{)}=\chi_{\sigma}\big{(}e\big{)% }=5
χσ((1 2)2(3 4)2)=χσ(e)=5\displaystyle\chi_{\sigma}\big{(}(1\;2)^{2}(3\;4)^{2}\big{)}=\chi_{\sigma}\big% {(}e\big{)}=5
χσ((1 2 3)2)=χσ((1 3 2))=χσ((1 2 3))=1\displaystyle\chi_{\sigma}\big{(}(1\;2\;3)^{2}\big{)}=\chi_{\sigma}\big{(}(1\;% 3\;2)\big{)}=\chi_{\sigma}\big{(}(1\;2\;3)\big{)}=-1
χσ((1 2 3)2(4 5)2)=χσ((1 2 3))=1\displaystyle\chi_{\sigma}\big{(}(1\;2\;3)^{2}(4\;5)^{2}\big{)}=\chi_{\sigma}% \big{(}(1\;2\;3)\big{)}=-1
χσ((1 2 3 4)2)=χσ((1 3)(2 4))=1\displaystyle\chi_{\sigma}\big{(}(1\;2\;3\;4)^{2}\big{)}=\chi_{\sigma}\big{(}(% 1\;3)(2\;4)\big{)}=1
χσ((1 2 3 4 5)2)=χσ((1 3 5 2 4))=0;\displaystyle\chi_{\sigma}\big{(}(1\;2\;3\;4\;5)^{2}\big{)}=\chi_{\sigma}\big{% (}(1\;3\;5\;2\;4)\big{)}=0;

these formulas hold for both σ=ρ\sigma=\rho and σ=sgnρ\sigma=\mathrm{sgn}\rho. Similarly, χρ2=χsgnρ2\chi_{\rho}^{2}=\chi_{\mathrm{sgn}\rho}^{2}, so WLOG we may assume that σ=ρ\sigma=\rho. Thus, using the above to compute the characters:

ee (1 2)(1\;2) (1 2)(3 4)(1\;2)(3\;4) (1 2 3)(1\;2\;3) (1 2 3)(4 5)(1\;2\;3)(4\;5) (1 2 3 4)(1\;2\;3\;4) (1 2 3 4 5)(1\;2\;3\;4\;5)
ρρ\rho\otimes\rho 25 1 1 1 1 1 0
Sym2ρ\mathrm{Sym}^{2}\rho 15 3 3 0 0 1 0
ρρ\rho\wedge\rho 10 -2 -2 1 1 0 0

Starting with Sym2ρ\mathrm{Sym}^{2}\rho, we can see directly from the character table that χSym2ρ,χId>0\langle\chi_{\mathrm{Sym}^{2}\rho},\chi_{{\,\mathrm{Id}}}\rangle>0; to make this more precise, we compute

χSym2ρ,χId=1120(115+103+153+200+200+301+240)=1.\langle\chi_{\mathrm{Sym}^{2}\rho},\chi_{{\,\mathrm{Id}}}\rangle=\frac{1}{120}% \big{(}1\cdot 15+10\cdot 3+15\cdot 3+20\cdot 0+20\cdot 0+30\cdot 1+24\cdot 0% \big{)}=1.

We can also see that (π,W0)(\pi,W_{0}) is a subrepresentation:

χSym2ρ,χπ=1120(1154+1032+1530+2001+200(1)+3010+240(1))=1,\langle\chi_{\mathrm{Sym}^{2}\rho},\chi_{\pi}\rangle=\frac{1}{120}\big{(}1% \cdot 15\cdot 4+10\cdot 3\cdot 2+15\cdot 3\cdot 0+20\cdot 0\cdot 1+20\cdot 0% \cdot(-1)+30\cdot 1\cdot 0+24\cdot 0\cdot(-1)\big{)}=1,

and (ρ,V)(\rho,V):

χSym2ρ,χρ=1120(1155+1031+1531+200(1)+200(1)+301(1)+2400)=1,\langle\chi_{\mathrm{Sym}^{2}\rho},\chi_{\rho}\rangle=\frac{1}{120}\big{(}1% \cdot 15\cdot 5+10\cdot 3\cdot 1+15\cdot 3\cdot 1+20\cdot 0\cdot(-1)+20\cdot 0% \cdot(1)+30\cdot 1\cdot(-1)+24\cdot 0\cdot 0\big{)}=1,

and sgnρ\mathrm{sgn}\rho:

