Working with

weighted complete intersections

A. R. Iano-Fletcher

Contents
1 Introduction 102
2 Acknowledgments 103
3 Notation 103
4 Preamble 105
5 Denitions and theorems on weighted projective spaces 105
6 Denitions and theorems on weighted complete intersections108
7 Cohomology of weighted complete intersections 114
8 Quasismoothness 116
9 Cyclic singularities and counting points 123
10 Determination of singularities on weighted complete
intersections 128
11 Preamble 132
12 Weighted curve hypersurfaces 133
13 Weighted surface complete intersections 136
14 Weighted 3-fold complete intersections 145
15 Canonically embedded weighted 3-folds 150
16 Q-Fano 3-folds 154
101
102 Working with weighted complete intersections

17 The plurigenus formula 162

18 The Reid table method 163
References 171

1 Introduction
This article contains the following:
(I) A presentation of the basic denitions, theorems and techniques of
weighted complete intersections, along with many examples. This in-
formation was collected from a variety of sources (mainly [WPS]) but
also includes some original results.
(II) Lists of various types of weighted complete intersections of dimensions
1, 2 and 3 with isolated canonical cyclic quotient singularities.
Weighted complete intersections occur naturally in many disguises. Enriques'
famous example of a surface of general type for which '4KS is not birational
can be expressed as the weighted complete intersection S10 in P(1; 1; 2; 5).
For certain classes of variety V of general type (e.g., minimal surfaces
of general type) the canonical maps 'nKV : V ! Ve are birational onto the
canonical model Ve for large enough n. Dene the canonical ring RV by
M
RV = H0(V; nKV ):
n0
The ring RV is known to be nitely generated in these cases, although not
necessarily generated in degree 1. So Ve ' Spec RV is a subvariety of some
weighted projective space.
Weighted complete intersections are similar to complete intersections in
ordinary projective space Pn but are usually singular and hence have some
pathologies. However they are still very easy to visualise and to work with;
their basic invariants are calculated using combinatorics. So they form a large
quagmire of good examples. This article sets out to familiarise the reader with
weighted complete intersections and to give certain combinatoric conditions
for their important properties. Some of these are already known (see [Da],
[Di], [Du], [WPS], etc.), but some are new. This constitutes Part I.
In Part II we present various lists of weighted complete intersections of
dimension 1, 2 and 3; all with at worst isolated canonical cyclic quotient
singularities. The canonical 3-fold weighted complete intersections are inter-
esting since they are all canonical models (see [R1], [R2], [R4], Section 2.5)
 A. R. Iano-Fletcher 103

and hence are of interest for classication purposes as well as in their own
right. These were all calculated using a set of combinatoric conditions and a
computer. We also give a complete list of the 95 families of weighted hyper-
surface K3 surfaces (see [R1], Section 4.5) found by Reid in 1979 after a long
hand calculation. We also calculate the corresponding singularities.
Another method originally used by Reid to produce examples of K3 sur-
faces is to be found in Section 18. It is used to produce canonically and
anticanonically embedded canonical 3-folds. From the Poincare series of the
graded ring corresponding to a weighted complete intersection, the degrees
of the generators and the relations can be determined. This technique uses
repeated dierencing to evaluate the power series. Using the Riemann{Roch
formula for canonical 3-folds (see Section 17) a Poincare series can be pro-
duced from a list (or record ) of invariants, which we hope will correspond to
either a canonically or an anticanonically embedded canonical 3-fold. Clearly
there will be a large number of rejected records and hence this is very hit-
and-miss. However in practice it works very well.
This article started life as the third chapter of my Ph.D. thesis [F2] and
grew.

2 Acknowledgments
I would like to thank Miles Reid for all his help and A. Dimca and A. Parusin-
ski for many useful conversations. My thanks to Maria Iano and Duncan
Dicks for reading through previous versions and suggesting changes. I would
also thank all those at the Mathematics Institute, University of Warwick, the
mathematics department of the University of Leicester, and the Max-Planck-
Institut fur Mathematik, Bonn. I am grateful to Prof. Hirzebruch and the
institute for the invitation and their kind hospitality during 1987 and 1988.

3 Notation
All varieties will be assumed to be quasiprojective over an algebraically closed
eld K of characteristic zero. Let V be such a variety, of dimension m.
 K  is the multiplicative group of nonzero elements of K .
 Z, Q are the rings of integers and rational numbers respectively.
 Zr is the Abelian group f0; 1; : : : ; r 1g under addition modulo r.
 Zr is the group of units of Zr under multiplication modulo r.
 fa; : : : ; bb; : : : ; cg is a list with the element b omitted.
104 Working with weighted complete intersections

 Am is ane m-space.
 Pm is projective m-space.
 P(a0 ; : : : ; am) is used to denote weighted projective space with weighting
a0; : : : ; am . When there is no ambiguity this is denoted simply by P.
 V 0 is the nonsingular locus of V .
 OV is the sheaf of regular functions on V .

1V =
1V=K is the sheaf of regular 1-forms on V 0 .

nV = n
1V=K is the sheaf of regular n-forms on V 0 .
 !V =
mV is the sheaf of regular canonical dierentials on V 0 .
 KV is the canonical divisor corresponding to !V = OV (KV ).
 Let L be a coherent sheaf on V . Then
(a) hi (L) = hi (V; L) = dim Hi(V; L),
P
(b) (L) = i( 1)i hi(L)
(c) and 'L is the rational map corresponding to the sheaf L.
 Let D be a Cartier divisor on V . Then
(a) hi (D) = hi (OV (D)),
P
(b) (D) = i( 1)i hi(OV (D)).
(c) and 'D is the rational map corresponding to the sheaf OV (D).
 In particular 'nKV is called the nth canonical map.
 pg (V ) = h0 (!V ) is the geometric genus of V .
 Pn(V ) = h0 (!V
n) is the nth plurigenus of V . For negative n these are
referred to as the anti-plurigenera.
The words smooth and nonsingular are used interchangeably.
 A. R. Iano-Fletcher 105

Part I. Weighted complete intersections

4 Preamble
In this chapter we give a brief summary of the facts about weighted com-
plete intersections, along with many examples. We also prove necessary and
sucient conditions for a weighted hypersurface Xd in P(a0 ; : : : ; an) to be
quasismooth and well formed.
Sections 5 and 6 recap the main denitions and theorems about weighted
projective spaces and weighted complete intersections. Section 7 sets out
various facts about the cohomology of weighted complete intersections. Sec-
tion 8 contains necessary and sucient conditions for quasismoothness in
the hypersurface and codimension 2 cases. Information about cyclic quotient
canonical singularities in dimensions 1, 2 and 3 is to be found in Section 9,
along with two technical lemmas used to count points of intersection along
singular strata of P. Examples of how to calculate the singularities of various
weighted complete intersections are included in Section 10.

5 Denitions and theorems on weighted

projective spaces
We start by reviewing some denitions and notation concerned with weighted
complete intersections.
5.1 Denition Let a0 ; : : : ; an be positive integers; dene S = S (a0; : : : ; an)
to be the graded polynomial ring K [x0 ; : : : ; xn], graded by deg xi = ai . The
weighted projective space P(a0 ; : : : ; an) is dened by
P(a0 ; : : : ; an ) = Proj S:

5.2 Note Let x0 ; : : : ; xn be ane coordinates on A n+1 and let the group K 
act via:
(x0; : : : ; xn) = (a0 x0; : : : ; an xn):
Then P(a0 ; : : : ; an) is the quotient (A n+1 n f0g)=K  . Under this group ac-
tion x0 ; : : : ; xn are the homogeneous coordinates on P(a0 ; : : : ; an). Clearly
P(a0 ; : : : ; an ) is a rational n-dimensional projective variety.

5.3 Ane coordinate pieces Let fx0 ; : : : ; xng be the homogeneous co-
ordinates on P(a0 ; : : : ; an). The ane piece xi 6= 0 is isomorphic to A n = Zai .
Let " be a primitive ai th root of unity. The group acts via:
zj 7! "aj zj
106 Working with weighted complete intersections

for all j 6= i, on the coordinates fz0 ; : : : ; zbi; : : : ; zng of A n ; here zj is thought

of as xj =xiaj =ai . Compare this with the case of Pn where the ane coordinates
on xi 6= 0 are zj = xj =xi.
5.4 Examples (i) Pn = P(1; : : : ; 1).
(ii) Consider P(1; 1; 2) with homogeneous coordinates u, v and w. The ane
piece w = 1 is A 2 =Z2 with group action
u 7! u; v 7! v:
The coordinate ring R is given by:
R = K [u; v]Z2 = K [u2 ; v2; uv]
= K [x; y; z]=(xy z2 ):
Thus P(1; 1; 2) is the projective completion of the ordinary quadratic
cone xy = z2 in A 3 .
5.5 Lemma For all positive integers q we have
Proj S (a0; : : : ; an) ' Proj S (qa0; : : : ; qan):
Proof This follows from the fact that the 2 graded rings are isomorphic. 
From [EGA], Proposition 2.4.7 (see also [Hart], Exercise II.5.13) we have:
5.6 Lemma Let S be a graded ring; dene the truncation S (q) = Lm0 Sqm
to be the graded subring whose mth graded part is Sqm . Then there exists a
canonical isomorphism Proj S (q) ' Proj S .
This is called the qth Veronese embedding, and is used in the proof of the
following:
5.7 Lemma Let a0 ; : : : ; an be positive integers with no common factor. If
q = hcf(a1 ; : : : ; an) then
Proj S (a0; : : : ; an) ' Proj S (a0; a1=q; : : : ; an=q):
Proof Dene S 0 = Lm0 Sqm with the same grading as S . So S 0 ' S (q).
By the previous lemma we have Proj S 0 ' Proj S .
Suppose that xp0


xpnn is a monomial of degree mq for any m. Then
0

p0a0 +


+ pnan = qm, and so q j p0a0 . As the faig have no common factor,
q j p0. Hence x0 only appears in S 0 as x0q . Thus S 0 = K [x0q ; x1; : : : ; xn], which
is isomorphic to S (qa0; a1; : : : ; an). Therefore
Proj S (a0; : : : ; an) ' Proj S 0 ' Proj S (a0; a1=q; : : : ; an=q): 
 A. R. Iano-Fletcher 107

5.8 Quasire ections Let G be a nite group acting on a variety X . A

quasire ection is any element of G whose xed locus is a hyperplane. No
singularities are produced by the action of any group generated by quasi-
re ections.
The cancellation which occurs in Lemma 5.7 is nothing other than the
elimination of quasire ections from the actions of each Zai on the correspond-
ing ane coordinate piece.
This lemma leads to the following corollary from [WPS], 1.3.1 (see also
[De], Proposition 1.3):
5.9 Corollary P(a0 ; : : : ; an) ' P(b0 ; : : : ; bn ) for some fb0 ; : : : ; bng such that
hcf(b0 ; : : : ; bbi; : : : ; bn) = 1 for each i:

Proof By Lemma 5.5 we can cancel any common factor of the fai g. By
renumbering as necessary and by repeated applications of Lemma 5.7 we
can reduce P(a0 ; : : : ; an) to the case P(b0 ; : : : ; bn). A maximum of n + 1
applications of Lemma 5.7 are required. 

