1-Manifolds Solutions PDF
1-Manifolds Solutions PDF
1-Manifolds Solutions PDF
Spring 2011
Problem 3. For the south pole taken as the center of the projection,
everything is similar to the results obtained above (see the solution to Prob-
lem 2). Since the north pole N = (0, 0, 1) and the south pole S = (0, 0, 1)
exchange places under the transformation z 7 z, all the formulas can be
obtained from the corresponding formulas for the stereographic projection
from the north pole by a change of signs. Answers:
(a) The coordinates (u0 , v 0 ) are defined for all points of S 2 except for the
1
Theodore Voronov. Differentiable Manifolds. Spring 2011
2u0 2v 0 1 (u0 )2 (v 0 )2
x= , y= , z= .
(u0 )2 + (v 0 )2 + 1 (u0 )2 + (v 0 )2 + 1 (u0 )2 + (v 0 )2 + 1
(c) Both coordinate systems (u, v) and (u0 , v 0 ) are defined on S 2 \ {N, S}.
To find the change of coordinates u = u(u0 , v 0 ), v = v(u0 , v 0 ), we substitute
x, y, z as functions of u0 , v 0 to the expressions for u and v. For example,
x 2u0 (u0 )2 +(v 0 )2 +1 u0
for u we obtain u = 1z = (u0 )2 +(v 0 )2 +1 2((u0 )2 +(v 0 )2 ) = (u0 )2 +(v 0 )2 , because
0 2
1(u ) (v ) 0 2 0 2
2((u ) +(v ) ) 0 2
1 z = 1 (u0 )2 +(v 0 )2 +1 = (u0 )2 +(v 0 )2 +1 . And similarly for v. So the change of
u0 v0
u= , v= .
(u0 )2 + (v 0 )2 (u0 )2 + (v 0 )2
These formulas are defined for all (u0 , v 0 ) except for (u0 , v 0 ) = (0, 0), which
are the coordinates of the north pole N in the (u0 , v 0 )-system. Conversely,
the change of coordinates u0 = u0 (u, v), v 0 = v 0 (u, v) is given by
u v
u0 = , v0 = .
u2 + v2 u2 + v2
These formulas are defined for (u, v) 6= (0, 0), which are the coordinates of
the south pole S in the (u, v) coordinate system.
2
Theodore Voronov. Differentiable Manifolds. Spring 2011
For the inverse map, we solve for x and z having in mind the relation |x|2 +
z 2 = 1. We have x = u(1 z), hence we arrive at the quadratic equation
(denoting 1 z = w): |u|2 w2 + (1 w)2 = 1, or (1 + |x|2 )w2 2w = 0, from
which we obtain the unique solution w = 2/(|u|2 + 1), since w = 1 z 6= 0,
and therefore
2u
x=
|u|2 + 1
|u|2 1
z=
|u|2 + 1
|u(N ) |2 1
2u(N )
N : u(N ) 7 , ,
|u(N ) |2 + 1 |u(N ) |2 + 1
x
1
N : (x, z) 7 u(N ) = .
1z
(We have introduced the subscript N as a label of the chart.)
By considering the south pole S = (0, 1) as the center of the projection
instead of N , we can construct the chart S : Rn(S) S n \ {N } in the same
way. The formulas differ by the sign of the last coordinate only (since N and
S are swapped under z 7 z):
|u(S) |2 + 1
2u(S)
S : u(S) 7 , ,
|u(S) |2 + 1 |u(S) |2 + 1
x
1
S : (x, z) 7 u(S) = .
1+z
These two charts make an atlas because (S n \ {N }) (S n \ {S}) = S n . On
the intersection (S n \ {N }) (S n \ {S}) = S n \ {N, S} we can calculate the
changes of coordinates, i.e., the expression of u(N ) via u(S) and vice versa.
Note that S n \ {N, S} corresponds to Rn \ {0} at each chart. One can find
directly:
u(S)
u(N ) =
|u(S) |2
(check!). This is a smooth map (even analytic), therefore we have a structure
of a smooth manifold for S n .
3
Theodore Voronov. Differentiable Manifolds. Spring 2011
Uk := {(x1 : . . . : xn+1 ) | xk 6= 0} .