χSym2ρ,χsgnρ=1120(1155+103(1)+1531+200(1)+200(1)+3011+2400)=1.\langle\chi_{\mathrm{Sym}^{2}\rho},\chi_{\mathrm{sgn}\rho}\rangle=\frac{1}{120% }\big{(}1\cdot 15\cdot 5+10\cdot 3\cdot(-1)+15\cdot 3\cdot 1+20\cdot 0\cdot(-1% )+20\cdot 0\cdot(-1)+30\cdot 1\cdot 1+24\cdot 0\cdot 0\big{)}=1.

The sum of the dimensions of these four irreducible representations is 15, hence

(Sym2ρ,Sym2V)(Id,)(π,W0)(ρ,V)(sgnρ,V).(\mathrm{Sym}^{2}\rho,\mathrm{Sym}^{2}V)\cong({\,\mathrm{Id}},{\mathbb{C}})% \oplus(\pi,W_{0})\oplus(\rho,V)\oplus(\mathrm{sgn}\rho,V).

In a similar fashion, one can see directly from the character table that χρρ,χππ>0\langle\chi_{\rho\wedge\rho},\chi_{\pi\wedge\pi}\rangle>0 and χρρ,χsgnπ>0\langle\chi_{\rho\wedge\rho},\chi_{\mathrm{sgn}\pi}\rangle>0, and a more explicit computation verifies this:

χρρ,χππ=1120(1106+10(2)0+15(2)(2)+2010+2010+3000+2401)=1,\langle\chi_{\rho\wedge\rho},\chi_{\pi\wedge\pi}\rangle=\frac{1}{120}\big{(}1% \cdot 10\cdot 6+10\cdot(-2)\cdot 0+15\cdot(-2)\cdot(-2)+20\cdot 1\cdot 0+20% \cdot 1\cdot 0+30\cdot 0\cdot 0+24\cdot 0\cdot 1\big{)}=1,
χρρ,χsgnπ=1120(1104+10(2)(2)+15(2)0+2011+2011+3000+240(1))=1,\langle\chi_{\rho\wedge\rho},\chi_{\mathrm{sgn}\pi}\rangle=\frac{1}{120}\big{(% }1\cdot 10\cdot 4+10\cdot(-2)\cdot(-2)+15\cdot(-2)\cdot 0+20\cdot 1\cdot 1+20% \cdot 1\cdot 1+30\cdot 0\cdot 0+24\cdot 0\cdot(-1)\big{)}=1,

hence

(ρρ,2V)(sgnπ,W0)(ππ,2W0).(\rho\wedge\rho,\bigwedge{\!}^{2}V)\cong(\mathrm{sgn}\pi,W_{0})\oplus(\pi% \wedge\pi,\bigwedge{\!}^{2}W_{0}).

Problem 75. Let (π,4)(\pi,{\mathbb{C}}^{4}) be the permutation representation of S4S_{4}, and let (ρ,V)(\rho,V) be the two-dimensional irreducible representation of S4S_{4}.

  1. (a)

    Show that (Sym2π,Sym24)(\operatorname{Sym}^{2}\pi,\operatorname{Sym}^{2}{\mathbb{C}}^{4}) has a unique subrepresentation isomorphic to (ρ,V)(\rho,V).

  2. (b)

    Use the ρ\rho-isotopic projector to find that subrepresentation.