5.10 Examples (i) P(a; b) ' P1 for all a and b.

(ii) P(2; 3; 3) ' P(2; 1; 1).
(iii) Let f = x5 + y3 + z2 2 K [x; y; z] with weights 6, 10 and 15 respectively.
Dene X : (f = 0)  P = P(6; 10; 15). By the previous lemma P ' P2 .
P(6; 10; 15) ' P(6; 2; 3) ' P(3; 1; 3) ' P(1; 1; 1):

The monomials transform as:

(x5 ; y3; z2) 7! (x; y3; z2 ) 7! (x; y3; z) 7! (x; y; z):
Thus X  P ' (x + y + z = 0)  P2 = P1  P2 . Of course the
coordinate rings of the ane cones (see 6.1) over X  P and P1  P2
are not isomorphic.
In view of Corollary 5.9 we make the following:
5.11 Denition The expression P(a0 ; : : : ; an) is well formed if
hcf(a0 ; : : : ; abi; : : : ; an) = 1 for each i.
108 Working with weighted complete intersections

5.12 The quotient map Let T = K [y0 ; : : : ; yn], where the fyig all have
weight 1, and so Pn ' Proj T . Consider the inclusion map S ,! T given by:
xi 7! yiai for all i.
This induces a quotient map  : Pn ! P. In terms of the coordinates fYig on
Pn

[Y0; : : : ; Yn] 7! [Y0a0 ; : : : ; Ynan ]:

L
The map Pn ! P is a ramied Galois covering with Galois group i Zai .
5.13 Denition Let r > 0 and a1; : : : ; an be integers and let x1 ; : : : ; xn be
coordinates on A n . Suppose that Zr acts on A n via:
xi 7! "ai xi for all i,
where " is a xed primitive rth root of unity. A singularity Q 2 X is a
quotient singularity of type 1r (a1 ; : : : ; an) if (X; Q) is isomorphic to an analytic
neighbourhood of (A n ; 0)=Zr.
5.14 Notation Write Pi 2 P for the point [0; : : : ; 0; 1; 0; : : :; 0], where the
1 is in the ith position. We will call Pi a vertex, the 1-dimensional toric
stratum PiPj an edge, etc. The fundamental simplex (that is, the union of
all the coordinate hyperplanes P0 : : : Pbi : : : Pn) will be denoted by
.
5.15 The singular locus Psing of P Dene hi;j;::: = hcf(ai; aj ; : : : ). The
vertex Pi is a singularity of type a1i (a0 ; : : : ; abi; : : : ; an); this singularity is not
necessarily isolated. Each generic point P of the edge PiPj has an analytic
neighbourhood P 2 U which is analytically isomorphic to (0; Q) 2 A 1  Y ,
where Q 2 Y is a singularity of type h1i;j (a0 ; : : : ; abi; : : : ; abj ; : : : ; an). Similar
results hold for higher dimensional toric strata. The singularities only occur
on the fundamental simplex
.
Notice that codimP(Psing )  2.

6 Denitions and theorems on weighted com-

plete intersections
The rst few denitions come from [WPS].
6.1 Denition Let X be a closed subvariety of a weighted projective space
P and p : A n+1 nf0g ! P the canonical projection. The punctured ane cone
CX over X is given by CX = p 1(X ), and the ane cone CX over X is the
completion of CX in A n+1 .
 A. R. Iano-Fletcher 109

Notice that K  acts on CX to give X = CX =K  .

6.2 Lemma CX has no isolated singularities.
Proof If P 2 CX is singular then every point on the same bre of the
K  -action will be singular. 

6.3 Denition X in P(a0 ; : : : ; an) is quasismooth of dimension m if its ane

cone CX is smooth of dimension m + 1 outside its vertex 0.
When X  P is quasismooth, the singularities of X are due to the K  -
action and hence are cyclic quotient singularities. Notice that this denition
is not equivalent to the smoothness of the inverse image  1(X ) under the
Galois cover of Section 5.12 (e.g., X8 in P(2; 3; 5)).
Another important fact ([WPS], Theorem 3.1.6) is that a quasismooth
subvariety X of P is a V -manifold (that is, a complex space locally isomor-
phic to the quotient of a complex manifold by a nite group of holomorphic
automorphisms). This is used later to dene the canonical sheaf of X , which
is usually singular.
6.4 Denition Let I be a homogeneous ideal of the graded ring S and dene
XI to be:
XI = Proj S=I  P:
Suppose furthermore that I is generated by a regular sequence ffig of homo-
geneous elements of S . XI  P is called a weighted complete intersec-
tion of multidegree fdi = deg fi g. In this case, we denote by Xd1 ;:::;dc in
P = P(a0 ; : : : ; an ) a suciently general element of the family of all weighted
complete intersections of multidegree fdig.
Xd1 ;:::;dc in P(a0 ; : : : ; an) is of dimension n c. We will usually write
Cd1 ;:::;dc in P(a0 ; : : : ; ac+1) for a dimension 1 complete intersection and Sd1 ;:::;dc
in P(a0 ; : : : ; ac+2) for a surface.
6.5 Denition Xd in P(a0 ; : : : ; an) will be said to be a linear cone if d = ai
for some i (that is, the dening equation f can be written as f = xi + g).
Clearly in this case Xd in P(a0 ; : : : ; an) is isomorphic to P(a0 ; : : : ; abi; : : : ; an).
6.6 Examples (i) X46 in P(4; 5; 6; 7; 23) is a general element in the family
of all degree 46 hypersurfaces in P(4; 5; 6; 7; 23).
(ii) X8 in P(1; 1; 1; 1; 4) is a double cover of P3 branched along a smooth
octic surface.
110 Working with weighted complete intersections

6.7 The coecient convention When a general polynomial of a given

weighted homogeneous degree occurs in a calculation then it is usually written
without the nonzero coecients. For example the dening polynomial for X2
in P(1; 1; 1) is:
f = c0 x2 + c1xy + c2xz + c3y2 + c4yz + c5 z2
and will be simply written as:
f = x2 + xy + xz + y2 + yz + z2 :
6.8 The canonical sheaf !X All weighted complete intersections (and
weighed projective spaces) are V -manifolds (that is, locally quotients of A n
by a nite group action) and so the dualising sheaf !X is given by
!X ' i !X 0 ;
where i : X 0 ,! X is the inclusion of the smooth part X 0 into X . This sheaf
is a divisorial sheaf (see [R1], Appendix to Section 1, Theorem 7) and can be
written as
!X ' OX (KX );
where KX is a Q -Cartier divisor (that is, rKX is a Cartier divisor for some
nonzero integer r). In fact KX jX 0 is Cartier. For the general denition of
the canonical sheaf for varieties with at worst canonical singularities see [R4],
Section 1.4.
We now introduce an important concept which was not mentioned (and
possibly missed) by Dolgachev in [WPS].
6.9 Denition A subvariety X  P of codimension c is well formed if the
expression for P is well formed (see Denition 5.11) and X contains no codi-
mension c + 1 singular stratum of P.
This means that any codimension 1 stratum of X is either nonsingular on
P, or an intersection X \ S , where S is a codimension 1 stratum of P, that
is, codimX (X \ Psing )  2.
6.10 Well-formedness for hypersurfaces in P(a0 ; : : : ; an) The hyper-
surface Xd in P(a0 ; : : : ; an) is well formed if and only if
(i) hcf(a0; : : : ; abi; : : : ; abj ; : : : ; an) j d
(ii) hcf(a0; : : : ; abi; : : : ; an) = 1
 A. R. Iano-Fletcher 111

for all distinct i, j .

