inflection points and canonical forms of non-singular cubic curves
In this entry, we shall investigate the inflection pointsof non-singular
complex cubic curves and show that theequations of such curves can always be put in variouscanonical forms
. The reason for lumping these two subjectsin a single entry is that, being related, it is moreefficient to develop them both at the same time than it wouldbe to treat either one in isolation — on the one hand,putting the equation of the curve in a canonical formsimplifies calculations relating to inflection points whilst,on the other hand, information about inflection points allowsone to construct canonical forms.
We shall consider our curves as projective curves and describethem with homogenous equations; i.e. a curve is a locus in where for some third-order homogenous polynomial . For the purpose at hand, theterm “inflection point” may be taken to mean a point on thecurve where the tangent
intersects the curve with multiplicity
3 — a point on the curve will have this property if and onlyif it is a zero of the Hessian.
We begin by presenting a crude canonical form. With furtherrefinement, this will become the Weierstrass canonical form.
Theorem 1.
Suppose is a non-singular complex cubic curve and a pointon that curve. Then we may choose homogenous coordiantes in whichthe coordiantes of are and equation of the curvelooks as follows:
Proof.
The most general equation of a cubic is as follows:
Requiring that the curve pass through means that. Since our curve is not singular, it has a tangentat all points, in particular, a tangent through . Wemay ask that the equation of the tangent through be, which requires that . Since our curve isnot singular at , both and cannot be zeroso, without loss of generality, we may take .
We may then make the following linear transform:
Upon doing so, our equation assumes the form
where the coefficients are defined as follows:
∎
Next, we turn our attention to the Hessian to begin ourinvestigation of inflection points.
Theorem 2.
If is a non-singular complex cubic curve with homogenousequation , then the Hessian of does not equala multiple of .
Proof.
By the foregoing theorem, we know that there existsa system of coordinates in which
Let us examine the Hessian at the point :
If , then the Hessian cannot me a multiple of because . Assume then that . In thatcase, we must have because, if , then would factor as a product of and a quadratic polynomial,which is impossible because the curve is assumed notto be singular.11Were forsome quadratic polynomial , then the gradient of would vanish at the intersection of the line andthe conic , so the variety
described by would be singular at the point(s) ofintersection. Evaluating the Hessian on the line ,
Since cannnot be zero, this does not vanish identicallyand is clearly not a non-zero multiple of .∎
This theorem tells us that is the equation ofa cubic curve distinct from . Hence, we know that theremust be at least one and at most nine point of intersectionof these two curves, i.e. at least one and at most nineinflection points. Knowing that inflection points exist, wecan improve our canonical form by choosing as aninflection point. With a little tidying up, this will give usWeierstrass’ canonical form.
Theorem 3.
Suppose is a non-singular complex cubic curve and a pointof inflection of on that curve. Then we may choose homogenouscoordinates in which the coordiantes of are andequation of the curve looks as follows:
Proof.
By our earlier result, we know that there existcoordinates in which the equation of the curve is
and the coordinates of are . By thecalculation made in the proof of the previous theorem,we know that . Since is a pointof inflection, this implies that .
As noted in the previous proof, we cannot have because that would make our curve singular. Hence,we can divide by to make the following coordinatetransformation:
After canceling a factor of , our equation becomes
where
∎
Using this canonical form, it is easy to show that wehave the maximum number of inflection points possible.
Theorem 4.
A non-singular complex cubic curve has nine inflection points.
Proof.
Given an inflection point, the foregoing theorem tells usthat we can choose a coordinate system in which that pointhas coordinates and the equation of the curveassumes the form where
Computing the Hessian, we have
Computing its gradient at , we find that
For comparison, we have
Thus, we see that the curves and are both smooth at andintersect transversely, hence they have intersectionmultiplicity 1 there. Because we could choose ourcoordinates to place any inflection point at, this means that every intersection ofour curve with its Hessian has intersectionmultiplicity 1. Since both and are ofthird order, Bezout’s theorem implies that thereare nine distinct intersection points, i.e. ninedistinct points of inflection.∎
Knowing that there exist multiple inflection points,we will now cast the equation in a form which placestwo inflection points at priveleged locations.
Theorem 5.
Suppose is a non-singular complex cubic curve and and are points of inflection of on that curve. Then wemay choose homogeneous coordinates in which the coordinatesof are , the coordinates of are and the equation of the curve looks as follows:
Proof.
Since is a point of its inflection, the tangent to at intersects with multiplicity 3 at . Since is a cubic, this means that this tangent cannotintersect at any other points; in particular, itcannot pass through . Likewise, the tangent through cannot pass through . Hence, the tangent at and the tangent at must intersect at some third point which is not collinear with and .
