1 Introduction
Throughout this paper, G will denote a finite group. Also, all the representations considered will be taken to be over the field of complex numbers. The set of all ordinary irreducible characters of G is denoted as usual by
$\operatorname {\mathrm {Irr}}(G)$
, and
$\operatorname {\mathrm {Lin}}(G)$
will denote the group of linear characters of
$G.$
There are a number of results concerning
$\operatorname {\mathrm {Irr}}(G)$
and Frobenius–Schur indicators, three of which are reviewed here. For the first, see [Reference Isaacs3, pages 49–50].
Theorem 1.1. Define
$\theta _n: G\rightarrow \mathbb {Z}_{\geq 0}$
by
$\theta _n(x) = \vert \{g\in G: g^n = x\}\vert $
for
$n\in \mathbb {N}.$
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu1.png?pub-status=live)
where
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu2.png?pub-status=live)
is the nth Frobenius–Schur indicator of
$\chi $
and
$v_n(\chi )\in \mathbb {Z}.$
The second result is a consequence of this theorem (see [Reference Isaacs3, Corollary 4.6]).
Corollary 1.2. Let G have exactly t involutions. Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu3.png?pub-status=live)
An element of G is real if it is conjugate to its inverse and
$\chi \in \operatorname {\mathrm {Irr}}(G)$
is real if
$\chi (x)\in \mathbb {R}$
for all
$x\in G.$
For the third result connecting these two concepts, see [Reference Isaacs3, Problem 6.13].
Theorem 1.3. The number of real conjugacy classes of G is equal to the number of real
$\chi \in \operatorname {\mathrm {Irr}}(G).$
The purpose of this paper is to find generalisations of these three results if irreducible projective characters of G are considered instead of ordinary ones. To generalise Theorem 1.1 it will be necessary to define the nth Frobenius–Schur indicator of an irreducible projective character of
$G.$
A number of remarks and examples were made and given in [Reference Humphreys2, pages 27–28] to show that this and Theorem 1.3 do not have a straightforward generalisation to the projective character situation, but our approach overcomes those difficulties.
In Section 2, basic facts about projective representations of G with
$2$
-cocycle
$\alpha $
will be stated. The nth Frobenius–Schur indicator of an irreducible projective character of G is then defined and interpreted for
$n = 2.$
Using this, the generalisations sought of the three results will be found in Section 3, although for the last two restricted to the case when both
$\alpha $
and its cohomology class have order
$2$
.
2 Frobenius–Schur indicators for projective characters
All of the standard facts and concepts relating to projective representations below may be found in [Reference Karpilovsky4, Reference Karpilovsky5], or (albeit to a lesser extent) [Reference Isaacs3, Ch. 11] or [Reference Haggarty and Humphreys1].
Definition 2.1. A
$2$
-cocycle of G over
$\mathbb {C}$
is a function
$\alpha : G\times G\rightarrow \mathbb {C}^*$
such that
$\alpha (1, 1) = 1$
and
$\alpha (x, y)\alpha (xy, z) = \alpha (x, yz)\alpha (y, z)$
for all x, y,
$z\in G.$
The set of all such
$2$
-cocycles of G form a group
$Z^2(G, \mathbb {C}^*)$
under multiplication. Let
$\delta : G\rightarrow \mathbb {C}^*$
be any function with
$\delta (1) = 1.$
Then
$t(\delta )(x, y) = \delta (x)\delta (y)/\delta (xy)$
for all
$x, y\in G$
is a
$2$
-cocycle of G, which is called a coboundary. Two
$2$
-cocycles
$\alpha $
and
$\beta $
are cohomologous if there exists a coboundary
$t(\delta )$
such that
$\beta = t(\delta )\alpha .$
This defines an equivalence relation on
$Z^2(G, \mathbb {C}^*)$
, and the cohomology classes
$[\alpha ]$
form a finite abelian group, called the Schur multiplier
$M(G).$
Definition 2.2. Let
$\alpha $
be a
$2$
-cocycle of
$G.$
Then
$x\in G$
is
$\alpha $
-regular if
$\alpha (x, y) = \alpha (y, x)$
for all
$y\in C_G(x).$
Let
$\beta \in [\alpha ]$
. Then
$x\in G$
is
$\alpha $
-regular if and only if it is
$\beta $
