^{1}

^{*}

^{2}

This paper proposes an algorithm for the detection of improper parameterization of rational curves using the concept of Grobner bases. The advantage of the proposed algorithm lies in the fact that the Grobner bases can operate in both univariate and multivariate fields with specified ordering.

Rational curves (also called unicursal curves) play an important role in past and present computer development in such areas as geometric modeling and computer graphics.By definition, a rational curve is any curve that can be represented parameterically in the form, , where, , and

are polynomials [1,2].

Theoretically, given any curve either in the explicit, implicit or parametric form, it is important to know whether parameterization exists for the curve or not before attacking the problem of detecting improper parameterization. The above question is simple and can be answered by just checking for the genus of the curve. If the genus of a curve is zero, then it can be parameterized otherwise no parameterization exists for the curve [3-7]. Furthermore, it is only possible to use rational polynomial parametric equations to give an exact representation of a curve iff its genus is zero [8,9].

The concept of parameterization is very important in rational curves (our present interest) in particular and curves and surfaces in general; due to the fact that parametric representations are very easy to handle (implement) during computations [10,11].

A rational curve [

In 1986, an algorithm [

In this paper, we show that it is also possible to detect improper parameterization using the concept of Gröbner bases. We use a necessary and sufficient condition, i.e. the resultant must equal zero [

The remainder of this paper is organized as follows: Section 2 presents our algorithm for the detection of an improper parameterization. Section 3 reviews a numerical example [

In this section, we introduce our algorithm. First of all we would like to describe the definition and notation of some of the mostly used terms in this paper.

Definition 2.1: If a polynomial f in with coefficients in k is a linear combination of monomials, then the polynomial f can be written in the form

where the summation is taken over a finite number of n-tuples, where are nonnegative integers. This set of polynomials in with coefficients in k is denoted as.

Thus, polynomials in one, two and three variables lie in, and, respectively. Therefore, denotes a field in n-variables and a field with one variable is normally denoted by k.

Definition 2.2: An ideal is a subset which satisfies the following:

1)2) If, then, and 3) If and, then.

Definition 2.3: Let be polynomials in and let the subset I be an ideal. Then I can be written in the form

Definition 2.4: Let denote a nonzero polynomial.

1) By letting, and where for every j; then is called the least common multiple of and, written in the form , where and are the leading monomials of u and v respectively and and are the multidegrees of u and v respectively.

2) The S-polynomial of u and v is the combination

where and are the leading terms of u and v respectively.

Definition 2.5: is defined as the remainder on division of by the ordered s-tuple .

Corollary 2.1: Suppose are polynomials both of positive degrees, then f and g are said to have a common Gröbner basis if and only if

where denotes the resultant of f and g with respect to x and denotes the determinant of the Sylvester matrix of f and g with respect to x ([

Corollary 2.2: Given a Gröbner basis of an ideal or, if is any polynomial and ,then the following statements are true

where is a constant for, and ([

Given a plane rational curve of the form

It is well known from Luroth’s theorem [

where

If a nonsingular point exists, then

for some.

We let and where p is a parameter value of a nonsingular point on the curve, and determine the values of p that might describe the same point, if it exists, by developing the system of equations below

From Equation (5), we obtain the two equations below.

Clearly, and are polynomials in p and if Equation (5) is true then, Equations (6) and (7) should have a common Gröbner basis which can be written (Corollary 2.1) as

By changing the value of to, in Equation (8) we obtain another equation of the form

which implies that there are two polynomials: and with a common Gröbner basis.

If Equations (8) and (9) hold, our next task is to evaluate the common Gröbner basis (Corollary 2.2). Let denote the common Gröbner basis for and, and denotes the common Gröbner basis for and; and according to Corollary 2.2, we will represent and as and respectively and also represent the Gröbner basis and as and respectively and hence we can write:

Equation (3) can be evaluated using Equations (10) and (11), while the values of a, b, c and d must be determined by using the conditions: when, and when, since the endpoint interpolation property must be satisfied by both curves i.e. the properly and improperly parameterized curves.

After obtaining and our final task is to determine the coefficients of Equation (2) i.e. the properly parameterized rational curve. Let and have a maximum degree m, and u be the degree of the improperly parameterized rational curve. Hence, the degree of the properly parameterized curve shall be.

Finally; Equation (2) takes the form

We now use the method of undetermined coefficients [

The whole algorithm is summarized as follows:

1) At the beginning we pick values of p.

2) We follow the procedures above and compute.

a) If gives a one-to-one relationship between q and p (i.e. properly parameterized) the algorithm will CONTINUE by selecting a new value of p from step 1.

b) If doesn’t give a one-to-one relationship between q and p (i.e. improperly parameterized) the algorithm will TERMINATE.

Our algorithm will not terminate at step 2(a) iff the equation gives a one-to-one relationship between q and p i.e. properly parameterized; but will terminate at step 2(b) if does not give a oneto-one correspondence between q and p i.e. improperly parameterized.

In this paragraph, we would throw light on how the Gröbner basis of an ideal is computed using Buchberger’s algorithm ([

Input:

Output: a Gröbner basis for I, with

REPEAT

FOR each pair, in DO

IF THEN

UNTIL

At the initial phase of the algorithm, G is enlarge by adding the remainder for. If

, then u, v and are also in I and therefore, since we are dividing by, we obtain. It is interesting to note that G contains the given basis F of I and as such, G is a real basis of I. The algorithm terminates when, which means that

for.

Finally, it is necessary to note that if the polynomial ideal I is in the field, then the above algorithm will give only one Gröbner basis (i.e. common Gröbner basis) that is common to both ideal members.

In this section we solve a numerical example from [

We consider the rational cubic Bezier curve defined by the equations

and compute the parametric substitution that gives the equations below

We pick and. When, we have:

The resultant of these two polynomials is:

This means that and have a common Gröbner basis. Therefore, the of these two polynomials is:

When, we have:

The resultant of and is:

This tells us that there is a common Gröbner basis for the two polynomials. Hence, GB_{4} for the two polynomials is:

Our parametric substitution becomes

We now use the endpoint interpolation property i.e. when and when and determine a, b, c and d to compute our parameter. The first condition gives, and then select and. The second constraint gives, we choose and, and finally obtain our parameter.

In [

In this piece of work, we utilize a necessary and sufficient condition to confirm that the two polynomials have a common Gröbner basis.In Section 3, we see that the existence of common Gröbner basis for both values and is confirmed by the fact that both and equals zero. After confirming the existence of a common Gröbner basis the algorithm is then used to compute this common Gröbner basis.

From Section 3, we see also that there is a one-to-two correspondence between parameter values of q and p, i.e. one value of q gives two values of p as it is evident from the obtained parametric relation; and also the parameterization has a tracing index of two i.e. the curve is doubly traced.Therefore, the curve is improperly parameterized.

The algorithm in [

The advantages of our algorithm are as follows: Our algorithm always confirms the existence of a common Gröbner basis before computation. Secondly, it is based on the concepts of Gröbner bases, which requires operations in the field. We just need to indicate the ordering i.e. lexicographic order, graded lexicographic order, or graded reverse lexicographic order.

Finally, from our algorithm in Section 2, we conclude with no doubt that it is also possible to detect improper parameterization using the notion of Gröbner bases iff the polynomial ideal I is in the field.

The authors would like to thank the anonymous referees for their useful comments on this paper. We would also like to thank the Chinese Scholarship Council.