The general creation function for sheaves takes a graded module representing the sheaf and a scheme X on which it is supported. Special constructors are provided in the cases of the structure sheaf of X and the canonical sheaf of X, when X is locally Cohen-Macaulay and equidimensional. The user may also ask for Serre twists of a given sheaf. Other constructions deriving new sheaves from existing sheaves will be described in later sections.
Given an ordinary projective scheme X and a module M over the coordinate ring of the ambient of X, such that M is annihilated by the defining ideal of X, this function returns the sheaf defined by graded module M on scheme X.
Given an ordinary projective scheme X, this function returns the structure sheaf OX for X, which is the sheaf defined by the coordinate ring RX of X, as a module. If the intrinsic is called with a second integer-valued argument n, the object returned is a twisted version of the sheaf, that is, Serre's twisting sheaf OX(n), which has RX(n) as its associated graded module (see Section 5, Chapter II of [Har77]). These are all invertible sheaves on X and OX(1) is the sheaf OX(H) corresponding to the class of a hyperplane divisor H on X.
Given an ordinary projective scheme X, this function returns the canonical sheaf KX for X. The scheme X should be an ordinary projective scheme which is equidimensional and locally Cohen-Macaulay. That is, all of the primary components of X should have the same dimension and its local rings should all be Cohen-Macaulay rings. These conditions aren't checked by Magma as the necessary computations can be very expensive in general. A non-singular variety always satisfies these conditions, and many singular normal varieties do also. For example, any curve or normal surface will be equidimensional and locally Cohen-Macaulay. The stronger condition of being arithmetically Cohen-Macaulay, can be checked by invoking the intrinsic IsArithmeticallyCohenMacaulay with the structure sheaf of X as argument.Under these conditions, X has a canonical sheaf KX, defined up to isomorphism, which acts as a dualising sheaf. See Section 7, Chapter III of [Har77] and Chapter 21 of [Eis95] for the module-theoretic background. For non-singular varieties, the canonical sheaf is the usual one: the highest alternating power of the sheaf of Kahler differentials. The function returns the canonical sheaf of X. It is computed from the dual complex to the minimal free resolution of the coordinate ring of X.
If the intrinsic is invoked with an additional integer argument n, it returns the nth Serre twist (see below) of the canonical sheaf KX(n). For a non-singular variety of dimension d, the map into projective space corresponding to KX(d - 1) is the important adjunction map.
Given a sheaf S, the function returns the nth Serre twist of Para, Para(n) isomorphic to Para tensor OXOX(n). If M is a module giving Para, then M(n) gives Para(n).
> P<x,y,z,t> := ProjectiveSpace(Rationals(),3); > R := CoordinateRing(P); > X := Scheme(P,x^3+y^3+z^3+t^3); > OX := StructureSheaf(X);We first examine the underlying graded module of the structure sheaf.
> Module(OX); Reduced Module R^1/<relations> Relations: [x^3 + y^3 + z^3 + t^3]Observe that the canonical sheaf KX of X is isomorphic to the twist OX( - 1) of the structure sheaf.
> KX := CanonicalSheaf(X); > Module(KX); Reduced Module R^1/<relations> with grading [1] Relations: [x^3 + y^3 + z^3 + t^3] > Module(StructureSheaf(X,-1)); Reduced Module R^1/<relations> with grading [1] Relations: [x^3 + y^3 + z^3 + t^3]Note that the module column weights are the negations of the Serre twist indices!
> Module(Twist(OX,-1)); Reduced Module R^1/<relations> with grading [1] Relations (Groebner basis): [x^3 + y^3 + z^3 + t^3]The equations x=z, y=t define an (exceptional) line in X. We can get its structure sheaf as a sheaf on X using the basic Sheaf constructor. The associated invertible sheaf (L)(Y) of Y as a divisor on X can be obtained from the DivisorToSheaf intrinsic described later in the chapter.
> IY := ideal<R|[x+z,y+t]>; // ideal of line > OY := Sheaf(QuotientModule(IY),X); > Module(OY); Graded Module R^1/<relations> Relations: [x + z], [y + t] > Scheme(OY); Scheme over Rational Field defined by x^3 + y^3 + z^3 + t^3
Maximize: BoolElt Default: false
Given an ordinary projective scheme X, this function returns the sheaf of 1-differentials on X, Ω1X/k. The function computes the natural representing module for the sheaf coming from the embedding of X in projective space (see Section 8, Chapter II of [Har77]). If the parameter Maximize is true, then the maximal module representing this sheaf is computed and used to define it (see next section).
Maximize: BoolElt Default: false
For an ordinary projective scheme X, this function returns the sheaf of tangent vectors for X. The function computes the natural representing module for these sheaves coming from the embedding of X in projective space (see Section 8, Chapter II of [Har77]). If the parameter Maximize is true, then the maximal module representing this sheaf is computed and used to define it (see next section).Combining either of the above intrinsics with the IsLocallyFree intrinsic, this gives an alternative method for checking non-singularity on varieties that are known to be (locally) Cohen-Macaulay. It is best to use the sheaf of differentials since that is generally easier to compute. This approach can be much faster for varieties having high codimension than the usual Jacobian method.
The projective space P should be ordinary projective 4-space Prj4 over a field. The function returns the locally free rank 2 sheaf on P which represents the Horrocks-Mumford bundle (see [HM73]). The scheme of vanishing of a general global section of this sheaf is a two dimensional Abelian variety in P.