Solution:

  1. (1)

    The character of (π,4)(\pi,{\mathbb{C}}^{4}) is given by

    size: 1 6 3 8 6
    ee (1 2)(1\;2) (1 2)(3 4)(1\;2)(3\;4) (1 2 3)(1\;2\;3) (1 2 3 4)(1\;2\;3\;4)
    (π,4)(\pi,{\mathbb{C}}^{4}) 4 2 0 1 0

    Using the formula χSym2π(g)=12(χπ(g)2+χπ(g2))\chi_{\mathrm{Sym}^{2}\pi}(g)=\frac{1}{2}\big{(}\chi_{\pi}(g)^{2}+\chi_{\pi}(g% ^{2})\big{)}, hence

    size: 1 6 3 8 6
    ee (1 2)(1\;2) (1 2)(3 4)(1\;2)(3\;4) (1 2 3)(1\;2\;3) (1 2 3 4)(1\;2\;3\;4)
    Sym2π\mathrm{Sym}^{2}\pi 10 4 2 1 0

    The character of (ρ,V)(\rho,V) is given by

    size: 1 6 3 8 6
    ee (1 2)(1\;2) (1 2)(3 4)(1\;2)(3\;4) (1 2 3)(1\;2\;3) (1 2 3 4)(1\;2\;3\;4)
    (ρ,V)(\rho,V) 2 0 2 -1 0

    Hence, computing the inner product of χSym2π\chi_{\mathrm{Sym}^{2}\pi} with χρ\chi_{\rho}:

    χSym2π,χρS4=124(1102+640+322+81(1)+600)=1,\langle\chi_{\mathrm{Sym}^{2}\pi},\chi_{\rho}\rangle_{S_{4}}=\frac{1}{24}\big{% (}1\cdot 10\cdot 2+6\cdot 4\cdot 0+3\cdot 2\cdot 2+8\cdot 1\cdot(-1)+6\cdot 0% \cdot 0\big{)}=1,

    there is therefore exactly one subrepresentation of (Sym2π,Sym24)(\operatorname{Sym}^{2}\pi,\operatorname{Sym}^{2}{\mathbb{C}}^{4}) that is isomorphic to (ρ,V)(\rho,V).

  2. (2)

    The ρ\rho-isotopic projector is given by

    Sym2πρ=224gS4χρ(g)¯Sym2π(g)=112(2Id+2(g[(1 2)(3 4)]Sym2π(g))(g[(1 2 3)]Sym2π(g))).\mathrm{Sym}^{2}\pi_{\rho}=\frac{2}{24}\sum_{g\in S_{4}}\overline{\chi_{\rho}(% g)}\mathrm{Sym}^{2}\pi(g)=\frac{1}{12}\left(2{\,\mathrm{Id}}+2\left(\sum_{g\in% [(1\;2)(3\;4)]}\mathrm{Sym}^{2}\pi(g)\right)-\left(\sum_{g\in[(1\;2\;3)]}% \mathrm{Sym}^{2}\pi(g)\right)\right).

    Since VV has dimension two, the subrepresentation we are searching for will be spanned by any two linearly independent vectors in Sym2πρ(Sym24)\mathrm{Sym}^{2}\pi_{\rho}(\mathrm{Sym}^{2}{\mathbb{C}}^{4}). We will succesively apply Sym2πρ\mathrm{Sym}^{2}\pi_{\rho} to the standard basis vectors of Sym24\mathrm{Sym}^{2}{\mathbb{C}}^{4} until we have two non co-linear vectors;

    Sym2πρe12=112(2e12+2(e22+e32+e42)(e22+e32+e22+e42+e32+e42+e12+e12))=0\mathrm{Sym}^{2}\pi_{\rho}e_{1}^{2}=\frac{1}{12}\left(2e_{1}^{2}+2\big{(}e_{2}% ^{2}+e_{3}^{2}+e_{4}^{2}\big{)}-\big{(}e_{2}^{2}+e_{3}^{2}+e_{2}^{2}+e_{4}^{2}% +e_{3}^{2}+e_{4}^{2}+e_{1}^{2}+e_{1}^{2}\big{)}\right)=0

    (recall that

    [(1 2)(3 4)]={(1 2)(3 4),(1 3)(2 4),(1 4)(2 3)},[(1\;2)(3\;4)]=\{(1\;2)(3\;4),\;(1\;3)(2\;4),\;(1\;4)(2\;3)\},

    and

    [(1 2 3)]={(1 2 3),(1 3 2),(1 2 4),(1 4 2),(1 3 4),(1 4 3),(2 3 4),(2 4 3)},[(1\;2\;3)]=\{(1\;2\;3),\;(1\;3\;2),\;(1\;2\;4),\;(1\;4\;2),\;(1\;3\;4),\;(1\;% 4\;3),\;(2\;3\;4),\;(2\;4\;3)\},

    the sums are carried out in the order listed here.) We proceed with other basis vectors:

