Information Technology Reference
In-Depth Information
select g 1 , g 2 from G such that the normal form h of the S -polynomial of g 1 , g 2
is not 0. Finally we extend S by ( γ,G ), h and h . This process is repeated until
the normal form of all S-polynomials is 0.
As mentioned above the generic variant of the Grobner system computation
allows us to drop branches. Recall from the presentation of our quantifier elimi-
nation algorithm that we have to perform for each branch a separate quantifier
elimination. If we are on the top-level of our quantifier-elimination algorithm we
actually compute a Grobner system containing one single branch, because the
condition on line 3 of the algorithm “extend” is tautological in this situation.
This reduces, in general, both the computation time and the size of the output
formula dramatically. As a rule, observed from our sample computations, we
compute only a few assumptions which can often be easily interpreted.
3.2
Generic Equation Construction
In Section 2.3 we have discussed how to construct an equation from a set of
ordering relations. In this section we adapt this to the generic case.
Recall that we generate a complete case distinction about the highest coe -
cient of each h
H . The size of this case distinction can be reduced by making
appropriate assumptions as shown below.
For h
H let
n h =max {−
0 ,...,d h } Var( a h,i )
}∪ i
v 1 ,...,v m } .
1
∈{
⊆{
For all n h with h
H and n h
0weaddtheassumption a h,n h
= 0 to our theory
Θ . Let finally D
k =1 {
. Then we can proceed with the
transformation described in Section 2.3 using D instead of D .Notethat D
=
×
max(0 ,n h ) ,...,d h }
D
and often D
D .
3.3
Generic Type Formula Computation
In this section we discuss an approach to computing generic type formulas.
The type formula construction presented in Section 2.2 is a primitive version
of the method used in our highly optimized Hermitian quantifier elimination. We
actually compute a type formula τ d for a polynomial p = i =0 c i y i of degree d
recursively:
τ d ( c d ,...,c 0 ) .
The recursion basis are the simple type formulas up to the degree 3. The defini-
tion of τ d is similar to the definition of τ d , but assumes a non-vanishing constant
coecient which implies the absence of the zero 0. The formula τ d is actually a
disjunctive normal form. Each constituent has the following schema
τ d ( c d ,...,c 0 )
( c 0 =0)
τ d− 1
c k 1 k 1 0
∧···∧
c k l k l 0 ,
where
{
k 1 ,...,k l }⊆{
1 ,...,d
}
and k j ∈{
<, >
}
.
Search WWH ::




Custom Search