6.11 Well-formedness in codimension 2 The codimension 2 weighted
complete intersection Xd1;d2 in P(a0 ; : : : ; an) is well formed if and only if
(i) for all distinct i, j and k, with h = hcf(a0 ; : : : ; abi; : : : ; abj ; : : : ; abk ; : : : ; an),
either h j d1 or h j d2,
(ii) for all distinct i and j , with h = hcf(a0 ; : : : ; abi; : : : ; abj ; : : : ; an), then
h j d1 and h j d2,
(iii) hcf(a0 ; : : : ; abi; : : : ; an) = 1 for all i.
6.12 Well-formedness in higher codimensions The above conditions
can be generalised to higher codimensions. Xd1 ;:::;dc in P(a0 ; : : : ; an) is well
formed if and only if
(i) P(a0 ; : : : ; an) is well formed
(ii) for all  = 1; : : : ; c the highest common factor of any (n 1 c + ) of
the faig must divide at least  of the fdj g.
6.13 Note Dimca also denes well-formedness (see [Di]) under a dierent
name. He gives the following equivalent set of arithmetic conditions in the
quasismooth case. Dene:
m(h) = jfi : h j ai gj; k(h) = jfi : h j digj and
q(h) = dim X + 1 m(h) + k(h)
for all h 2 Z. Then the quasismooth weighted complete intersection Xd1 ;:::;dc
in P(a0 ; : : : ; an) is well formed if and only if q(p)  2 for all primes p. This
follows from a theorem essentially due to Hamm (see [Di], Proposition 2).
In fact a weighted complete intersection (not necessarily quasismooth) is
well formed if and only if q(h)  2 for all integers h  2. This is easy to show
from the conditions in Section 6.12.
6.14 The adjunction formulaPIf Xd P
;:::;dc in P(a0 ; : : : ; an ) is well formed
1
and quasismooth then !X ' OX ( di ai) (see [WPS], Theorem 3.3.4).
We dene the amplitude to be this dierence of sums, and usually denoted it
by .
6.15 Note The adjunction formula does not hold if the weighted complete
intersection is not well formed. We give two examples in dimensions 1 and 2
respectively.
112 Working with weighted complete intersections

(i) Consider the curve C7 in P(1; 2; 3). Let D  P2 be the curve  1 (C )

where  : P2 ! P is the quotient map (see Section 5.12) Then the curve
D is nonsingular of degree 7 and so is of genus 15. By the Hurwitz The-
orem (see [Hart], Corollary IV.2.4) we calculate that g(C ) = 1 and so
!C ' OC . On the other hand, since the amplitude is 1, this contradicts
the adjunction formula.
(ii) An example in dimension 2 is the surface S9 in P(1; 2; 2; 3). A quick
calculation shows that this surface is both quasismooth and nonsingular.
If it is well formed then the amplitude  = 1 and so KS2 = 34 . This
contradicts the fact that KS2 2 Z whenever S is nonsingular. In fact S9
in P is a smooth K3 surface.
6.16 Well-formedness in dimensions greater than 2 However, it
turns out that well-formedness only needs to be checked in dimensions 1
and 2. We have the following generalisation of a proposition due to Dimca
(see [Di], Proposition 6).
6.17 Theorem Let X = Xd1;:::;dc in P(a0 ; : : : ; an) be a quasismooth weighted
complete intersection of dimension greater than 2. Then
either X is well formed
or X is the intersection of a linear cone with other hypersurfaces (that is,
ai = d for some i and ).
6.18 Note (i) In case (ii) the weighted complete intersection is isomor-
phic to an intersection of lower codimension, that is, Xd1 ;:::;dc;:::;dc in
P(a0 ; : : : ; abi ; : : : ; an ) or possibly a weighted projective space.

(ii) Cases (i) and (ii) are not mutually exclusive. Consider the hypersurface
X2 in P(1; 1; 1; 1; 2) given by
X
f = z + xi xj :
i;j
This is both a linear cone and well formed, and is, of course, isomorphic
to P3 .
We need a preliminary result.
6.19 Lemma Let Z be the ane variety of all points P which satisfy the
determinantal condition:
01 1
g1 (P ) : : : g1m(P )
rank B
@ ... ... C
A  k;
1
gc (P ) : : : gc (P )
m
 A. R. Iano-Fletcher 113

where fgij g are general weighted homogeneous nonzero polynomials. If Z is

nonempty then codim Z  (m k)(c k).
This is an elementary fact (see [ACGH], p. 83).
6.20 Proof of Theorem 6.17 Let X = (f1; : : : ; fc)  P = P(a0 ; : : : ; an).
Suppose that P is well formed and assume that X is quasismooth with
dim X  3 but not well formed. So there is a singular stratum e of P
such that codimX (e \ X )  1.
If codimX (e \ X ) = 0 then X  e and so X is contained in some
coordinate hyperplane. Thus some of the dening polynomials are of the
form f = xi for some  and i. So X is the intersection of at least one linear
cone with other hypersurfaces.
So assume that codimX (e \ X ) = 1. By reordering we can assume that
e = (xk =


= xn = 0)  P
for some k. Let  = p 1 e  A n+1 n f0g, where p : A n+1 n f0g ! P is the
natural projection. Since codimX e = 1 then k = dim  = n c. As  is a
xed component of CX then we can write the ffg as:
X
n higher order terms
f = xigi (x0 ; : : : ; xk 1) + in xk ; : : : ; xn
i=k
for all  = 1; : : : ; c.
Dene MP to be the matrix
0 1
@f1 =@x0 (P ) : : : @f1 =@xn (P )
MP = B@ ... ... CA :
@fc =@x0 (P ) : : : @fc=@xn (P )
Singular points on CX occur whenever rank MP < c. Consider this matrix
restricted to :
0 1
0; : : : ; 0 g1k (P ) : : : gck (P )
MP 2 = B
@ ... ... ... C
A:
0; : : : ; 0 g1 (P ) : : : gc (P )
n n

So P 2  \ CX is singular whenever rank(gij )  c 1. Let Z be just this set.

If Z is empty then, in particular, 0 2= Z . As the entries of MP are all
weighted homogeneous polynomials, they must all be of degree 0. Thus,
using the coecient convention 6.7,
X higher order terms
f = xi + in xk ; : : : ; xn
114 Working with weighted complete intersections

for all  = 1; : : : ; c. So X is the intersection of a linear cone with other

hypersurfaces.
So assume that Z is nonempty. By the previous lemma, codim Z  n
k c + 2. Remembering that k = n c we have
dim Z  k (n k c + 2) = n c 2 = dim X 2  1:
Thus Z n f0g is nonempty and thus CX is not smooth away from the origin,
a contradiction. 

7 Cohomology of weighted complete

intersections
From [WPS], Section 3.4.3 we have:
7.1 Lemma Let X = (f1 ; : : : ; fc)  P(a0 ; : : : ; an) be a well-formed quasi-
smooth weighted projective complete intersection. Let A be the graded ring
S (a0; : : : ; an)=(f1; : : : ; fc) and An be the nth graded part of A. Then
8
>
<An if i = 0
H (X; OX (n)) ' >0
i if i = 1; : : : ; dim X 1
:A n  if i = dim X
for all n 2 Z.
In particular if S is a well-formed quasismooth weighted projective com-
plete intersection of dimension 2 then the following are equivalent:
(i) S is a K3 surface.
(ii) !S ' OS .
(iii) the amplitude  =  d
P P a = 0.
i i
For hypersurfaces we have the following result due to Steenbrink [S]:
7.2 Theorem Let X be the weighted hypersurface Xd in P(a0 ; : : : ; an) with
P a . Then
dening equation f and  = d i the Hodge structure is given by:
8
>
>0 if i + j 6= n 1 and i 6= j
>
<1   if i + j 6= n 1 and i = j
i;j
h (X ) = >dimK S(a0 ;:::;an ) if i + j = n 1 and i 6= j
>
>  f
 jd+
:dimK S(a0;:::;af
n) + 1 if i + j = n 1 and i = j
jd+
where f = (@f=@xi )i=0;:::;n is the Jacobian ideal of f .
 A. R. Iano-Fletcher 115

Proof This follows from [WPS], Section 4 and duality. 

7.3 Note The above formula satises the duality relations hi;j = hj;i =
hn 1 i;n 1 j for all i and j because
 S (a ; : : : ; a )   S (a ; : : : ; a ) 
dimK 0 n
= dimK 0 : n
jd+ f  f (n 1 j )d+
7.4 The Euler number The Euler number e(V ) of a variety V is dened
by
X
e(V ) = ( 1)i+j hi;j (V ):
i;j
For a smooth curve C we have e(C ) = deg KC = 2 2g. For a surface S,
with at worst Du Val singularities of types fQni gi where Q = A; B or E , we
have Noether's formula:
X
12(OS ) = KS2 + e(S ) + ni:
i
In particular the case of a K3 surface P S with Du Val singularities
P of types
fQni gi gives that h (S ) = 20 i ni and so e(S ) = 24 i ni.
1 ; 1
When X is a well-formed quasismooth weighted hypersurface of dimension
3 most of the Hodge numbers cancel or are zero and so
e(X ) = 2(1 h1;2 (X )):
7.5 Examples (i) The hypersurface S3 in P(1; 1; 1; 2) has Euler number
5. There are two ways to check this.
(a) It is easy to see that this surface has exactly one singularity, of
type 21 (1; 1) (that is, of Du Val type A1 ). Also the amplitude is 2
and KS2 = ( 2)2
32 = 6. By Noether's formula we have e(S3 ) = 5.
(b) Alternatively, the Hodge numbers are simple to calculate. Let w,
x, y and z be generators of weights 1, 1, 1 and 2 respectively in
S (1; 1; 1; 2). Then
 K [w; x; y; z] 
1 ; 1
h = dim (w2; x2 ; y2; w + x + y) = 2:
1
Thus the Hodge structure is:
hi;j i = 0 i = 1 i = 2
j=0 1 0 0
j=1 0 3 0
j=2 0 0 1
116 Working with weighted complete intersections

(c) Thus e(S3 ) = 1 + 3 + 1 = 5.

(ii) The hypersurface X10 in P(1; 1; 1; 2; 5) has the following Hodge struc-
ture.
hi;j i = 0 i = 1 i = 2 i = 3
j=0 1 0 0 1
j=1 0 1 145 0
j = 2 0 145 1 0
j=3 1 0 0 1
Let v, w, x, y and z be generators of weights 1, 1, 1, 2 and 5 respectively in
S (1; 1; 1; 2; 5). The only hard Hodge number is
 K [v; w; x; y; z] 
1 ; 2
h (X ) = dimK (v9; w9; x9 ; y4; z) = 145:
20
This gives an Euler number of 288.

