Let us first reduce to showing that is a set of generators for
the
-module
. This can be done by the following
proposition where notations from the proof of corollary
6.2 are used.
PROOF: We use [Do2] p. 74 ff.
First the reader may check that our definition of and
is identical to the one given there.
According to
[Do2] and 2.2c in [Do1] we have
where
is the
irreducible
module induced from
the linear character
of the Borel subgroup
(notations taken from [Do2]).
Here
runs through the set
of dominant weights
corresponding to the irreducibles occuring in
. If
denotes
the maximal torus of
we may consider the
weights
as the group-like elements in its coordinate
ring. More precisely
is of the form
as can be seen from the argumentation in [Do2]. Here,
is a partition of
in not more than
parts and
an
integer. Restricting to the symplectic group
we have
to set the coefficient of dilation
equal to
. Thus the restriction of
to the maximal torus of
is just the dominant
weight
for the symplectic group itself. Furthermore, it is easy to show that
restricting the
-module structure of
to
the symplectic group gives
the module
induced from the linear
character
of the Borel subgroup
of
(for details see
[Oe], 3.3.3). But the dimension of the latter one is known to
be the cardinality of
(see [Do3]
theorem 2.3 b for instance). Thus, we obtain
Observe that by theorem 4.1 the proof of 6.1
can be reduced to the
case
, since the definition of bideterminants over
and
respectively commutes with the isomorphism
when
is considered as a
-algebra.
Now, suppose we have shown that
generates
as a
-module.
Then the image of
in
under the epimorphism
considered in the proof of 6.2 is a set of
generators, too. By the above proposition it must be a basis of
. Consequentely, there can't be any relations among the
elements of
, especially none with integer coefficients, giving
the desired result.
The proof that is indeed a set of generators will follow from a
symplectic version of the famous straightening formula. For
convenience of the reader we will first state the algorithm leading to the
classical straightening formula. To do so, we put an order on the set
of partitions of
writing
if the dual
occurs before the dual
in the lexicographical order.
In this order the fundamental weight
is the largest element, whereas
is the
smallest one. We abbreviate
and define
resp.
to be the
-linear span in
of all
bideterminants
such that
resp.
. For
we set
. Clearly
.
Here, the order on is the lexicographical one according to the
given order on
, in our case
. A proof of the proposition can be found for
example in [Mr], 2.5.7.
Let us now state the symplectic analogue. First consider the algebra
where
is the ideal in
generated by the
coefficient of dilation. It is graded since
is a homogenous element
but not a bialgebra because an augmentation map is missing. In fact, it turns
out that (in the case where
is an algebraically closed field)
it is the coordinate ring of the semigroup
of
noninvertible elements in the symplectic monoid (see remark
).
Let us abbreviate its submodule of homogenous elements of
degree
by
and define
and
in the same manner as above. Further,
define a map
by
, where
and order
writing
if and only if
appears before
in the lexicographical order. Next, we
obtain an order
on
in a lexicographical way, as
well:
Finally, this gives a new order on
via the embedding
given by
. Now we are able to state the symplectic
straightening algorithm:
Before proving this, let us deduce that is a set of
generators for
. First note that multiplication by
the coefficient of dilation
leads to an exact sequence for
Therefore, by induction on we can reduce to showing that
is a set of generators for
. For this claim it is
enough to show that
is a set of generators of
for each partition
. This can be deduced from the straightening algorithm
7.3: Since
is a finite set, the elemination of multi-indices
not being
-symplectic standard in an expression
must terminate. This gives the straightening formula
concerning the right-hand-side argument of
:
Now, there is an algebra automorphism on
induced
by matrix transposition and given by
on
generators. Of course it is an anti-automorphism of coalgebras.
It can be readily seen that
and
. Therefore, it factors to an automorphism of
which will be denoted by the same symbol. From the definition of
bideterminants we see
.
Applying to the congruence relation in
7.4 we see that
a non
-symplectic standard entry
on the left-hand-side
entry of a bideterminant can be eleminated, too, not
affecting the right-hand-side entry. Thus, it must be possible to
write
as a sum of bideterminants
modulo
where
. Therefore, the proof of
theorem 6.1 is finished as soon as proposition
7.3 is established.
Also, the symplectic straightening formula is related to the treatment
of symplectic Schur-modules in [Do3] and [Ia], section 6,
as can be seen from the proof of lemma 8.1 below.
Concerning the latter paper it should be noted
that the algebra
defined there
as the coordinate ring of the symplectic group itself is only filtered by
but not graded by
. In fact, it can be deduced from the above
remark that
is the corresponding graded algebra,
that is
as
-vector spaces.