Skip to content

Commit

Permalink
Updated MA4204
Browse files Browse the repository at this point in the history
  • Loading branch information
sahasatvik committed Jan 19, 2024
1 parent ec888b1 commit ac7c68f
Show file tree
Hide file tree
Showing 2 changed files with 147 additions and 8 deletions.
Binary file modified MA4204/notes.pdf
Binary file not shown.
155 changes: 147 additions & 8 deletions MA4204/notes.typ
Original file line number Diff line number Diff line change
Expand Up @@ -152,18 +152,18 @@ the presence of $sigma$ is clear from context.
]
#example[
The only one dimensional representation of $S_3$ in $C^*$ is the sign
The only one dimensional representation of $S_3$ in $CC^times$ is the sign
homomorphism. To see this, consider an arbitrary homomorphism $sigma: S_3 ->
C^*$. Note that $ker(sigma)$ must be a normal subgroup of $S_3$, hence must
be one of ${e}, A_3, S_3$. The third option yields the trivial
representation $sigma = id_(C^*)$, and the first option gives the
contradiction $S_3 tilde.equiv im(sigma) subset C^*$ (the right side is
CC^times$. Note that $ker(sigma)$ must be a normal subgroup of $S_3$, hence
must be one of ${e}, A_3, S_3$. The third option yields the trivial
representation $sigma = id_(CC^times)$, and the first option gives the
contradiction $S_3 tilde.equiv im(sigma) subset CC^times$ (the right side is
abelian while the left is not). This leaves $ker(sigma) = A_3$, i.e.
$sigma(g) = 1$ for all even permutations $g in S_3$. The remaining elements
of $S_3$ (the odd permutations) must be sent to $-1$, since for any odd
permutation $h in S_3$, the permutation $h^2$ is even, so $sigma(h)^2 =
sigma(h^2) = 1$. The result is precisely the sign homomorphism $
sigma: S_3 -> C^*, wide
sigma: S_3 -> CC^times, wide
g mapsto cases(
+1 &"if" g in A_3,\
-1 & "if" g in.not A_3.
Expand Down Expand Up @@ -497,7 +497,7 @@ $
for $i = k$, $j = ell$; summing over $1 <= i <= n$, $1 <= j <= m$, we
have $
0 = sum_(i j) sum_(g in G) [sigma'(g)]_(i i) [sigma(g)^(-1)]_(j j)
&= sum_(g in G) ((sum_i [sigma(g)]_(i i)) (sum_j [sigma(g)^(-1)]_(j j))) \
&= sum_(g in G) ((sum_i [sigma'(g)]_(i i)) (sum_j [sigma(g)^(-1)]_(j j))) \
&= sum_(g in G) chi_V (g) chi_(V') (g^(-1)) \
&= "ord"(G) ip(chi_V, chi_V').
$
Expand All @@ -510,14 +510,18 @@ $
"ord"(G) delta_(i j) delta_(k l)
$ for all $1 <= i, j, k, ell <= n$. Following a similar process as before, $
"ord"(G) ip(chi_V, chi_V')
&= sum_(g in G) ((sum_i [sigma(g)]_(i i)) (sum_j [sigma(g)^(-1)]_(j j))) \
&= sum_(g in G) ((sum_i [sigma'(g)]_(i i)) (sum_j [sigma(g)^(-1)]_(j j))) \
&= sum_(i j) sum_(g in G) [sigma'(g)]_(i i) [sigma(g)^(-1)]_(j j) \
&= sum_(i j) 1/n "ord"(G) delta_(i j) \
&= "ord"(G)
$
This completes the proof.
]
With this, the characters of irreducible representations form an orthonormal
subset of class functions on $G$. To check whether a representation $V$ is
irreducible or not, it is enough to verify that $ip(chi_V, chi_V) = 1$.
#corollary[
The number of irreducible representations of $G$ (up to isomorphism) is at
most the number of conjugacy classes of $G$.
Expand Down Expand Up @@ -550,3 +554,138 @@ $
Observe that the rows of this table are orthogonal; indeed, so are the
columns!
] <ex_S3_table>
Given any representation $V$ of $G$, we can use Maschke's Theorem
(@thm_maschke) to decompose it as a direct sum of (non-isomorphic) irreducible
representations $V_1, ..., V_k$, with multiplicities $m_1, ..., m_k$. By
representing the elements of $G$ as matrices in block diagonal form, we can
derive the following result.
#lemma[
Let $V_1, ..., V_k$ be irreducible representations of $G$, and let $
V tilde.equiv m_1 V_1 plus.circle dots.c plus.circle m_k V_k.
$ Then, $
chi_V tilde.equiv m_1 chi_V_1 + dots.c + m_k chi_V_k.
$ The multiplicities can be recovered as $m_i = ip(chi_V, chi_V_i)$.
]
This immediately tells us that $chi_V = chi_V'$ if and only if $V tilde.equiv
V'$. Furthermore, we have the relation $
ip(chi_V, chi_V) = sum_i m_i^2.
$
#example[
Let $S_4$ act on $CC^4$ by permuting the basis vectors ${e_1, e_2, e_3,