8 Quasismoothness
In this section we prove conditions for quasismoothness for hypersurfaces and
codimension 2 weighted complete intersections.
First we consider the problem of a hypersurface.
8.1 Theorem The general hypersurface Xd in P = P(a0 ; : : : ; an) of degree
d, where n  1 is quasismooth if and only if
either (1) there exists a variable xi for some i of weight d (that is, X is a
linear cone)
or (2) for every nonempty subset I = fi0; : : : ; ik 1g of f0; : : : ; ng,
either (a) there exists a monomial xMI = xmi0 0


xmik k 11 of degree d,
or (b) for  = 1; : : : ; k, there exist monomials
xMI  xe = xmi0 0;


ximk k 11; xe
of degree d, where fe g are k distinct elements.
8.2 Note If Xd is a linear cone then f can be written as f = xi + g for some
xi and Xd is clearly quasismooth. So we need only consider the case where f
is not linear in any of the variables (that is, deg xi = ai 6= d for all i).
 A. R. Iano-Fletcher 117

Proof Assume that Xd in P is not a linear cone. Let F be the linear system
of all homogeneous polynomials of degree d with respect to the weights ai .
Let f 2 F be a suciently general polynomial. Dene Xd : (f = 0)  P.
i
C?X ! A n+1
??n f0g
?y y
i
Xd ! P

Note that the point 0 is a base point and is usually singular; as this point
does not lie in CX this does not aect quasismoothness. By Bertini's Theorem
(see [Hart], Remark III.10.9.2) the only singularities of the general CX lie on
the base locus of the linear system F . Any component of the base locus is
just a coordinate k-plane for some k = 0; : : : ; n. So the general hypersurface
Xd is quasismooth if and only if the general hypersurface CX is nonsingular
at each point of its intersection with every coordinate k-plane contained in
the base locus.
Let  be a coordinate k-plane for some k = 1; : : : ; n. By renumbering,
assume that  is given by xk =


= xn = 0, corresponding to the subset
I = f0; : : : ; k 1g. Let 0   be the open toric stratum where x0 ; : : : ; xk 1
are nonzero. Expand f in terms of the coordinates xk ; : : : ; xn:
Xn higher order terms
f = h(x0 ; : : : ; xk 1) + xi gi(x0 ; : : : ; xk 1) + in xk ; : : : ; xn :
i=k
Assume that one of conditions (a) and (b) hold for I . If (a) holds (that
is, h is nonzero) then  is not part of the base locus, and so by Bertini's
Theorem 0 contains no singular points. Geometrically this means that CX
intersects 0 transversally and so 0 is normal to the hypersurface at the
points of intersection.
Assume that only (b) holds. So h  0 and   CX . By (b) there are
at least k ofT the gi which are nonzero. Singular points occur exactly on the
locus Z = i (gi = 0)  0 , which is an intersection of at least k free linear
systems on 0 . Thus dim Z  0. As Z is a quasicone, it is at worst the origin
(compare Lemma 6.2). Therefore CX is nonsingular along 0 .
As one of these two conditions holds for every nonempty subset I , CX is
nonsingular.
Conversely assume that conditions (a) and (b) do not hold for all I . Let
I be a subset for which these two conditions fail. Without loss of generality
assume that I = f0; : : : ; k 1g. Let  be the corresponding coordinate
k-plane xk =


= xn = 0. As (a) and (b) do not hold
X n higher order terms
f = xi gi(x0 ; : : : ; xk 1) + in xk ; : : : ; xn
i=k
118 Working with weighted complete intersections

and at most k 1 of the gi are nonzero.

As above, singular points occur exactly on the intersection
\
Z= (gi = 0) \ :
ik

Since there are at most k 1 of the gi which are nonzero, dim Z  k (k 1) =

1. Thus Z is nonempty and so CX is singular on .
Therefore conditions (a) and (b) are both sucient and necessary for
quasismoothness when Xd in not a linear cone. 

8.3 Note (i) The only quasismooth cones are the linear cones. Suppose a
variable xi does not occur in the dening equation f . So CX ' CX 0  A 1
where X 0 : (f = 0)  P(a0 ; : : : ; abi; : : : ; an). Suppose that CX 0 has a
singularity at the origin. Thus CX 0  A 1 has a line of singularities along
0  A 1 ; a contradiction. So CX 0 is nonsingular at the origin and so f
must be linear in a variable; this is the linear cone case.
(ii) Without loss of generality we can assume in (b) that e 2 f0; : : : ; ng I ,
since otherwise this is condition (a).
(iii) For 2jI j  n + 1 condition (b) implies condition (a), since there are
simply not enough variables xi .
(iv) Condition (b), with jI j = 1, of the theorem gives that for all i = 0; : : : ; n
there must exist a monomial xni xei , for some ei, of degree d. This is
equivalent to requiring that CX is smooth along the coordinate axes
(that is, Xd is quasismooth at the vertices) and is in practice the most
substantial case. Weighted hyperspaces (and polynomials) which satisfy
this condition will be said to be semi-quasismooth.
(v) CX contains no coordinate stratum of dimension  (n + 1)=2 except
possibly in the linear cone case.
So we have the following corollaries for curves, surfaces and 3-folds.
8.4 Corollary The curve Cd in P(a0 ; a1 ; a2), where d > ai , is quasismooth if
and only if the following hold for all i:
(1) there exists a monomial xni xei , for some ei , of degree d.
(2) there exists a monomial of degree d which does not involve xi.
 A. R. Iano-Fletcher 119

Proof Since d > ai for all i, Xd is not a linear cone. Conditions (1) and (2)
come from considering the conditions of the above theorem for jI j = 1 and
jI j = 2 respectively. 
The proofs of the following corollaries are similar to the above.
8.5 Corollary The surface Sd in P(a0 ; : : : ; a3), where d > ai, is quasismooth
if and only if the following hold:
(1) for all i there exists a monomial xnixei for some ei of degree d.
(2) for all distinct i, j
either there exists a monomial xmi xnj of degree d,
or there exist monomials xni 1 xmj 1 xe1 and xni 2 xmj 2 xe2 of degree d such
that e1 and e2 are distinct.
(3) there exists a monomial of degree d which does not involve xi .
8.6 Corollary The 3-fold Xd in P(a0 ; : : : ; a4), where d > ai, is quasismooth
if and only if the following hold:
(1) for all i there exists a monomial xnixei of degree d.
(2) for all distinct i, j
either there exists a monomial xmi xnj of degree d,
or there exist monomials xni 1 xmj 1 xe1 and xni 2 xmj 2 xe2 of degree d such
that e1 and e2 are distinct.
(3) there exists a monomial of degree d which does not involve either xi or
xj .
In the codimension 2 case we have:
8.7 Theorem Suppose the general codimension 2 weighted complete inter-
section Xd1 ;d2 in P = P(a0 ; : : : ; an), where n  2, of multidegree fd1; d2g is
not the intersection of a linear cone with another hypersurface. Xd1 ;d2 in P is
quasismooth if and only if for each nonempty subset I = fi0 ; : : : ; ik 1g of
f0; : : : ; ng one of the following holds:
(a) there exists a monomial xMI 1 of degree d1 and there exists a monomial
xMI 2 of degree d2
(b) there exists a monomial xMI of degree d1, and for  = 1; : : : ; k 1 there
exist monomials xMI m uxem u of degree d2, where feg are k 1 distinct
elements.
120 Working with weighted complete intersections

(c) there exists a monomial xMI of degree d2 , and for  = 1; : : : ; k 1 there

exist monomials xMI m uxem u of degree d1, where fe g are k 1 distinct
elements.
(d) for  = 1; : : : ; k, there exist monomials xIM xe1 of degree d1 , and xIM xe2
1 2

of degree d1, such that fe1g are k distinct elements, fe2 g are k distinct
elements and fe1 ; e2g contains at least k + 1 distinct elements.
Proof Let F1 and F2 be linear systems of all homogeneous polynomials of
degrees d1 and d2 respectively with respect to the weights a0 ; : : : ; an. Let
f1 2 F1 and f2 2 F2 be suciently general polynomials. Dene
X = Xd1 ;d2 : (f1 = f2 = 0)  P:
We have the following commutative diagram:
i
C?X ! A n+1
??n f0g
?y y
i
X ! P

The only singularities that can occur in the general member of the family
occur on the coordinate strata. So as in the proof of quasismoothness for
hypersurfaces, X is quasismooth if and only if CX is smooth along all the
coordinate strata.
Assume that one of conditions (a), (b), (c) or (d) holds for each nonempty
subset I . Let  be a coordinate k-plane for some k. By renumbering, we can
assume that  is given by xk =


= xn = 0, corresponding to the subset
I = f0; : : : ; k 1g. As before let 0 be the open toric strata where x0 ; : : : ; xk 1
are all nonzero. Expand both f1 and f2 in terms of the coordinates xk ; : : : ; xn:
X
n higher order terms
f = h(x0 ; : : : ; xk 1) + xigi (x0 ; : : : ; xk 1) + in xk ; : : : ; xn
i=k
for  = 1; 2.
Suppose (a) holds. So h1 and h2 are nonzero on 0. If either h1 or h2
involves only one monomial then 0 \ CX is empty. This includes the case
when k = 1. So without loss of generality assume that h1 and h2 each involve
at least 2 monomials and hence k  2. 0 is not part of the base locus of
F1 or F2 . By Bertini's Theorem (f1 = 0) and (f2 = 0) are nonsingular on
0. Since (h1 = 0) and (h2 = 0) are free linear systems on 0, (h1 = 0) and
(h2 = 0) intersect transversally. Thus, at each point of (h1 = h2 = 0) \ 0,
there exist two distinct normals. Therefore CX is nonsingular along 0.
 A. R. Iano-Fletcher 121

Suppose (b) holds. So h1 is nonzero and there are at least k 1 of the

fg1i g
which are nonzero. So 0 is not part of the base locus for F1 , and so
by Bertini's Theorem we have that (f1 = 0) is nonsingular on 0 . Singular
points occur exactly on the locus
\
Z = (h1 = 0) (g2i = 0)  0;
i
which is an intersection of at least k 1 free linear systems on (h1 = 0) \ 0 .
Thus dim Z  0 and hence is at worst the origin. Therefore CX is nonsingular
along 0 .
The case where condition (c) holds is similar to the case for condition (b).
Suppose that only condition (d) holds. We have
X
n higher order terms
f = xigi (x0 ; : : : ; xk 1) + in xk ; : : : ; xn
i=k
for  = 1; 2. The normal directions, perpendicular to the plane , to the
hypersurfaces are (g1k ; : : : ; g1n) and (g2k; : : : ; g2n). Dene the matrix MP by
gk(P ) : : : gn(P )
MP = 1k 1 :
g2 (P ) : : : g2 (P )
n