    Sym2πρe1e2\displaystyle\mathrm{Sym}^{2}\pi_{\rho}e_{1}e_{2} =112(2e1e2+2(e1e2+e3e4+e3e4)(e2e3+e1e3+e2e4+e1e4+e2e3+e2e4+e1e3+e1e4))\displaystyle=\frac{1}{12}\left(2e_{1}e_{2}+2\big{(}e_{1}e_{2}+e_{3}e_{4}+e_{3% }e_{4}\big{)}-\big{(}e_{2}e_{3}+e_{1}e_{3}+e_{2}e_{4}+e_{1}e_{4}+e_{2}e_{3}+e_% {2}e_{4}+e_{1}e_{3}+e_{1}e_{4}\big{)}\right)
    =112(4e1e2+4e3e42e1e32e1e42e2e32e2e4)=b1.\displaystyle=\frac{1}{12}\left(4e_{1}e_{2}+4e_{3}e_{4}-2e_{1}e_{3}-2e_{1}e_{4% }-2e_{2}e_{3}-2e_{2}e_{4}\right)=b_{1}.
    Sym2πρe1e3\displaystyle\mathrm{Sym}^{2}\pi_{\rho}e_{1}e_{3} =112(2e1e3+2(e2e4+e1e3+e2e4)(e1e2+e2e3+e2e3+e3e4+e3e4+e1e4+e1e4+e1e2))\displaystyle=\frac{1}{12}\left(2e_{1}e_{3}+2\big{(}e_{2}e_{4}+e_{1}e_{3}+e_{2% }e_{4}\big{)}-\big{(}e_{1}e_{2}+e_{2}e_{3}+e_{2}e_{3}+e_{3}e_{4}+e_{3}e_{4}+e_% {1}e_{4}+e_{1}e_{4}+e_{1}e_{2}\big{)}\right)
    =112(4e1e3+4e2e42e1e22e1e42e2e32e3e4)=b2;\displaystyle=\frac{1}{12}\left(4e_{1}e_{3}+4e_{2}e_{4}-2e_{1}e_{2}-2e_{1}e_{4% }-2e_{2}e_{3}-2e_{3}e_{4}\right)=b_{2};

    this is not in b1{\mathbb{C}}b_{1}, hence (Sym2π,span{b1,b2})(\mathrm{Sym}^{2}\pi,\mathrm{span}\{b_{1},b_{2}\}) is the subrepresentation we are looking for.

Problem 76. A group GG of order 168168 has conjugacy classes C1C_{1}, C2C_{2}, C3C_{3}, C4C_{4}, C7AC_{7A} and C7BC_{7B}, where each conjugacy class is labelled by the order of any element in that class (so, for example, any element of C7AC_{7A} or C7BC_{7B} has order 77). The following shows one of the rows of the character table of GG.

size:12156422424classC1C2C3C4C7AC7B(π1,V)31011+72172\begin{array}[]{c|cccccc}\text{size:}&1&21&56&42&24&24\\ \text{class}&C_{1}&C_{2}&C_{3}&C_{4}&C_{7A}&C_{7B}\\ \hline\cr(\pi_{1},V)&3&-1&0&1&\frac{-1+\sqrt{-7}}{2}&\frac{-1-\sqrt{-7}}{2}\\ \end{array}
  1. (a)

    Show that if xx is an element of C7AC_{7A} or C7BC_{7B}, then xx is conjugate to x2x^{2}.

  2. (b)

    Find the character table of GG.

