Math 8205
Spring 1998
Solutions to Exercises 3 and 6
Last updated Thursday, June 18, 1998.
Contents
Solution to Exercise 3a
By Exercise 6.4 in the text, every line in Q = G(1,3)
P5 is of the form
L =
p,H,
where p is a point in P3
and H is a plane in P3.
Therefore, we obtain a map of sets
:
W -> F1(Q)
G(1,5) by sending (p,H) -->
p,H .
So, the main point is to show that it is a regular map, i.e.
that the Plücker coordinates of L
G(1,5) are given, at least locally,
as polynomials in the coordinates of p and
H. Now, in W
P3
(P3)*
we have an equality of points:
(p,H) = ([x0,...,
x3],[y0,...,
y3]).
and the inclusion p
H is expressed by the following equation:
x0y0
+ ... + x3y3
= 0.
Since regularity is a local property, we may assume that p
is in the open subset x0
0. Then [y0,..., y3]
[1,0,0,0],
so we may assume without loss of generality that H
(P3)* lies in the open subset
y3
0.
To find the Plücker coordinates of L
G(1,5), it is enough to find the coordinates of two points of
L,
or equivalently, the Plücker coordinates of two lines, say
1
and
2,
which lie in H and pass through p.
To specify
1
and
2,
it suffice sto find two points of H such that
p doesn't lie on the line joining them.
One possibility is to choose the points [0, y3,
0,-y1] and [0,
0, y3, -y2]
respectively.
This gives us the Plücker coordinates of
1
and
2
as the 2 by 2 minors of the matrices :
and
respectively. These minors obviously are polynomials in the homogeneous
coordinates of p and H.
To finish, we use these minors as the entries of a 2 by
6 matrix whose 2 by 2 minors will be the
Plücker coordinates of L.
Those, again, will certainly be polynomials in the homogeneous coordinates
of p and H.
Back
to the exercises.
Solution to Exercise 3b
It's fairly clear that our map W -> F1(Q)
is bijective. Therefore we only need to find dim(W).
We consider the map
:
W -> (P3)*
induced by the projection of P3
(P3)*
to the second fiber.
Clearly,
is surjective. Since the inverse image of H
(P3)* is (isomorphic to) the
plane H
P3,
every fiber of
is of dimension = 2. We conclude that dim(W)
= 3 + 2 = 5.
Back
to the exercises.
Solution to Exercise 6a
The tangential variety TX = Tan(X) has dimension
2r, where r = dim(X).
Also recall that the differential (d
)p
: T p(X) -> T
(p)(P2r
- 1) is not injective if and only if L
tX,p
is nonempty,
where tX,p
is the embedded tangent space of X at p.
-
Case 1: dim(TX) < 2r. In this case,
L
TX is empty
for a general line L
P2r+1,
and there are no points where the differential fails to be injective.
-
Case 2: dim(TX) = 2r. In this case,
L
TX is a
finite set of points for a general line L
P2r+1.
This implies that there are only finitely many points where the
differential is not injective,
provided that we can choose L so that each
point of the intersection L
TX
lies on only finitely many (embedded) tangent spaces of X.
To check this last point, consider the closed subset Y
X
P2r+1,
consisting of all pairs (p,z) such that z
tX,p .
-
By considering the projection to the first factor, we see that dim(Y)
= 2r.
-
TX is the image of Y under the projection
to the second factor. Let
:
Y -> TX be induced by this projection.
If z
TX, then the points of the fiber
-1(z)
correspond bijectively to points p
X such that z
tX,p.
Since we're assuming that dim(TX) = 2r, a
general fiber of
is finite. Thus
-1(z)
is finite for every point in the complement of some proper closed subset
V0
TX. Since dim(V0)
< 2r, L
V0
is empty for a general line L
P2r+1.
Choosing L so that L
TX
is finite and L
V0
is empty, we conclude that there are at most finitely many points
p such that (d
)p
is not injective.
Back
to the exercises.
Solution to Exercise 6b
The secant variety Sec(X) has dimension
2r + 1, where r = dim(X).
-
Case 1: dim(Sec X) < 2r. In this case,
L
Sec(X)
is empty for a general line L
P2r+1, and
=
L
is injective.
-
Case 2: dim(Sec X) = 2r. In this case,
L
Sec(X)
is a finite set of points for a general line L
P2r+1.
Each point in some dense open subset of Sec(X) lies on
a 1-dimensional family of secant lines.
Doing something similar to what we did in Case 2 of part a.,
we show that we can choose L so that each point of
L
Sec(X)
lies on a 1-dimensional family of secant lines. This implies that
the "double point set" D is 1-dimensional.
-
Case 3: dim(Sec X) = 2r + 1. In this case
Sec(X) = P2r+1,
so that L
Sec(X).
So, we have to show that we can choose L so that
each point of L lies on only finitely many secant
lines of X.
Let W0
(X
X -
)
P2r+1
be the set of triples (p,q,z) such that z
<p,q> (= the line joining p and
q),
and let W be the closure of W0
in X
X
P2r+1.
By considering the projection to X
X
-
, we
see that dim(W) = dim(W0
) = 2r + 1.
Now, let
:
Y -> Sec(X) be induced by projection of the product
to P2r+1.
Then a general fiber of
is finite.
Let W1
W be the set of points where the local fiber dimension is
1. Clearly, dim(W1)
< dim(W) = 2r + 1.
We consider the restriction of
to W1 . The key point
that dim(
(W1))
< dim(W1).
(Just note that a general fiber of the restriction has dimension
1.) It follows that dim(
(W1))
2r - 1.
Therefore, if L is a general line in P2r+1,
the intersection L
(W1)
is empty, and the double point set of
L is of dimension =
1.
Back
to the exercises.
Comments and inquiries to: roberts@math.umn.edu
Back to my homepage: http://www.math.umn.edu/~roberts