Singular points occur exactly on the locus Z = fP : rank MP  1g. As there

are at least k monomials of the form xMI xe of degree d, at least k of the fgi g
are nonzero. As these are free on 0 , each row of the matrix MP is nonzero
for each P 2 0 . Furthermore this matrix for any P 2 Z has at least k + 1
nonzero columns, since there are at least k + 1 distinct elements in fe1 ; e2g.
By renumbering we can assume that the rst k + 1 columns of M P are not
identically zero on 0 .
Fix P 2 0. Without loss of generality we can assume that g1k (P ) 6= 0.
If g2k (P ) = 0 then g2i (P ) 6= 0 for some i > k, and so M P has rank 2. In this
case P 2 CX is nonsingular. Suppose that g2k (P ) 6= 0. Dene a = g1k (P ),
b = g2k (P ) and
\
ZP = (ag2i (Q) bg1i (Q) = 0)  0 :
i>k
Notice that P 2 ZP if and only if rank MP  1, which is equivalent to P 2 CX
being singular. Since ZP is the intersection of k free linear systems on 0 ,
dim ZP  0 and so ZP is at worst the origin. In particular P 2= ZP and hence
P 2 CX is nonsingular. Therefore CX is nonsingular along 0 .
As one of these four conditions holds for every nonempty subset I , CX is
nonsingular.
122 Working with weighted complete intersections

Conversely assume that none of the conditions (a), (b), (c) or (d) hold
for some nonempty subset I . Without loss of generality we can assume that
I = f0; : : : ; k 1g. Let  be the corresponding coordinate plane xk =


=
xn = 0. There are three cases:
(i)  6 CXd1 . So h1 is nonzero and there are at most k 2 of the fg2i g
T are nonzero. The singular points are exactly the locus Z = (h1 =
which
0) i(g2i = 0). However
dim Z  k (k 2) 1 = 1;
and so Z contains more than the origin. Thus CX is singular along .
(ii)  6 CXd2 . Similarly in this case CX is singular along .
(iii)   CXd1 \ CXd2 . In this case both h1 and h2 are identically zero. So
X
n higher order terms
f = xi gi (x0 ; : : : ; xk 1) + in xk ; : : : ; xn
i=k
for  = 1; 2. As condition (d) does not hold, one of two cases occurs:
either for some  there are at mostTk 1 of the fgi g which are nonzero.
Thus the intersection Z = i(gi = 0) has dimension at least 1
and so these fgi g have a common solution. Therefore the matrix
 g k (P ) : : : g n (P ) 
MP = 1k 1
g2 (P ) : : : g2 (P )
n

has rank less than 2 for some P 2 Z and hence CX is singular
along .
or there are at most k distinct elements in fe1 ; e2g. Thus there are
at most k nonzero columns in the matrix MP . Let Z = fP :
rank MP  1g. Therefore
dim Z  k (k 1) = 1;
and so contains more than just the origin. Therefore CX is singular
along .
So if one of these four conditions are not satised for every subset I then CX
is singular. 
8.8 Corollary Suppose Xd ;d in P is quasismooth and is not the intersection
1 2
of a linear cone with another hypersurface. We have the following:
 A. R. Iano-Fletcher 123

(i) Every variable xi occurs in at least one of the dening equations.

(ii) All but at most one variable are in both equations.
(iii) If xi does not appear in one dening equation then there exists a mono-
mial xmi occurring in the other equation.
Proof
(i) This follows from the previous theorem with jI j = 1.
(ii) Suppose, after renumbering, that x0 and x1 are not involved in f1 . Then
none of the conditions can hold for I = f0; 1g, a contradiction.
(iii) Suppose that xi does not appear in f1 . Conditions (a), (b) and (d)
cannot hold and so there must be a monomial xmi of degree d2. Geo-
metrically if one of the hypersurfaces is singular along a coordinate axis,
because the equation fi does not involve that variable, then the other
hypersurface cannot pass through that axis. 

9 Cyclic singularities and counting points

In this section we give combinatorial conditions for cyclic quotient singulari-
ties to be isolated and canonical (see [R4], Denition 1.1 for the denitions of
canonical and terminal singularities). The last two lemmas of this section are
used to count the number of intersections along 1 and 2 dimensional strata.
We also give an alternative proof of the rst of these lemmas in terms of the
Minkowski mixed volume of integral polyhedra.
9.1 Lemma A canonical curve point is smooth.
This is clear since canonical singularities are normal. For dimension 2 we
have:
9.2 Lemma The following are equivalent:
(i) Q in S is a cyclic quotient canonical surface singularity.
(ii) Q is of type 1r (a; a) for some index r and a coprime to r.
(iii) Q is of type 1r (1; 1) for some index r.
The above singularities are Du Val singularities of type Ar 1.
For 3-folds we have the following due to White, Morrison, Stevens, Danilov
and Frumkin:
124 Working with weighted complete intersections

9.3 Lemma The following are equivalent:

(i) S is an isolated cyclic quotient terminal 3-fold singularity.
(ii) S is of type 1r (b0 ; b1; b2 ), for some positive integers r, b0 , b1 , b2, with
r  2, r and bi coprime and r j bi + bj for a pair of distinct i, j .
(iii) S is of the form 1r (1; 1; b) for some r  2 and b coprime to r.
The following two lemmas are very useful for calculating the number and
arrangement of singularities on a complete intersection.
9.4 Lemma Let x and y be of weight a0 and a1 respectively, such that
hcf(a0 ; a1) = 1. Suppose f (x; y) is a homogeneous polynomial of degree d,
semi-quasismooth (see Note 8.3, iv) and suciently general. Let P0 = [1; 0]
and P1 = [0; 1]. Then Xd : (f = 0) in P(a0 ; a1) is a nite set and:
(i) Pi is in Xd if and only if ai - d for i = 0; 1,
(ii) there are exactly b a0da1 c other points in Xd .
Proof Notice that xa =ya is an invariant of the group action of K  on A 2 0
1 0

which denes P(a0 ; a1). There are four cases:

(i) a0 j d and a1 j d. Then f is of the form
f = xd=a0 +


+ yd=a1 ;
written using the coecient convention (see Section 6.7). So
f = xa1
 d=a a 0 1

yd=a1 y0a +


+ 1;

which has exactly a0da1 roots.

(ii) a0 - d and a1 j d. Since Xd is semi-quasismooth, f is of the form
f = y(x(d a1 )=a0 +


+ y(d a1 )=a1 ):

The solution y = 0 gives the point P0.

 (d
f = xa1 a1 )=a0 a1
yd=a1 y0a +


+ 1:

This has exactly n = da0 aa11 roots. So d = na0 a1 + a1. As a0 - d then

a0 > 1, and so a1 < a0 a1. Thus n = b a0da1 c.
 A. R. Iano-Fletcher 125

(iii) a0 j d and a1 - d. Similar to (ii).

(iv) a0 - d and a1 - d.
f = xy(x(d a0 a1 )=a0 +


+ y(d a0 a1)=a1 )
So the two vertices P0 and P1 are solutions. Also
 
f = xa1 (d a0 a1 )=a0 a1 +


+ 1;
xyd=a1 y0a
which has exactly n = d aa00a1 a1 roots on P n fP0; P1g. So d = na0 a1 +
(a0 + a1 ). As a0 - d and a1 - d then a0 ; a1  2 and not both equal to 2.
Thus
a0 a1 = (a0 1)(a1 1) 1 + a0 + a1a0 + a1 :
Therefore n = b a0da1 c. 

9.5 Lemma Suppose that x0 , x1 and x2 have weights a0 , a1 and a2 , where

hcf(a0 ; a1; a2 ) = 1. Let f and g be suciently general semi-quasismooth ho-
mogeneous polynomials in K [x0 ; x1 ; x2] of degrees d and e respectively. Sup-
pose that Xd;e : (f = 0; g = 0) in P(a0 ; a1; a2 ) is a nite set. Let
 ni;j be the number of points of Xd;e along the edge PiPj ,
 hi;j = hcf(ai; aj ),
 ni be the number of points at the vertex Pi (that is, ni = 0; 1),
 N be the number of points in P n
.
Then:
de = X ni + X ni;j + N:
a0 a1 a2 i ai i>j hi;j
9.6 Note (i) Xd;e in P is not automatically nite (consider for example
X5;9 in P(1; 2; 4)).
(ii) Similar results hold for higher codimensions and involve induction on
the dimension.
(iii) Notice that Lemma 9.4 can be deduced from the above (consider Xd;1
in P(a0 ; a1 ; 1)).
(iv) This also has connections with the Minkowski mixed volumes of Newton
polyhedra (see after the proof).
126 Working with weighted complete intersections

Proof Let  : P2 ! P be the quotient map dened in Section 5.12. Let

F = f and G = g. Since Xd;e is nite, V (F ) and V (G) have no common
components. By Bezout's theorem Y = V (F; G) in P2 consists of exactly de
points counted with multiplicity.
The restriction of  to P2 n
is a0a1 a2-to-1, onto P n
. As there are N
points on P n
this accounts for a0 a1 a2N points on P2 n
.
The restriction of  to the line QiQj is ai aj =hi;j -to-1, onto PiPj . Without
loss of generality assume that hi;j j d but that hi;j - e. Let k be such that
fi; j; kg = f0; 1; 2g. Notice that xk j g, or else there would exist a monomial
xai xbj of degree e, contradicting hi;j - e. Then f and g are of the form:
f = xmi xj + xmj xi +



g = xk (xni 0 + xmj 0 +


):
Thus F and G are of the form:
F = Ximai Xjaj + Xjnaj Xiai +



0
G = Xkak (Xin0ai + Xjm aj +


):
We localise F and G by setting Xi = 1, to give the corresponding ane
equations F and G. Let [Xi ; Xj ; Xk ] = [1; ; 0] be a point of intersection
along the line Qi Qj . The multiplicity  of the intersection is given by:
 = mult(F; G; [1; ; 0])
= mult(F; G; (; 0))
= mult(Xiai + Ximai +


; Xkak ; (; 0))
= mult(Xi0 +


; Xkak ; (0; 0))
= ak ;
where Xi0 = Xi  . So this line contributes (ni;j ak )aiaj =hi;j points (counted
with multiplicity) to Bezout's theorem.
Consider the vertex Qi . If Pi is contained in X then ai - d and ai - e. As
X is semi-quasismooth, ai j d aj and ai j e ak for distinct i, j , and k. So
f and g are of the form:
f = xni xj +



g = xmi xk +



Thus:
F = Xinai Xj +



G = Ximai Xk +



 A. R. Iano-Fletcher 127

The intersection multiplicity  at Qi is:

 = mult(F; G; Qi):
Localising at Xi = 1 gives:
 = mult(F; G; (0; 0))
= mult(Xjaj +


; Xkak +


; (0; 0))
= aj ak :
Clearly Xjaj and Xkak are the smallest degree monomials in F and G. So this
gives a contribution of aj ak ni.
Combining the above gives:
X X ni;j aiaj ak
de = niaj ak + h + Na0 a1a2 ;
distinct i;j;k i>j; k6=i;j i;j

which rearranges to give the formula in the lemma. 