-regular. If x is
$\alpha $
-regular then so too are
$x^{-1}$
and any conjugate of
$x,$
so from the latter one may refer to the
$\alpha $
-regular conjugacy classes of
$G.$
Definition 2.3. Let
$\alpha $
be a
$2$
-cocycle of
$G.$
Then an
$\alpha $
-representation of G of dimension n is a function
$P:G\rightarrow \mathrm {GL}(n, \mathbb {C})$
such that
$P(x)P(y) = \alpha (x, y)P(xy)$
for all x,
$y\in G.$
Observe that if P is an
$\alpha $
-representation of
$G,$
then
$P(g)P(x)P(g)^{-1}\! =\! f_{\alpha }(g, x)P(gxg^{-1})$
and
$P(x)^m = p_{\alpha }(x, m)P(x^m)$
for all
$g, x\in G$
and
$m\in \mathbb {N},$
where
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu4.png?pub-status=live)
An
$\alpha $
-representation is also called a projective representation of G with
$2$
-cocycle
$\alpha $
and its trace function is its
$\alpha $
-character. Let
$\operatorname {\mathrm {Proj}}(G, \alpha )$
denote the set of all irreducible
$\alpha $
-characters of G. The relationship between
$\operatorname {\mathrm {Proj}}(G, \alpha )$
and
$\alpha $
-representations is much the same as that between
$\operatorname {\mathrm {Irr}}(G)$
and ordinary representations of G (see [Reference Karpilovsky4, page 184] for details). Next
$x\in G$
is
$\alpha $
-regular if and only if
$\xi (x)\not = 0$
for some
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha )$
and
$\vert \!\operatorname {\mathrm {Proj}}(G, \alpha )\vert $
is the number of
$\alpha $
-regular conjugacy classes of
$G.$
For
$[\beta ]\in M(G)$
there exists
$\alpha \in [\beta ]$
such that
$o(\alpha ) = o([\beta ])$
and
$\alpha $
is a class-function
$2$
-cocycle, that is, the elements of
$\operatorname {\mathrm {Proj}}(G, \alpha )$
are class functions. If
$\alpha $
is a class-function
$2$
-cocycle of G, then
$x\in G$
is
$\alpha $
-regular if and only if
$f_{\alpha }(g, x) = 1$
for all
$g\in G.$
The nth Frobenius–Schur indicator of
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha )$
can now be defined and agrees with the normal definition if
$\alpha $
is trivial.
Definition 2.4. Let
$\alpha $
be a
$2$
-cocycle of G of finite order. Then the nth Frobenius–Schur indicator
$v_n^{\alpha }(\xi )$
for
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha )$
and
$n\in \mathbb {N}$
is given by
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu5.png?pub-status=live)
If
$\alpha $
is a
$2$
-cocycle of finite order of
$G,$
then this allows the construction of the
$\alpha $
-covering group H of G (see [Reference Karpilovsky4, Ch. 4, Section 1] or [Reference Haggarty and Humphreys1, page 191]). Let
$\omega $
be a primitive
$o(\alpha )$
th root of unity and let
$A = \langle \omega \rangle .$
The set of elements of H may be taken to be
$\{ar(x): a\in A, x\in G\}$
, and H is a group under the binary operation
$ar(x)br(y) = ab\alpha (x, y)r(xy)$
for all
$a, b\in A$
and all
$x, y\in G.$
This is a central extension of G:
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu6.png?pub-status=live)
with
$\pi (r(x) )= x$
for all
$x\in G.$
It also has the following important property. Let P be an
$\alpha ^i$
-representation of G for
$i\in \mathbb {Z}$
. Then
$R(ar(x)) = \lambda ^i(a)P(x)$
for all
$a\in A$
and all
$x\in G$
is an ordinary representation of
$H,$
where
$\lambda \in \operatorname {\mathrm {Lin}}(A)$
with
$\lambda (\omega ) = \omega ;$
moreover, P is irreducible if and only if R is. Here R is said to linearise P (or to be the lift of P). Let
$\operatorname {\mathrm {Irr}}(H\vert \lambda ^i) = \{\chi \in \operatorname {\mathrm {Irr}}(H): \chi _A = \chi (1)\lambda ^i\}$
for
$i\in \mathbb {Z}.$
Then the linearisation process outlined means that for each such i there exists a bijection from
$\operatorname {\mathrm {Irr}}(H\vert \lambda ^i)$
to
$\operatorname {\mathrm {Proj}}(G, \alpha ^i)$
defined by
$\chi \mapsto \xi ,$
where
$\chi (r(x)) = \xi (x)$
for all
$x\in G$
and it is convenient to say that
$\chi $
linearises
$\xi $
.
