Complete intersection points on general surfaces in \(\mathbb P^{3}\) (Q1957998)

The paper concerns the study of complete intersection points in 3-projective space. In a previous paper the authors proposed a new approach to the problem of studying complete intersection subvarieties of hypersurfaces. This approach uses a mix of projective geometry and commutative algebra and is more direct than usual methods for addressing the general problem. With their approach they are able to give a complete description of the situation for complete intersections of codimension \(r\) in \(\mathbb{P}^n\) that lie on a general hypersurface of degree \(d\) whenever \(2r \leq n+2\). In this paper, the authors are interested in the first case not covered by their previous results: the case \(n=3\) and \(r=3\). More detailed, we say that \(X \subset \mathbb{P}^3\) is a complete intersection \(0-\)dimensional scheme if its ideal is \(I_X=(F,G,H)\) where the forms \(F\), \(G\) and \(H\) are a regular sequence in the ring \(R=\mathbb{C}[x_0, \dots, x_3]\). Moreover, if \(\deg F=a\), \(\deg G=b\) and \(\deg H=c\) then we say that \(X\) is a complete intersection of type \((a,b,c)\). Assume that \(a \leq b \leq c\) and write CI\((a,b,c)\) to describe a complete intersection of type \((a,b,c)\). The basic question of this paper is: for which integers \(a,b,c\) and \(d\) does the general degree \(d\) surface of \(\mathbb{P}^3\) contain a CI\((a,b,c)\)? There are case where the answer is straightforward. If \(d=c\), then the answer is clearly affirmative because we are cutting a complete intersection curve of type \((a,b)\) with a surface of degree \(d\) (and similarly for \(d=a\) and \(d=b\)). If \(d<a\), then the answer is negative because no form of degree less than \(a\) belongs to the ideal of a CI\((a,b,c)\); and similarly for \(a < d < b\), since a generic form is irreducible. If \(b<d<c\) then we are actually looking for a complete intersection of type \((a,b)\) on the generic degree \(d\) surface and this case is covered in the previous paper of the authors. Hence, they focus on the following refinement of the question: for which integers \(a,b,c\) and \(d\), \(a \leq b \leq c < d\), does the general degree \(d\) surface of \(\mathbb{P}^3\) contain a CI\((a,b,c)\)? The answer is given by the following Theorem. For nonnegative integers \(a,b,c,d\) sucht that \(a \leq b \leq c <d\), the following statements hold. \(\bullet\) If \(a\leq 4\), then the generic degree \(d\) surface of \(\mathbb{P}^3\) contains a CI\((a,b,c)\). \(\bullet\) If \(a=5\) and \(b\leq 11\), then the generic degree \(d\) surface of \(\mathbb{P}^3\) contains a CI\((5,b,c)\); if \(a=5\), \(b=12\) and \(c=12\), then the generic degree \(d\) surface of \(\mathbb{P}^3\) contains a CI\((5,12,12)\); if \(a=5\), \(b=12\) and \(c\geq 13\), then the generic surface of degree \(d\geq 2c+15\) does not contain a CI\((5,12,c)\); if \(a=5\) and \(b\geq 13\), then the generic surface of degree \(d\geq b+c+2\) does not contain a CI\((5,b,c)\). \(\bullet\) If \(a=6\) and \(b\leq 7\), then the generic degree \(d\) surface of \(\mathbb{P}^3\) contains a CI\((6,b,c)\); if \(a=6\), \(b=8\) and \(c=8,9\), then the generic degree \(d\) surface of \(\mathbb{P}^3\) contains a CI\((6,8,c)\); if \(a=6\), \(b=8\) and \(c\geq 10\), then the generic surface of degree \(d\geq 2c+12\) does not contain a CI\((6,b,c)\); if \(a=6\) and \(b\geq 9\), then the generic surface of degree \(d\geq b+c+3\) does not contain a CI\((6,b,c)\). \(\bullet\) If \(a\geq 7\), then the generic surface of \(\mathbb{P}^3\) of degree \(d\geq a+b+c-3\) surface does not contain a CI\((a,b,c)\). The previous theorem gives a complete asymptotic solution to the existence problem for CI\((a,b,c)\) on a general surface of degree \(d\) in \(\mathbb{P}^3\). More precisely, the authors state the following result: Corollary. Let \(a \leq b \leq c <d\) be integers. Then, the generic degree \(d\) surface contains a CI\((a,b,c)\) if: \(\bullet\) \(a\leq 4\); \(\bullet\) \(a=5\) and \(b\leq 11\); \(\bullet\) \(a=5\), \(b=12\) and \(c=12\); \(\bullet\) \(a=6\) and \(b\leq 7\); \(\bullet\) \(a=6\), \(b=8\) and \(c=8,9\). For any other \(a,b,c\) and \(d>>0\) (depending on \(a,b,c\)), the generic degree \(d\) surface does not contain a CI\((a,b,c)\). Finally the authors state the following Conjecture. Given nonnegative integers \(a,b,c\) and \(d\) such that \(a \leq b \leq c < d\), there exists a function \(d(a,b,c)\), possibly assuming the value \(+\infty\), such that the generic degree \(d\) surface in \(\mathbb{P}^3\) contains a CI\((a,b,c)\) if and only if \(d < d(a,b,c)\).