An alternative proof of the above two lemmas is via Newton polyhedra
and the Minkowski mixed volume (see both [Be] and [Ku]).
9.7 Denition An integral polyhedron S is a polyhedron in R n with vertices
in Zn. The n-dimensional volume of S will be denoted by Vn(S ), where the
volume of the unit parallelepiped is 1.
9.8 Denition For each m = (m1 ; : : : ; mn) 2 Zn dene
xm = xm1


xmn n :
1

Let f 2 K [x1 ; x1 1; : : : ; xn; xn 1 ] be a Laurent polynomial. Then

X
f= cmxm ;
m2Zn
where all but a nite number of the fcmg are zero. The Newton polyhedron
Newton(f ) of f is the convex hull of fm 2 Zn : cm 6= 0g, and is an integral
polyhedron.
9.9 Denition Let S = fSi : i = 1; : : : ; ng be a set of integral polyhedra.
The Minkowski mixed volume V (S ) of S is given by:
X X
V (S ) = ( 1)n 1 Vn(Si) + ( 1)n 2 Vn(Si + Sj ) +
i>j



+ Vn(S1 +


+ Sn)
where Si + Sj = fsi + sj : si 2 Si; sj 2 Sj g.
128 Working with weighted complete intersections

This is the classical formula, up to a multiple of n!.

Let T n be the n-dimensional torus (K  )n. This corresponds to the open
toric stratum in P. Let F be a system of n suciently general Laurent
polynomials ffi : T n ! K g with corresponding Newton polyhedra S = fSig.
The roots of these n polynomials in T n are isolated. Let L(F ) be the number
of such roots, counted with multiplicity. Then [Be], Theorem A gives:
L(F ) = V (S ):
9.10 Alternative proof of Lemma 9.4 Let Td=a 1 be the torus x0 x1 6= 0 in
P = P(a0 ; a1 ). Suppose that a0 ; a1 j d. Then f = x0 0
+


+ xd=a
1 . So
1

Nf = Newton(f ) = [(d=a0; 0); (0; d=a1)];

where [P; Q] denotes the line segment in Z2 from P to Q. So V1(Nf ) + 1 is
the number of integral points on Nf , that is, the number of solutions to
f(;  ) 2 Z2 :   0;   0; a0 + a1 = dg:
For a solution (;  ) we have  = (d a1 )=a0 2 Z, that is, d  a1 mod
a1. As a0 and a1 are coprime, then a1 is invertible modulo a0 , with inverse s.
So   ds mod a0 , that is,  = ds + na0 for some n. Also 0    d=a1. So
ds  n  d ds :
a0 a0 a1 a0
There are a0da1 + 1 such solutions. Thus f has a0da1 roots on the torus T 1 in P.
Similarly when a0 - d, etc. 
Lemma 9.5 can be proved using analogous methods.

10 Determination of singularities on
weighted complete intersections
In this section we shall determine the singularities of three weighted complete
intersections, presenting the calculations in detail. These examples are a
good introduction to the theorems giving arithmetic conditions for weighted
complete intersections to have at worst isolated canonical singularities.
10.1 The surface S = S36 in P(7; 8; 9; 12) We shall see that this surface
has four singularities, one each of type A2 , A3, A6 and A7. The Euler number
of such a K3 surface is 6, the lowest Euler number found in any of the lists
of weighted complete intersection K3 surfaces.
 A. R. Iano-Fletcher 129

Let w, x, y and z be the homogeneous coordinates on P = P(7; 8; 9; 12)

of weights 7, 8, 9 and 12 respectively. Let f be a general polynomial of
homogeneous degree 36. Using the coecient convention (see Section 6.7) we
have:
f = w4x + x3 z + y4 + z3 + others:
So S is well formed and, by Theorem 8.1, is quasismooth. Hence its singular-
ities arise only due to the singularities of P and occur only on the edges and
vertices of P. Consider the vertices.
(1) P0: f contains no monomial of the form wn for any n and so P0 2 S .
Consider the ane piece (w = 1). The point P0 2 S looks like:
(fe = f (1; x; y; z) = x +


= 0)  A 3 ="
where " is a primitive 7th root of unity and acts on the coordinates of
A 3 via:
x 7! "8x = "x
y 7! "9y = "2y
z 7! "12z = "5z:
Notice that @f =@x = w4 +


is nonzero at P0. By the Inverse Function
Theorem y and z are local coordinates around P0 2 S . This gives a
singularity of type 71 (2; 5), which is Du Val of type A6.
(2) P1: Again f contains no monomial of type xn and so P1 2 S . As above,
this gives a Du Val singularity of type A7.
(3) P2, P3: Since f contains the monomials y4 and z3 then P2 , P3 62 S .
There are only two singular edges in P, P1P3 which is analytically iso-
morphic to K   14 (3; 1) and P2P3 which is K   31 (2; 1).
(4) P1P3 : Since f jP1P2 = x3z + z3 = z(x3 + z2 ) then S does not contain the
edge P1P3. As x 6= 0 and z 6= 0 on the edge P1P3 then the ane piece
(z = 1) contains all of the intersection points. Since (@f =@x)jz=1 =
x2 +


is nonzero then w and y are local coordinates on S at each
of the points of S \ P1P3 . This is clear geometrically since S is a
general element of all degree 36 hypersurfaces and so it must cross this
line transversally. Thus each point is a singularity locally analytically
isomorphic to A 2 =" where the coordinates of A 2 are w and y and " is a
4th root acting via:
w 7! "7w = "3w
y 7! "9y = "y:
This gives a Du Val singularity of type A3 .
130 Working with weighted complete intersections

(5) We must now count the number of intersection points on this edge.
Each point of the intersection is given by the equation x3 + z2 = 0 in
P(8; 12). This is just X24 in P(8; 12), that is, X6 in P(2; 3). Either from
rst principles or from Lemma 9.4 we can see that this is exactly one
point.
(6) P2P3: As above, there is exactly one Du Val singularity, which is of type
A2, along this edge.
10.2 The 3-fold X = X46 in P(4; 5; 6; 7; 23) The canonical 3-fold hyper-
surface X46 in P(4; 5; 6; 7; 23) has 3 singularities of type 12 (1; 1; 1), 1 of type
1 (3; 1; 1), 1 of type 1 (4; 1; 2), 1 of type 1 (5; 1; 1), and 1 of type 1 (6; 1; 3).
4 5 6 7
The singularities are checked as follows. Let v, w, x, y and z be the
homogeneous coordinates of P = P(4; 5; 6; 7; 23) of weights 4, 5, 6, 7 and 23
respectively. Let f be a general polynomial of homogeneous degree 46. Then,
using the coecient convention, we write f in the form:
f = v10 x + w8x + x7 v + y6v + z2 + others:
This is well formed and quasismooth (see Theorem 8.1). So the singularities
of the hypersurface occur only on the edges and at the vertices of P. Consider
the vertices in reverse order:
(1) P4: Since f contains the monomial z2 with nonzero coecient, f (P4) 6=
0 and so P4 62 X46 .
(2) P3: There is no monomial of the form yn for any n in f , and so P3 2 X46.
Consider the ane piece (y = 1). P3 2 X46 looks like:
(fe = f (v; w; x; 1; z) = v +


= 0)  A 4 =";
where " is a primitive 7th root of unity and acts as:
(v; w; x; z) 7! ("4v; "5w; "6x; "23 z):
Notice that @f=@v = y6 +


is nonzero at P3. By the Inverse Function
Theorem w, x and z are local coordinates on X46 around P3 2 X46.
Thus the singularity here is of type 71 (5; 6; 23). This is equivalent to
1
7 (6; 1; 3), which is terminal.
(3) P2: Again there is no monomial of the form xn for any n in f , and so
P2 2 X46. Consider the ane piece (x = 1). P2 2 X46 looks like:
(fe = f (v; w; 1; y; z) = v +


= 0)  A 4 =";
 A. R. Iano-Fletcher 131

where " is a primitive 6th root of unity and acts as:

(v; w; y; z) 7! ("4v; "5w; "7y; "23z):
Notice that @f=@v = x7 +


is nonzero at P3. By the Inverse Function
Theorem, w, y and z are local coordinates on X46 around P2 2 X46 .
Thus the singularity here is of type 61 (5; 7; 23). This is equivalent to
1 (5; 1; 1), which is terminal.
6
(4) P1: P1 2 X46 is locally f = x +


= 0 and gives a terminal singularity
of type 51 (4; 1; 2).
(5) P0: P0 2 X46 is locally f = x +