Now x is
$\alpha $
-regular if and only if
$\omega ^ir(x)$
and
$\omega ^jr(x)$
are not conjugate for all i and j with
$0\leq i<j\leq o(\alpha )-1.$
So for counting purposes there are exactly
$o(\alpha )$
conjugacy classes of H that map under
$\pi $
to the conjugacy class of an
$\alpha $
-regular element of G and fewer than this for an element that is not
$\alpha $
-regular. If
$o(\alpha ) = o([\alpha ]),$
then
$A\leq H'$
and the mapping
$\alpha ^i\mapsto [\alpha ^i] = [\alpha ]^i$
for
$i =0,\ldots , o(\alpha )-1$
is a bijection.
Lemma 2.5. Let
$\alpha $
be a
$2$
-cocycle of G of finite order and let H be the
$\alpha $
-covering group of
$G.$
If
$r(x)\in H$
is real, then so too is
$x.$
Conversely if
$x\in G$
is real, then
$r(x)$
is real if and only if there exists
$g\in G$
such that
$gxg^{-1} = x^{-1}$
and
$f_{\alpha }(g, x) = \alpha (x, x^{-1})^{-1}.$
Proof. If
$r(x)$
is real with
$r(g)r(x)r(g)^{-1} = r(x)^{-1},$
it follows that
$f_{\alpha }(g, x)r(gxg^{-1}) = \alpha (x, x^{-1})^{-1}r(x^{-1}),$
so that in particular
$gxg^{-1} = x^{-1}$
and x is real. The converse is now obvious.
Lemma 2.6. Let
$\alpha $
be a
$2$
-cocycle of G of finite order and let H be the
$\alpha $
-covering group of
$G.$
Let
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha ^i)$
for
$i\in \mathbb {Z}$
and let
$\chi \in \operatorname {\mathrm {Irr}}(H\vert \lambda ^i)$
linearise
$\xi .$
Then
$v_n^{\alpha ^i}(\xi ) = v_n(\chi ).$
Proof. Using the notation introduced,
$r(x)^n = p_{\alpha }(x, n)r(x^n)$
for
$n\in \mathbb {N}.$
So from Theorem 1.1,
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu7.png?pub-status=live)
since
$v_n(\lambda ^i) = v_1(\lambda ^{ni})$
from Theorem 1.1, so that
$v_n(\lambda ^i) =1$
if
$o(\lambda ^{ni}) = 1$
and is
$0$
otherwise.
Let
$\alpha $
be a
$2$
-cocycle of G of finite order and let H be the
$\alpha $
-covering group of
$G.$
Consider another transversal of A in
$H, \{s(x): x\in G\}$
with
$s(1) = 1,$
where
$s(x) = \delta (x)r(x)$
for
$\delta (x)\in A.$
This gives rise to a new
$2$
-cocycle
$\beta \in [\alpha ]$
with
$\beta = t(\delta )\alpha $
and for which
$o(\beta )$
divides
$o(\alpha ).$
Let
$\chi \in \operatorname {\mathrm {Irr}}(H\vert \lambda ^i)$
. Then
$\chi $
linearises
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha ^i)$
and
$\xi '\in \operatorname {\mathrm {Proj}}(G, \beta ^i),$
where
$\xi '(x) = \lambda ^i(\delta (x))\xi (x)$
for all
$x\in G.$
Now
$s(x)^n = r(x)^n$
for
$n\equiv 0 \pmod {o(\alpha )}$
and so, from the proof of Lemma 2.6,
$v_n^{\alpha ^i}(\xi ) = v_n^{\beta ^i}(\xi ')$
for
$n\equiv 0 \pmod {o(\alpha )}.$
If
$o(\alpha ) = o([\alpha ]),$
then
$o(\beta ) = o(\alpha )$
and H is also the
$\beta $
-covering group of
$G.$
Using this notation,
$\{s(x): x\in G\}$
can be chosen to be conjugacy-preserving, that is,
$s(x)$
and
$s(y)$
are conjugate in H whenever x and y are conjugate in G (see [Reference Karpilovsky5, Lemma 4.1.1] or [Reference Haggarty and Humphreys1, Proposition 1.1]) and this choice makes
$\beta $
a class-function
$2$
-cocycle.