e_4}$, and let $(sigma, V)$ denote the induced (regular) representation. Note that
each matrix $sigma(g)$ is a permutation, hence its trace $chi_V(g)$ is
precisely the number of elements of ${1, 2, 3, 4}$ fixed by the action of
$g$. With this, we can compute $chi_V$ for each conjugacy class (identified
by its cycle type) as follows.
#align(
center,
tablex(
columns: 6,
inset: (x: 1em, y: 0.6em),
stroke: table-stroke,
align: center + horizon,
auto-hlines: false,
auto-vlines: false,
[*$bold(S_4)$*], vlinex(), $e$, $(a b) times 6$, $(a b)(c d) times 3$, $(a b c) times 8$, $(a b c d) times 6$,
hlinex(),
[Trivial], 1, 1, 1, 1, 1,
[Sign], 1, -1, 1, 1, -1,
$V$, 4, 2, 0, 1, 0,
)
)
Compute $
ip(chi_V, chi_V) = 1/24 (4^2 + 6 dot.c 2^2 + 8 dot.c 1^2) = 2,
$ whence $V$ is not irreducible. Indeed, we know that $W_1 = "span"{e_1 +
e_2 + e_3 + e_4}$ is a trivial subrepresentation of $V$ of dimension $1$.
Furthermore, $2 = 1^2 + 1^2$ is the only way of writing $2$ as a sum of
squares of integers, so $V$ must decompose into precisely two irreducible
subrepresentations with both multiplicities $1$. This means that $V
tilde.equiv W_1 plus.circle W_3$ for some irreducible representation $W_3$
of dimension 3. Using $chi_V = chi_W_1 + chi_W_3$, we can compute the
character $chi_W_3$ and obtain the following.
#align(
center,
tablex(
columns: 6,
inset: (x: 1em, y: 0.6em),
stroke: table-stroke,
align: center + horizon,
auto-hlines: false,
auto-vlines: false,
[*$bold(S_4)$*], vlinex(), $e$, $(a b) times 6$, $(a b)(c d) times 3$, $(a b c) times 8$, $(a b c d) times 6$,
hlinex(),
$W_3$, 3, 1, -1, 0, -1,
)
)
Next, we move on to a different representation of $S_4$: consider all
subsets of size 2 of ${1, 2, 3, 4}$ (of which there are 6), and consider the
action on this collection induced by the permutations on the set ${1, 2, 3,
4}$. Let $(tau, V')$ denote the induced representation. Again, $chi_V'(g)$
is the number of 2-subsets fixed by the action of $g$. For instance, an
element $(a b) in S_4$ will only fix 2-subsets ${a, b}, {c, d}$, while an
element $(a b c) in S_4$ fixes no 2-subset. With this, we have the
following.
#align(
center,
tablex(
columns: 6,
inset: (x: 1em, y: 0.6em),
stroke: table-stroke,
align: center + horizon,
auto-hlines: false,
auto-vlines: false,
[*$bold(S_4)$*], vlinex(), $e$, $(a b) times 6$, $(a b)(c d) times 3$, $(a b c) times 8$, $(a b c d) times 6$,
hlinex(),
$V'$, 6, 2, 2, 0, 0,
)
) Compute $ip(chi_V', chi_V') = 3 = 1^2 + 1^2 + 1^2$. Again, we may compute
$ip(chi_V', chi_W_1) = 1$ and $ip(chi_V', chi_W_3) = 1$, which tells us that
$V' tilde.equiv W_1 plus.circle W_3 plus.circle W_2$ for some irreducible
representation $W_2$ of dimension $2$. Using $chi_V' = chi_W_1 + chi_W_3 +
chi_W_2$, we can compute the character $chi_W_2$.
#align(
center,
tablex(
columns: 6,
inset: (x: 1em, y: 0.6em),
stroke: table-stroke,
align: center + horizon,
auto-hlines: false,
auto-vlines: false,
[*$bold(S_4)$*], vlinex(), $e$, $(a b) times 6$, $(a b)(c d) times 3$, $(a b c) times 8$, $(a b c d) times 6$,
hlinex(),
$W_2$, 2, 0, 2, -1, 0,
)
) We now have $4$ irreducible characters of $S_4$; indeed, we may combine
$W_3$ with the sign representation to get another irreducible representation
$W'_3$, completing the character table of $S_4$.
#align(
center,
tablex(
columns: 6,
inset: (x: 1em, y: 0.6em),
stroke: table-stroke,
align: center + horizon,
auto-hlines: false,
auto-vlines: false,
[*$bold(S_4)$*], vlinex(), $e$, $(a b) times 6$, $(a b)(c d) times 3$, $(a b c) times 8$, $(a b c d) times 6$,
hlinex(),
[Trivial], 1, 1, 1, 1, 1,
[Sign], 1, -1, 1, 1, -1,
$W_2$, 2, 0, 2, -1, 0,
$W_3$, 3, 1, -1, 0, -1,
$W'_3$, 3, -1, -1, 0, 1,
)
) The last trick uses the following proposition.
]
#proposition[
Let $(sigma, V)$ and $(tau, CC^times)$ be representations of $G$. Then,
$(tau sigma, V)$ is also a representation of $G$, where $(tau sigma)(g) =
tau(g) sigma(g)$. Furthermore, $chi_(tau sigma) = chi_tau chi_sigma$.
]

0 comments on commit ac7c68f

Please sign in to comment.