= 0 and gives a terminal singularity
of type 41 (3; 1; 1).
Consider the edges of P. An edge PiPj is singular if and only if h =
hcf(ai ; aj ) 6= 1. In which case it is analytically equivalent to K  
h (a0 ; :: : ; ab1i ; : : : ; abj ; : : : ; a4 ). So only the edge P0 P2 is singular and looks
1
like K  2 (1; 1; 1). Since 2 = hcf(4; 6) j 46, the hypersurface does not
contain this line. Lemma 9.4 is used on X46 in P(4; 6), after cancelling
the common factor, to give three points of intersection. Alternatively,
fjP0P2 = uxg36(u; x) = uxg3(u3; x2 );
where g36 and g3 are polynomials of degree 36 and 3 respectively. There
are exactly three solutions to g3 = 0, and so there are three points
of intersection. So X46 crosses P0P2 transversally and hence there are
three singularities, each of type 12 (1; 1; 1), along P0P2 .
10.3 The 3-fold X12;14 in P(2; 3; 4; 5; 6; 7) The family of codimension two
complete intersections X12;14 in P(2; 3; 4; 5; 6; 7) is an anticanonically embed-
ded Fano 3-fold having only the following isolated terminal singularities: 1 of
type 15 (4; 1; 2), 2 of type 13 (2; 1; 1) and 7 of type 21 (1; 1; 1).
The singularities are checked as follows. Let u, v, w, x, y and z be the
homogeneous coordinates of weights 2, 3, 4, 5, 6 and 7 respectively. Let
f , g be homogeneous polynomials of degrees 12 and 14 respectively. Then
X = (f = g = 0)  P = P(2; 3; 4; 5; 6; 7).
Consider the vertices of the weighted projective space P. Since 5 - 12 and
5 - 14, P3 2 X . So
f = x2 u +



g = x2 w +



Thus fv; y; zg are local coordinates around P3, which is therefore a singularity
of type 51 (3; 6; 7), that is, 15 (4; 1; 2). There are no other vertices contained in
X.
Consider the 1-dimensional loci of P.
132 Working with weighted complete intersections

(1) P0P2: h = hcf(2; 4) = 2 and

f = u6 + w3 +



g = u7 + w2y +



Then the local coordinates are fv; x; zg and the singularities are of type
1
2 (1; 1; 1). There are three such intersection points (by Lemma 9.4 ap-
plied to X6 in P(1; 2)).
(2) P0P4: Likewise h = hcf(2; 6) = 2 and
f = u6 + y2 +



g = u7 + u5w + y2u +



(f = 0) in P(1; 3) is two points by Lemma 9.4. So there are two singu-
larities, each of type 21 (1; 1; 1), along P0P4.
(3) P2P4: There is exactly one singularity on this line, of type 12 (1; 1; 1).
(4) P1P4: This time h = hcf(3; 6) = 3 and
f = v4 + y2 +



g = v4u + y2u +



So there are two of type 31 (1; 1; 1) on P1P4 .
Consider the only singular 2-dimensional locus, P0P2 P4, of P where h =
hcf(2; 4; 6) = 2. By Lemma 9.5, there are seven intersection points (some of
which have already been counted), all of type 12 (1; 1; 1).

Part II. Lists of various weighted complete

intersections
11 Preamble
The aim of this chapter is to produce lists of hypersurface and codimen-
sion 2 weighted complete intersections of dimension at most 3 with at worst
isolated canonical singularities. We present various theorems giving combi-
natoric conditions on the weights and degrees of such intersections. From
these conditions we can produce lists of intersections (along with their corre-
sponding singularities). In most cases a computer was used for its speed and
inability to become bored.
 A. R. Iano-Fletcher 133

Sections 12 and 13 treat the cases of dimension 1 and 2 respectively and

give corresponding lists. Section 14 deals with the 3-fold case (both hyper-
surfaces and codimension 2) and Sections 15 and 16 deal with the particular
cases of canonical 3-folds and Q -Fano 3-folds respectively. Section 18 gives an
alternative method for producing canonically and anticanonically embedded
3-fold complete intersections using the Poincare series of a ring.

12 Weighted curve hypersurfaces

12.1 Theorem A weighted curve complete intersection is smooth if and only
if it is quasismooth.

Proof Any 1-dimensional cyclic quotient singularity is of type 1r (a) for some
coprime r and a. Let x be the coordinate on A 1 . The group Zr acts via:
x 7! "a x;
where " is a primitive rth root of unity. So
= ' Spec K [x]Zr ' Spec K [xr ] ' Spec K [x] ' A 1 :
A 1 Zr

So this is nonsingular. Notice that this group action is just a quasire ection
(see Section 5.8). 
From [OW], Corollary 3.5 we have a formula for the genus of dimension 1
hypersurfaces.
12.2 Theorem Let Cd in P(a0 ; a1 ; a2) be a nonsingular curve. Then the
genus g is given by:
2 X hcf(ai; aj ) X
2 !
g = 21 a ad a d ai aj
hcf(d; ai)
+ ai 1 :
0 1 2 i>j i=0

12.3 Theorem A weighted curve Cd in P(a0 ; a1 ; a2) is well formed, not a

linear cone and quasismooth if and only if for each i the following three
conditions hold:
(1) ai < d,
(2) ai j d,
(3) hcf(ai ; aj ) = 1 for all distinct i, j .
134 Working with weighted complete intersections

Proof C is well formed if and only if ai j d for all i and hcf(ai; aj ) = 1 for
all distinct i, j (see Section 6.10). These are conditions (2) and (3).
Suppose C is not a linear cone and quasismooth. Then conditions (1)
holds. Also ai j d ae for some e. But this is already satised by condition
(2).
The converse follows immediately from conditions (1), (2) and (3). 
12.4P Smooth weighted curve hypersurfaces with amplitude  =
d ai = 0 We list the only smooth weighted curves of codimension 1 with
 = 0 satisfying the above conditions.
Curve D
C3  P(1; 1; 1) 3P
C4  P(1; 1; 2) 2P
C6  P(1; 2; 3) P
All are elliptic curves (that is, g = 1 and ! ' OC ) and are given by Proj RC
where RC is:
M
RC = H0 (OC (nD));
n0
and D is given in the above table.
12.5 The calculation The above curves are the only ones satisfying the
conditions of Theorem 12.3. This is demonstrated as follows.
Order the faig by a0  a1  a2. conditions (2) and (3) of Theorem 12.3
give a0 a1a2 j d. Let d = a2. As  = 0 then 3a2  a0 + a1 + a2 = d = a2.
So   3 (that is,  = 2; 3).
(i)  = 2. So a0a1 j 2. Either (a0 ; a1) = (1; 1) (that is, C4 in P(1; 1; 2)) or
(a0; a1 ) = (1; 2) (that is, C6 in P(1; 2; 3)).
(ii)  = 3. So a0a1 j 3. Either (a0 ; a1) = (1; 1) (that is, C3 in P(1; 1; 1)) or
(a0; a1 ) = (1; 3) in which case a2 = 2 < a1, a contradiction.
12.6 The ring RC Consider an elliptic curve C and the divisor D = 2P ,
where P is any point on C . By Riemann{Roch,
h0 (nD) h1 (nD) = deg(nD) + (1 g):
As D > K  0, then h1 (nD) = 0 for all n  1. Also g = 1 and so
h0 (nD) = deg(nD) = 2n:
 A. R. Iano-Fletcher 135

Thus h0 (D) = 2 and h0 (2D) = 4. Let x0 ; x1 be a basis for H0 (D). Then x20 ,
x0 x1 and x21 are linearly independent elements of H0(2D). As h0 (2D) = 4
then there exists an extra element y of degree 4.
Consider the map:
'n : H0(D)
H0((n 1)D) ! H0 (nD):
Notice that x0 and x1 have no common base points. By the base-point-free
pencil trick (see [ACGH], p. 126),
Ker 'n ' H0 ((n 1)D D) = H0((n 2)D);
which has dimension 2(n 2). Also H0(D)
H0 ((n 1)D) has dimension
2  2(n 1). So dim Im 'n = 2n, and hence 'n is onto for all n  2. This
means that H0 (nD) is generated by H0 (D) and H0 ((n 1)D).
We thus have the following table of bases for the H0(nD).
n h0 (nD) monomials
1 2 x0 ; x1
2 4 x20 ; x0x1 ; x21 ; y
3 6 x30 ; x20x1 ; x0 x21 ; x31; x0 y; x1y
4 8 x40 ; x30x1 ; x20 x21 ; x0x31 ; x41 ; x20y; x0x1 y; x21y; y2
Notice that H0(4D) has dimension 8, but there are 9 monomials. Since '4
is onto then the rst eight in the list are linear independent. So there must
be a relation of the form:
f = y2 + yh2(x0 ; x1) g4(x0 ; x1 );
where h2 and g4 are homogeneous polynomials of degrees 4 and 2 respectively.
The number Nn of monomials in H0(nD) is given by:
jnk  
Nn = 1 + n 2 n + 21 :

Suppose that f was the only relation, then the dimension of the module
generated by the monomials of degree n is Nn 1
Nn 4 = 2n, which is the
same as h0 (nD).
Thus the ring R is K [x0 ; x1 ; y]=(f ), where xi has weight 1 and y has weight
2, that is, the curve is C4 in P(1; 1; 2). This technique should be compared to
that in [M], Lecture 1, p. 17{21] and to Weierstrass normal form.
136 Working with weighted complete intersections

12.7P Smooth weighted curve hypersurfaces with amplitude  =

d ai = 1 There are only two such curves which satisfy the conditions of
Theorem 12.3:
curve genus !C
C4  P(1; 1; 1) 3 OC (1)
C6  P(1; 1; 3) 2 OC (1)
These were calculated in a similar way to those of Section 12.5 and the genera
by the formula in Theorem 12.2.