The next result is an immediate corollary of Lemma 2.6 from [Reference Isaacs3, page 58].
Corollary 2.7. Let
$\alpha $
be a
$2$
-cocycle of G with
$o(\alpha ) = o([\alpha ]) = 2.$
Let
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha ).$
Then
$v_2^{\alpha }(\xi ) = 0$
or
$\pm 1.$
Moreover,
$v_2^{\alpha }(\xi ) = 0$
if and only if
$\xi $
is nonreal,
$v_2^{\alpha }(\xi ) =1$
if and only if
$\xi $
is afforded by a real
$\alpha $
-representation, and
$v_2^{\alpha }(\xi ) = -1$
if and only if
$\xi $
is real but is not afforded by any real
$\alpha $
-representation of
$G.$
Lemma 2.6 also explains why the second Frobenius–Schur indicator is defined to be
$0$
when
$o(\alpha )> 2,$
but another rationale follows. If
$\alpha (x, y)\not \in \mathbb {R}$
and P is an
$\alpha $
-representation of
$G,$
then at least one of the three matrices
$P(x), P(y)$
and
$P(xy)$
must contain a nonreal entry.
Example 2.8. Consider the elementary abelian group
$G = C_p\times C_p$
for p a prime number, which has
$M(G)\cong C_p$
(see [Reference Karpilovsky4, Proposition 10.7.1]). Let
$\alpha $
be any
$2$
-cocycle of G with
$o([\alpha ]) = p.$
Then the only
$\alpha $
-regular element of G is the identity element and consequently the only element
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha )$
has
$\xi (1) = p$
and
$\xi (x) = 0$
for
$x\ne 1$
(see [Reference Karpilovsky5, Theorem 8.2.21]). So
$\xi $
is integer-valued, but is not afforded by any real
$\alpha $
-representation for
$p\geq 3$
from the remark preceding this example. If
$o(\alpha )\geq 3$
and is finite, let H be the
$\alpha $
-covering group of G and let
$\chi \in \operatorname {\mathrm {Irr}}(H\vert \lambda )$
linearise
$\xi .$
Then
$\chi $
is nonreal since
$\lambda $
is nonreal.
It can be concluded from Example 2.8 that the results of Corollary 2.7 do not hold in general for any group G with a
$2$
-cocycle of finite order greater than
$2$
and in this case
$v_2^{\alpha }(\xi ) = 0$
for all
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha )$
can only be interpreted as meaning that each
$\xi $
is not afforded by any real
$\alpha $
-representation of
$G.$
It should be noted that in general the value of
$v_n^{\alpha }(\xi )$
for
$n\equiv 0\pmod {o(\alpha )}$
depends upon the choice of
$\alpha $
, even if
$o(\alpha ) = o([\alpha ]) =2$
, as the next example illustrates.
Example 2.9. Let
$G = C_2\times C_2.$
It is well known that G has two Schur representation groups (also known as covering groups) up to isomorphism, namely D and Q, the dihedral and quaternion groups of order
$8$
, respectively. The character tables of these two groups are identical, and the irreducible characters
$\chi $
and
$\chi '$
of degree
$2$
of each linearise
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha )$
and
$\xi '\in \operatorname {\mathrm {Proj}}(G, \alpha ')$
respectively, where
$\alpha $
and
$\alpha '$
are the
$2$
-cocycles of G constructed from D and Q of order
$2$
with
$o([\alpha ]) = o([\alpha ']) = 2.$
Now
$\xi $
and
$\xi '$
are identical and integer-valued from Example 2.8; however,
$v_2^{\alpha }(\xi ) = v_2(\chi ) = 1,$
whereas
$v_2^{\alpha '}(\xi ') = v_2(\chi ') = -1.$
Using Lemma 2.6 other results concerning
$v_n$
carry over to
$v_n^{\alpha },$
as in the next lemma.