We may choose our coordinates so as to place at , at and at . The equation of ourcurve is , where
Since and , we must have and in order for and to lie on. The equation of the line is .Restricting to this line, we have
Since is an inflection point, the curve must havethird order contact with this tangent line, the restrictionof to this line should have a triple root at ,i.e. should be a multiple of . Hence, and.
Likewise, the equation of the line is .Restricting to this line, we have
Again, since is an inflection point, we must havea triple root, so this quantity should be a multipleof , so we must have and .
Summarizing our progress so far, we have found that
Next, we note that cannot be zero because thatwould imply that had a singularity at .Likewise, we must not have be zero because thatwould mean a singularity at . Also, were ,we would have which wouldhave a singularity at . Thus, by rescaling, , and if necessarry, we can take, so our equation assumes the form
∎
Upon examining the form of the equation just derivedmore closely, we learn an important geometric fact:
Theorem 6.
Suppose that is a non-singular complex cubic curveand that and are inflection points of . Thenthe line intersects in a third point , whichis also an inflection point.
Proof.
By the foregoing result, we know that we can write theequation for as with
with located at and located at .The line then has equation . Restricting to this line, , so, the third point of intersection of with , hascoordinates .
Next, we compute the Hessian determinant at :
Since it equals zero, is an inflection point.∎
Since our curve is of the third order, a line cannotintersect more than three points. Hence, we see thata line passing through two inflection points of anon-singular complex cubic plane curve must pass throughexactly three inflection points. As it turns out, thisobservation, together with the fact that there are exactlynine inflection points suffices to determine the locationsof the inflection points up to collineation. However, ratherthan pursuing this line of reasoning here, we will insteadcast the equation of the curve in a form which makes it easyto locate all the inflection points and makes the symmetry
of the curve apparent.
Theorem 7.
Suppose is a non-singular complex cubic curve. Then wemay choose homogeneous coordinates the equation of the curveassumes the form
Proof.
We know that there exists a straight line which intersects in exactly three points, all of which are inflection pointsof . Choose a homogenous coordinate system in which theequation of is . Since the three points of intersectionof with are distinct, we may place them at any threelocations; we shall choose, , and , where is a primitive cube root of unity.
With this choice, the equation of becomes
By making the change of variables
we simplify the equation of the curve to
where
Note that this transformation leaves the coordinates of the threepoints of intersection of and unchanged.
Next, we impose the requirement that the three points lying on beinflecton points of . Setting to zero, the Hessian becomes
In order for the three points intersection of and to beinflection points, we must have this be a multiple of .This requires that and both be zero.
Were zero as well, our cubic would have a singularity at .Since is assumed to not be singular, that means that , sowe can make the further transform
to put the equation for in the form .∎
While we have shown that we can transform the equation of anynon-singular curve into our canonical form, there remains thepossibility that a curve whose equation is expressible inour canonical form may be singular or degenerate. We willnow examine this possibility.
Theorem 8.
A equation describes a singularcurve if and only if .
Proof.
The curve described by the equation is singularif and only if there exists a point on the curve at which allthree partial derivatives of go zero. Taking derivatives
and doing a little bit of algebraic
manipulation, this meansthat, for the curve to be singular, there must be a non-trivialsolution to the system of equations
Multiplying the last three equations together, we obtain, or. If , the onlyway for this to be satisfied is to have either or or . However, suppose that . Then, by the last twoequations, we would also have and , so the solutionwould be trivial. Likewise, if , then it follows that and ; and, if , it follows that and .Hence, when , we only have the trivial solution,so the curve is non-singular.
We will finish by explicitly showing that the curve degenerateswhen , i.e. when or or. (As before, is a primitive cuberoot of unity.) In each of these cases, we can factor ourequation into a product of three linear terms:
Thus, when , our curve degenerates into a triangle(whose vertices are the singularities).∎
Having obtained this canonical form, it is quite easy to exhibit allnine inflection points.
Theorem 9.
Given a non-singular complex cubic plane curve, thereexists a coordinate system in which the inflectionpoints of that curve have the following coordinates:
As previously, denotes a primitive third root of unity.
Proof.
For convenience, set andlet be times the Hessian of .Computing, we have
Thus the Hessian is of the same form, but with replaced by. Next, we form two combinations of ande :
As we have just seen, the condition for the curve not to besingular is exactly that , so we may cancelto conclude that
and
The latter equation will be satisfied if either or or . When , the former equation reducesto , which gives us the solutions
Likewise, when , it reduces to withthe solutions
and, when , it reduces to , withsolutions
∎