Solution:

  1. (1)

    Since xx has order 77, so must x2,x3,,x6x^{2},x^{3},\ldots,x^{6}. hence each of these is in either C7AC_{7A} or C7BC_{7_{B}}. In particular, the pigeonhole principle gives that at least two of {x,x2,x3}\{x,x^{2},x^{3}\} are conjugate to each other. We consider the three possibilities: (i) x,x2x,x^{2} are conjugate: we are done, as this is what we are trying to show (ii) x,x3x,x^{3} are conjugate. Writing x3=axa1x^{3}=axa^{-1}, we have x2=x9=x3x3x3=ax3a1=a2xa2x^{2}=x^{9}=x^{3}\cdot x^{3}\cdot x^{3}=ax^{3}a^{-1}=a^{2}xa^{-2}, hence xx and x2x^{2} are also conjugate. (iii) x2x^{2} and x3x^{3} are conjugate. Writing x3=bx2b1x^{3}=bx^{2}b^{-1}, we then have x=x15=(x3)5=(bx2b1)5=b5x10b5=b5x3b5=b6x2b6x=x^{15}=(x^{3})^{5}=(bx^{2}b^{-1})^{5}=b^{5}x^{10}b^{-5}=b^{5}x^{3}b^{-5}=b^{% 6}x^{2}b^{-6}, i.e. xx and x2x^{2} are also conjugate.

  2. (2)

    The trivial representation is of course irrecducible. By Proposition 31 (Lecture 12, cf. Problem 54), (π1,V)(\pi_{1}^{*},V^{*}) is also irreducible, with character χπ1¯\overline{\chi_{\pi_{1}}}. We have now found three rows of the character table:

    size:12156422424classC1C2C3C4C7AC7B(Id,)111111(π1,V)31011+72172(π1,V)31011721+72\begin{array}[]{|c|c|c|c|c|c|c|}\hline\cr\text{size:}&1&21&56&42&24&24\\ \hline\cr\text{class}&C_{1}&C_{2}&C_{3}&C_{4}&C_{7A}&C_{7B}\\ \hline\cr({\,\mathrm{Id}},{\mathbb{C}})&1&1&1&1&1&1\\ \hline\cr(\pi_{1},V)&3&-1&0&1&\frac{-1+\sqrt{-7}}{2}&\frac{-1-\sqrt{-7}}{2}\\ \hline\cr(\pi_{1}^{*},V^{*})&3&-1&0&1&\frac{-1-\sqrt{-7}}{2}&\frac{-1+\sqrt{-7% }}{2}\\ \end{array}

    We now consider the representations (π1π1,2V)(\pi_{1}\wedge\pi_{1},\bigwedge^{2}V) and Sym2π,Sym2V\mathrm{Sym}^{2}\pi,\mathrm{Sym}^{2}V. Recall that

    χπ1π1(g)=12(χπ1(g)2χπ1(g2)),χSym2π(g)=12(χπ1(g)2+χπ1(g2))\chi_{\pi_{1}\wedge\pi_{1}}(g)=\frac{1}{2}(\chi_{\pi_{1}}(g)^{2}-\chi_{\pi_{1}% }(g^{2})),\qquad\chi_{\mathrm{Sym}^{2}\pi}(g)=\frac{1}{2}(\chi_{\pi_{1}}(g)^{2% }+\chi_{\pi_{1}}(g^{2}))

    If xx has order two, then x2=ex^{2}=e, if xx has order three, so does x2x^{2}, if xx has order four, then x2x^{2} has order two, and from part (1), we know that if xx is in C7AC_{7A} or C7BC_{7B}, then so is x2x^{2}. This lets us compute:

    size:12156422424classC1C2C3C4C7AC7Bπ1π131011721+72Sym2π1620011\begin{array}[]{c|cccccc}\text{size:}&1&21&56&42&24&24\\ \text{class}&C_{1}&C_{2}&C_{3}&C_{4}&C_{7A}&C_{7B}\\ \hline\cr\pi_{1}\wedge\pi_{1}&3&-1&0&1&\frac{-1-\sqrt{-7}}{2}&\frac{-1+\sqrt{-% 7}}{2}\\ \hline\cr\mathrm{Sym}^{2}\pi_{1}&6&2&0&0&-1&-1\\ \end{array}

    From this we see that π1π1\pi_{1}\wedge\pi_{1} is simply π1\pi_{1}^{*}, and that Sym2π1\mathrm{Sym}^{2}\pi_{1} is a new irreducible representation.