Lemma 2.10. Let
$\alpha $
be a
$2$
-cocycle of G of finite order. Let
$\xi \in \operatorname {\mathrm {Proj}}(G, \alpha )$
and let
$\mu \in \operatorname {\mathrm {Lin}}(G)$
with
$\mu ^n$
trivial for
$n\in \mathbb {N}.$
Then
$v_n^{\alpha }(\xi )\in \mathbb {Z}$
and
$v_n^{\alpha }(\mu \xi ) = v_n^{\alpha }(\xi ).$
Proof. Let H be the
$\alpha $
-covering group of G and
$\chi \in \operatorname {\mathrm {Irr}}(H\vert \lambda )$
linearise
$\xi $
. Then
$v_n^{\alpha }(\xi )\in \mathbb {Z}$
from Lemma 2.6 and Theorem 1.1. Now let
$\nu \in \operatorname {\mathrm {Lin}}(H)$
linearise
$\mu .$
Then
$\nu \chi $
linearises
$\mu \xi $
and
$\nu ^n$
is trivial, so
$v_n^{\alpha }(\mu \xi ) = v_n(\nu \chi ) = v_n(\chi ) = v_n^{\alpha }(\xi )$
using [Reference Isaacs3, Lemma 4.8] and Lemma 2.6.
3 Frobenius–Schur indicator applications
Let
$\alpha $
be a
$2$
-cocycle of G of finite order and define
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu8.png?pub-status=live)
From Lemma 2.10,
$\theta _n^{\alpha }$
is an integral linear combination of
$\alpha $
-characters of G and so
$\theta _n^{\alpha }(x) = 0$
if x is not
$\alpha $
-regular. If, in addition,
$\alpha $
is a class-function
$2$
-cocycle, then
$\theta _n^{\alpha }$
is a class function. If
$o(\alpha ) = 1,$
then
$\theta _n^{\alpha } = \theta _n$
as in Theorem 1.1.
By analogy with the definition in Theorem 1.1, define
$\theta _n^+: G\rightarrow \mathbb {Z}_{\geq 0}$
by
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu9.png?pub-status=live)
for
$n\in \mathbb {N}$
. This function is used in the generalisation of Theorem 1.1.
Theorem 3.1. Let
$\alpha $
be a
$2$
-cocycle of G with
$o(\alpha ) = o([\alpha ])$
of finite order m and let
$n\in \mathbb {N}$
with
$n\equiv 0\pmod {m}.$
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu10.png?pub-status=live)
Proof. Let H be the
$\alpha $
-covering group of
$G.$
Then, using Theorem 1.1 and Lemma 2.6,
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu11.png?pub-status=live)
for all
$x\in G.$
Continuing with the notation and hypotheses in Theorem 3.1, suppose
$g\in G$
with
$g^n = x$
and let
$y\in C_G(x).$
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu12.png?pub-status=live)
Now if m is a prime number and x is not
$\alpha $
-regular, then
$r(x)$
is conjugate to
$ar(x)$
for all
$a\in A.$
So if
$r(y)r(x)r(y)^{-1} = ar(x),$
then the mapping
$g\mapsto ygy^{-1}$
defines a bijection from
$\{g\in G: p_{\alpha }(g, n) = 1~ \text {and}~g^n = x\}$
to
$\{g\in G: p_{\alpha }(g, n) = a~ \text {and}~g^n = x\},$
which explains why
$m\theta _n^+(x) = \theta _n(x)$
in this scenario.
The next result is a special case of Theorem 3.1 that generalises Corollary 1.2.
Corollary 3.2. Let
$\alpha $
be a
$2$
-cocycle of G with
$o(\alpha ) = o([\alpha ]) = 2.$
Let H be the
$\alpha $
-covering group of G and let H and G have exactly t and s involutions, respectively. Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu13.png?pub-status=live)
Proof. Using Corollary 1.2 and the proof of Theorem 3.1,
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250110135933449-0162:S0004972724000571:S0004972724000571_eqnu14.png?pub-status=live)
The final aim is to generalise Theorem 1.3, which involves an analysis of the real conjugacy classes of
$G.$
Lemma 3.3. Let
$\alpha $
be a class-function
$2$
-cocycle of G with
$o(\alpha ) = o([\alpha ]) = 2.$
Let H be the
$\alpha $
-covering group of G with its associated central subgroup
$A = \langle -1\rangle $
and transversal
$\{r(x): x\in G\}.$
Let
$x\in G$
be real. Then
$r(x)$
is nonreal if and only if x is
$\alpha $
-regular and
$\alpha (x, x^{-1}) = -1.$
Proof. If x is
$\alpha $
-regular, then
$r(x)$
is real if and only if
$\alpha (x, x^{-1}) = 1$
from Lemma 2.5. On the other hand, if x is not
$\alpha $
-regular, then there exists
$y\in C_G(x)$
such that
$r(y)r(x^{-1})r(y)^{-1} = -r(x^{-1}).$
Now if
$gxg^{-1} = x^{-1},$
then either
$f_{\alpha }(g, x)$
or
$f_{\alpha }(yg, x)$
equals
$\alpha (x, x^{-1})^{-1}$
and so
$r(x)$
is real from Lemma 2.5.