(here 1 stands on the kth place). We used subscript for denoting the number
of chart and put it in brackets to avoid confusion. Define maps k : Rnk Uk ,
kl = 1 n k n l
k l : {u(l) Rl | u(l) 6= 0} {u(k) Rk | u(k) 6= 0}
(where 1 appears at the kth place in the LHS and at the lth place, in the
RHS), from where we obtain
where
ui(l)
ui(k) = .
uk(l)
for all i. (Two special cases of this formula are i = k where we get uk(k) = 1
as it should be, and i = l where ul(k) = u1k , since ul(l) = 1.)
(l)
In small dimensions, we have two charts for RP 1 and three charts for
RP 2 . For RP 1 , the charts are 1 : R U1 where U1 = {(x1 : x2 ) | x1 6= 0)},
1 : x(1) 7 (1 : x(1) ), and 2 : R U1 where U2 = {(x1 : x2 ) | x2 6= 0)},
2 : x(2) 7 (x(2) : 1), and the changes of coordinate are
1 1
x(2) = , x(1) = .
x(1) x(2)
4
Theodore Voronov. Differentiable Manifolds. Spring 2011
5
Theodore Voronov. Differentiable Manifolds. Spring 2011
S 1 RP 1 , (x, y) 7 (x : 1 y) RP 1 ,
S 2 CP 1 , (x, y, z) 7 (x + iy : 1 z) CP 1 ,
and
2x1 x2 (x1 )2 (x2 )2
1 1 1 2
RP S , (x : x ) 7 (x, y) = , ,
(x1 )2 + (x2 )2 (x1 )2 + (x2 )2
2w1 w2 |w1 |2 |w2 |2
1 2 1 2
CP S , (w : w ) 7 (x + iy, z) = , .
|w1 |2 + |w2 |2 |w1 |2 + |w2 |2
1
To get the first set, one writes (x1 : x2 ) = xx2 : 1 = (u : 1) = 1y x
: 1 , so
finally (x1 : x2 ) = (x, 1 y), and similarly for the complex case. To get the
1 1
second set, one has to plug u = xx2 or u + iv = w w2
into the inverse formulas
for the stereographic projection and get rid of denominators. In particular,
for (cos : sin ) RP 1 , we get as the image (x, y) = (sin 2, cos 2) S 1 .
6
Theodore Voronov. Differentiable Manifolds. Spring 2011
7
Theodore Voronov. Differentiable Manifolds. Spring 2011
A 7 gA
where g GL(2), and A and A0 specify the same subspace L if and only
if A0 = gA. Therefore the set of 2-planes (linear subspaces) in R4 can be
identified with the set of equivalence classes of 2 4 matrices A of rank 2
as above w.r.t. the equivalence relation A gA for all g GL(2). The
analogy with the projective space is clear: one takes matrices with linearly-
independent rows instead of non-zero vectors and multiplies/ divides them
by invertible square matrices instead of non-zero numbers. In detail, charts
for G2 (R4 ) can be introduced as follows. Define U34 G2 (R4 ) as the set of
planes specified by matrices A for which the square matrix made of the third
and fourth columns of A is invertible (this does not change under the left
multiplication by g GL(2)). There is a chart
4 b11 b12 b11 b12 1 0
: Mat(22) U34 G2 (R ) , B = 7 [A] =
b21 b22 b21 b22 0 1
is the inverse map. We may put extra labels on and B denoting them
as (34) and B (34) , and introduce similar charts (12) , (13) , (23) , and (24)
corresponding to other choices of columns of A giving invertible 2 2 sub-
matrices. It is clear that the union of all images of these charts is the whole
G2 (R4 ) because of the rank condition satisfied by the matrices A. So it is an
atlas. As for the changes of coordinates, they are given by algebraic opera-
tions with matrix entries, hence are smooth. We conclude that G2 (R4 ) is a
smooth manifold of dimension 2 2 = 4.
(c) In general, the Grassmann manifold Gk (Rn ) is indeed a manifold of
dimension k (n k). The equivalence classes of matrices A Mat(k n) of
rank k can be considered as homogeneous coordinates (A A0 A0 = gA
for g GL(k)). A coordinate chart can be obtained by dividing A from
the left by an invertible k k submatrix with fixed column numbers j1 <
. . . < jk , so that the genuine (inhomogeneous) coordinates in each such
8
Theodore Voronov. Differentiable Manifolds. Spring 2011
chart are given by the entries of a k (n k) matrix. There are nk such