Thus, locally an analytic Riemannian metric on a two-dimensional real mani-
fold can always be brought into the form (9.98). Coordinates x; y in this form are
called isothermal coordinates. If two metrics g
1
and g
2
are related as g
2
¼ fg
1
where f is a positive function, then it is readily seen from (9.4, 9.5) that all angles
have the same values in both metrics. The Riemannian geometries with such two
metrics are called conformal, and mappings F : M
1
! M
2
between geometries M
1
and M
2
which preserve all angles are called conformal mappings. Hence, an
analytic Riemannian metric on a two-dimensional manifold is always locally
conformal to the Euclidean metric. If there are coordinates w ¼ u þiv in which
ds
2
may be written as ds
2
¼j/ðwÞj
2
dwd
w, then either dw is proportional to dz
(orientation preserving) or to dz (orientation reversing). In the first case w is an
analytic function of z.IfFðzÞ is an analytic complex function with dF=dz ¼ f ðzÞ,
then locally ds
2
¼ dFd
F and Re F and Im F are Cartesian local coordinates in the
above considered manifold M.
The metric (9.98) defines a one-dimensional complex Riemannian manifold
which is also called a Riemannian surface. Note that ðx
1
þ ix
2
Þ¼fdzmeans
x
1
x
¼ x
2
y
¼ Re f ; x
1
y
¼x
2
x
¼ Im f . It is seen that every two-dimensional orient-
able real manifold allows for a complex manifold structure with a locally
Euclidean metric that makes it into a Riemannian surface.
Let M be a complex manifold of dimension m and consider the tangent space
T
z
ðMÞ on M at point z ¼ðz
1
; ...; z
m
Þ. The linear mapping J : T
z
ðMÞ!T
z
ðMÞ :
JX ¼ iX or JX ¼iX has the obvious property J
2
¼ J J ¼Id
T
z
ðMÞ
. Conversely,
if V is a vector space, then a linear transformation J : V ! V; J
2
¼Id
V
is called
a complex structure on V. Treat V as a real vector space, then two vectors X and
JX cannot be proportional to each other, JX ¼ kX with real k, because then
J
2
X ¼ k
2
X; k
2
0, against the assumption. Hence, X and JX span a two-dimen-
sional subspace E
X
of V which is clearly invariant under J; JE
X
¼ E
X
, and every
invariant subspace of J in V is two-dimensional. A complex structure can only
exist on V, if as a real vector space it is even-dimensional. If V
is the dual space
of V, then a complex structure J on V induces a complex structure (also called J)
on V
by the definition
hx; JXi¼hJx; Xi for all x 2 V
; X 2 V: ð9:99Þ
From this definition, J
2
¼Id
V
readily follows.
An example from physics is the time reversal operator
b
T in quantum
mechanics. As an operator in the space of spinor quantum states W it has the
property
b
T
2
¼Id and hence is a complex structure. If a Hamiltonian
b
H com-
mutes with
b
T, then
b
HW ¼ WE implies
b
Hð
b
TWÞ¼
b
T
b
HW ¼ð
b
TWÞE, and all
eigenvalues E of
b
H are twofold degenerate. This is called Kramers degeneracy.
Much more general implications of a complex structure on quantum physics are
discussed in [6].
9.7 Complex, Hermitian and Kählerian Manifolds 337