Let P be an
$\alpha $
-representation of G of dimension
$n.$
Then for all
$g, x\in G$
,
$P(g)P(x)P(x^{-1})P(g)^{-1}$
equals
$f_{\alpha }(g, x)f_{\alpha }(g, x^{-1})\alpha (gxg^{-1}, gx^{-1}g^{-1})I_n,$
but it also equals
$\alpha (x, x^{-1})I_n.$
Thus if
$\alpha $
is a class-function
$2$
-cocycle of G and x is
$\alpha $
-regular, then
$\alpha (x, x^{-1}) = \alpha (gxg^{-1}, gx^{-1}g^{-1})$
for all
$g\in G.$
In the context of Lemma 3.3 and using this result, let
$k_0, k^+$
and
$k^-$
denote the number of conjugacy classes
$\mathcal {C}$
of G that are respectively (a) real and not
$\alpha $
-regular, (b) real and
$\alpha $
-regular with
$\alpha (x, x^{-1}) = 1$
for all
$x\in \mathcal {C},$
and (c) real and
$\alpha $
-regular with
$\alpha (x, x^{-1}) = -1$
for all
$x\in \mathcal {C}.$
Theorem 3.4. Let
$\alpha $
be a class-function
$2$
-cocycle of G with
$o(\alpha ) = o([\alpha ]) = 2.$
Then the number of real elements of
$\operatorname {\mathrm {Proj}}(G, \alpha )$
is
$k^+ - k^-.$
Proof. Let H be the
$\alpha $
-covering group of
$G.$
The number of real conjugacy classes of G and H is
$k_0 + k^+ + k^-$
and
$k_0 + 2k^+$
, respectively, from Lemma 3.3 and previous remarks. Thus from Theorem 1.3 the number of real elements of
$\operatorname {\mathrm {Proj}}(G, \alpha )$
is the second number minus the first.
If
$\alpha '$
is a
$2$
-cocycle of G with
$o(\alpha ') = o([\alpha ']) = 2,$
then we may let H be the
$\alpha '$
-covering group of
$G.$
As explained after Lemma 2.6: (a) there exists a change of transversal so that the resultant
$2$
-cocycle
$\alpha $
of G is a class-function
$2$
-cocycle with
$o(\alpha ) = 2$
and
$\alpha \in [\alpha '];$
(b) the numbers of real elements of
$\operatorname {\mathrm {Proj}}(G, \alpha )$
and
$\operatorname {\mathrm {Proj}}(G, \alpha ')$
are equal, with this number given by Theorem 3.4.
Example 3.5. Every element of the symmetric group
$S_4$
is real,
$M(S_4)\cong C_2$
and
$S_4$
has two Schur representation groups up to isomorphism (see [Reference Morris6, Theorem 1]). One is the binary octahedral group, and the three elements of
$\operatorname {\mathrm {Proj}}(S_4, \alpha )$
constructed from this group, for a class-function
$2$
-cocycle
$\alpha $
with
$o(\alpha ) = o([\alpha ]) = 2,$
are all real (see [Reference Morris6, page 70]), so
$k^+ = 3$
and
$k^- = 0.$
The other Schur representation group is
$\mathrm {GL}(2, 3)$
, and only one element of
$\operatorname {\mathrm {Proj}}(S_4, \alpha ')$
constructed from this group, for a class-function
$2$
-cocycle
$\alpha '$
with
$o(\alpha ') = 2$
and
$\alpha '\in [\alpha ],$
is real (see [Reference Humphreys2, Remark (ii), pages 27–28] or [Reference Morris6, page 56]), so here
$k^+ = 2$
and
$k^- = 1.$