    We are now missing two irreducible representations. Recalling the difference of squares rule, we consider the representation (π1,V)(π1,V)(\pi_{1},V)\otimes(\pi_{1}^{*},V^{*}), which has character χπ1χπ1¯\chi_{\pi_{1}}\overline{\chi_{\pi_{1}}}:

    size:12156422424classC1C2C3C4C7AC7Bπ1π1910122\begin{array}[]{c|cccccc}\text{size:}&1&21&56&42&24&24\\ \text{class}&C_{1}&C_{2}&C_{3}&C_{4}&C_{7A}&C_{7B}\\ \hline\cr\pi_{1}\otimes\pi_{1}^{*}&9&1&0&1&2&2\end{array}

    Since all the entries in the table are non-negative and the representation has positive dimension, we must have χπ1π1,χId>0\langle\chi_{\pi_{1}\otimes\pi_{1}^{*}},\chi_{{\,\mathrm{Id}}}\rangle>0, so this is not irreducible. However, when we compute this inner product, we get

    χπ1π1,χId=1168(9+21+42+242+242)=1,\langle\chi_{\pi_{1}\otimes\pi_{1}^{*}},\chi_{{\,\mathrm{Id}}}\rangle=\frac{1}% {168}\left(9+21+42+24\cdot 2+24\cdot 2\right)=1,

    hence

    (π1,V)(π1,V)(Id,)(ρ,W).(\pi_{1},V)\otimes(\pi_{1}^{*},V^{*})\cong({\,\mathrm{Id}},{\mathbb{C}})\oplus% (\rho,W).

    Subtracting off the trivial character from χπ1π1\chi_{\pi_{1}\otimes\pi_{1}^{*}} gives

    size:12156422424classC1C2C3C4C7AC7B(ρ,W)801011\begin{array}[]{c|cccccc}\text{size:}&1&21&56&42&24&24\\ \text{class}&C_{1}&C_{2}&C_{3}&C_{4}&C_{7A}&C_{7B}\\ \hline\cr(\rho,W)&8&0&-1&0&1&1\end{array}

    Computing the the norm of χρ\chi_{\rho} gives

    χρ2=1168(164+561+241+241)=1,\|\chi_{\rho}\|^{2}=\frac{1}{168}\big{(}1\cdot 64+56\cdot 1+24\cdot 1+24\cdot 1% \big{)}=1,

    which shows that ρ\rho is irreducible! We are now missing just one irreducible representation:

    size:12156422424classC1C2C3C4C7AC7B(Id,)111111(π1,V)31011+72172(π1,V)31011721+72Sym2π1620011(ρ,W)801011\begin{array}[]{|c|c|c|c|c|c|c|}\hline\cr\text{size:}&1&21&56&42&24&24\\ \hline\cr\text{class}&C_{1}&C_{2}&C_{3}&C_{4}&C_{7A}&C_{7B}\\ \hline\cr({\,\mathrm{Id}},{\mathbb{C}})&1&1&1&1&1&1\\ \hline\cr(\pi_{1},V)&3&-1&0&1&\frac{-1+\sqrt{-7}}{2}&\frac{-1-\sqrt{-7}}{2}\\ \hline\cr(\pi_{1}^{*},V^{*})&3&-1&0&1&\frac{-1-\sqrt{-7}}{2}&\frac{-1+\sqrt{-7% }}{2}\\ \hline\cr\mathrm{Sym}^{2}\pi_{1}&6&2&0&0&-1&-1\\ \hline\cr(\rho,W)&8&0&-1&0&1&1\\ \end{array}

    We then immediately know that this representation (σ,U)(\sigma,U) must have dimension 77. Column orthogonality then gives

    size:12156422424classC1C2C3C4C7AC7B(σ,U)711100\begin{array}[]{c|cccccc}\text{size:}&1&21&56&42&24&24\\ \text{class}&C_{1}&C_{2}&C_{3}&C_{4}&C_{7A}&C_{7B}\\ \hline\cr(\sigma,U)&7&-1&1&-1&0&0\end{array}