13 Weighted surface complete intersections

In this section we give necessary and sucient conditions for surface weighted
complete intersections of codimension 1 and 2 to be quasismooth, well formed,
and have at worst canonical singularities. We also include lists of such inter-
sections.
13.1 Theorem Let Sd inPP = P(a0 ; a1; a2; a3 ) be a general hypersurface of
degree d and let  = d ai. Sd is quasismooth, well formed with at worst
canonical quotient singularities and is not a linear cone if and only if all the
following hold:
(1) For all i,
(i) d > ai .
(ii) there exists e such that ai j d ae (that is, there exists a monomial
xni xe of degree d).
(iii) there exists a monomial of degree d which does not involve xi .
(iv) if ai - d, then ai j .
(2) For all distinct i, j , with h = hcf(ai; aj ), then
(i) h j d.
(ii) h j .
(iii) one of the following holds:
either there exists a monomial xmi xnj of degree d,
or there exist monomials xni 1 xmj 1 xe1 and xni 2 xmj 2 xe2 of degree d
such that e1 and e2 are distinct.
for all distinct i, j , k, hcf(ai; aj ; ak ) = 1.
13.2 Note Since the hypersurface is well formed then !S = OS ().
 A. R. Iano-Fletcher 137

Proof Let f be a general homogeneous polynomial of degree d in variables

x0 ; : : : ; x3; dene Sd : (f = 0)  P.
Sd is quasismooth and not a linear cone if and only if conditions (1i),
(1ii), (1iii) and (2iii) hold (see Corollary 8.5).
Suppose furthermore that conditions (1iv), (2i), (2ii) and (3) hold. As Sd
is quasismooth the only singularities are due to the K  -action and hence are
cyclic quotient singularities on the fundamental simplex
 P. By condition
(3) only vertices and edges need be checked.
Consider Pi 2 Sd. By renumbering we can assume that i = 0. So a0 - d.
Condition (1ii) gives that there exists an e 6= 0 such that a0 j d ae . Without
loss of generality we can assume that e = 1. So f is of the form f = xn0 x1 +


.
Thus @f=@x1 is nonzero at P0. By the Inverse Function Theorem x2 and x3 are
local coordinates. So P0 2 Sd is of type a10 (a2 ; a3). However d = a0 +


+a3 +
and so a0 j a2 + a3 + . By condition (1iv), a0 j a2 + a3. Let h = hcf(a0 ; a2).
So h j a3 and hence, by condition (3), h = 1. Therefore P0 2 Sd is a canonical
singularity.
Consider the edge PiPj . Again by renumbering assume that i = 0 and
j = 1. f restricted to P0P1 is:
X
f= xn0 xm1 ;
where the sum is taken over the set f(n; m) : na0 + ma1 = dg. If a0 - d
then a0 j d ae for some e 6= 0. If e 6= 1 then h = hcf(a0; a1 ) j ae and by
condition (4) h = 1. Then P0P1 is nonsingular. So assume that either a0 j d or
a0 j d a1 . Hence f is not identically zero on P0P1, and so Sd \ P0 P1 is nite.
Each point in this intersection is of type h1 (a2 ; a3). Since d = a0 +


+ a3 + 
and h j  then h j a2 + a3 . Also hcf(h; a2) = 1. Thus each point is canonical.
Therefore Sd in P has at worst canonical singularities.
Conversely assume that Sd is quasismooth, well-formed, not a linear cone
and has at worst only canonical singularities. Suppose ai - d. By renumbering
we can assume that i = 0. So P0 2 Sd and a0 j d ae for some e. Without
loss of generality assume that e = 1. As above the singularity at P0 2 Sd is
of type a10 (a2; a3 ). Since this is canonical we have a0 j a2 + a3 and so a0 j .
This is condition (1iv).
Suppose h = hcf(ai; aj ). By renumbering assume that i = 0 and j = 1.
As Sd is well formed then h j d, which is condition (2i). So P0P1 \ Sd is a
nite intersection, where each point is of type h1 (a2 ; a3). This is canonical and
so h j . This is condition (2ii).
Suppose h = hcf(ai; aj ; ak ). Without loss of generality assume that i = 0,
j = 1 and k = 2. Let h0 = hcf(a0; a1). So h0 j d. Hence the line P0 P1 contains
singularities of type h10 (a2; a3 ). As these are canonical h = hcf(h0 ; a2) = 1.
This is condition (3). 
138 Working with weighted complete intersections

13.3 Reid's 95 codimension 1 K3 surfaces In 1979, Reid produced the

list of all families of codimension 1 weighted K3 surfaces; 95 in all (see [R1],
Section 4.5). The full list follows along with their respective singularities.
Weighted K3 Singularities
No. 1 X4  P(1; 1; 1; 1)
No. 2 X5  P(1; 1; 1; 2) A1
No. 3 X6  P(1; 1; 1; 3)
No. 4 X6  P(1; 1; 2; 2) 3  A1
No. 5 X7  P(1; 1; 2; 3) A1; A2
No. 6 X8  P(1; 1; 2; 4) 2  A1
No. 7 X8  P(1; 2; 2; 3) 4  A1 ; A2
No. 8 X9  P(1; 1; 3; 4) A3
No. 9 X9  P(1; 2; 3; 3) A1; 3  A2
No. 10 X10  P(1; 1; 3; 5) A2
No. 11 X10  P(1; 2; 2; 5) 5  A1
No. 12 X10  P(1; 2; 3; 4) 2  A1 ; A2; A3
No. 13 X11  P(1; 2; 3; 5) A1; A2; A4
No. 14 X12  P(1; 1; 4; 6) A1
No. 15 X12  P(1; 2; 3; 6) 2  A1 ; 2  A2
No. 16 X12  P(1; 2; 4; 5) 3  A1 ; A4
No. 17 X12  P(1; 3; 4; 4) 3  A3
No. 18 X12  P(2; 2; 3; 5) 6  A1 ; A4
No. 19 X12  P(2; 3; 3; 4) 3  A1 ; 4  A2
No. 20 X13  P(1; 3; 4; 5) A2; A3; A4
No. 21 X14  P(1; 2; 4; 7) 3  A1 ; A3
No. 22 X14  P(2; 2; 3; 7) 7  A1 ; A2
No. 23 X14  P(2; 3; 4; 5) 3  A1 ; A2; A3; A4
No. 24 X15  P(1; 2; 5; 7) A1; A6
No. 25 X15  P(1; 3; 4; 7) A3; A6
No. 26 X15  P(1; 3; 5; 6) 2  A2 ; A5
No. 27 X15  P(2; 3; 5; 5) A1; 3  A4
No. 28 X15  P(3; 3; 4; 5) 5  A2 ; A3
No. 29 X16  P(1; 2; 5; 8) 2  A1 ; A4
No. 30 X16  P(1; 3; 4; 8) A2; 2  A3
No. 31 X16  P(1; 4; 5; 6) A1; A4; A5
No. 32 X16  P(2; 3; 4; 7) 4  A1 ; A2; A6
No. 33 X17  P(2; 3; 5; 7) A1; A2; A4 ; A6
No. 34 X18  P(1; 2; 6; 9) 3  A1 ; A2
No. 35 X18  P(1; 3; 5; 9) 2  A2 ; A4
No. 36 X18  P(1; 4; 6; 7) A3; A1; A6
No. 37 X18  P(2; 3; 4; 9) 4  A1 ; 2  A2; A3
 A. R. Iano-Fletcher 139

No. 38 X18  P(2; 3; 5; 8) 2  A1 ; A4; A7

No. 39 X18  P(3; 4; 5; 6) 3  A2 ; A3; A1 ; A4
No. 40 X19  P(3; 4; 5; 7) A2 ; A3; A4; A6
No. 41 X20  P(1; 4; 5; 10) A1 ; 2  A4
No. 42 X20  P(2; 3; 5; 10) 2  A1 ; A2; 2  A4
No. 43 X20  P(2; 4; 5; 9) 5  A1 ; A8
No. 44 X20  P(2; 5; 6; 7) 3  A1 ; A5; A6
No. 45 X20  P(3; 4; 5; 8) A2 ; 2  A3; A7
No. 46 X21  P(1; 3; 7; 10) A9
No. 47 X21  P(1; 5; 7; 8) A4 ; A7
No. 48 X21  P(2; 3; 7; 9) A1 ; 2  A2; A8
No. 49 X21  P(3; 5; 6; 7) 3  A2 ; A4; A5
No. 50 X22  P(1; 3; 7; 11) A2 ; A6
No. 51 X22  P(1; 4; 6; 11) A3 ; A1; A5
No. 52 X22  P(2; 4; 5; 11) 5  A1 ; A3; A4
No. 53 X24  P(1; 3; 8; 12) 2  A2 ; A3
No. 54 X24  P(1; 6; 8; 9) A1 ; A2; A8
No. 55 X24  P(2; 3; 7; 12) 2  A1 ; 2  A2 ; A6
No. 56 X24  P(2; 3; 8; 11) 3  A1 ; A10
No. 57 X24  P(3; 4; 5; 12) 2  A2 ; 2  A3 ; A4
No. 58 X24  P(3; 4; 7; 10) A1 ; A6; A9
No. 59 X24  P(3; 6; 7; 8) 4  A2 ; A1; A6
No. 60 X24  P(4; 5; 6; 9) 2  A1 ; A4; A2 ; A8
No. 61 X25  P(4; 5; 7; 9) A3 ; A6; A8
No. 62 X26  P(1; 5; 7; 13) A4 ; A6
No. 63 X26  P(2; 3; 8; 13) 3  A1 ; A2; A7
No. 64 X26  P(2; 5; 6; 13) 4  A1 ; A4; A5
No. 65 X27  P(2; 5; 9; 11) A1 ; A4; A10
No. 66 X27  P(5; 6; 7; 9) A4 ; A5; A2; A6
No. 67 X28  P(1; 4; 9; 14) A1 ; A8
No. 68 X28  P(3; 4; 7; 14) A2 ; A1; 2  A6
No. 69 X28  P(4; 6; 7; 11) 2  A1 ; A5; A10
No. 70 X30  P(1; 4; 10; 15) A3 ; A1; A4
No. 71 X30  P(1; 6; 8; 15) A1 ; A2; A7
No. 72 X30  P(2; 3; 10; 15) 3  A1 ; 2  A2 ; A4
No. 73 X30  P(2; 6; 7; 15) 5  A1 ; A2; A6
No. 74 X30  P(3; 4; 10; 13) A3 ; A1; A12
No. 75 X30  P(4; 5; 6; 15) A3 ; 2  A1; 2  A4 ; A2
No. 76 X30  P(5; 6; 8; 11) A1 ; A7; A10
No. 77 X32  P(2; 5; 9; 16) 2  A1 ; A4; A8
No. 78 X32  P(4; 5; 7; 16) 2  A3 ; A4; A6
No. 79 X33  P(3; 5; 11; 14) A4 ; A13
140 Working with weighted complete intersections

No. 80 X34  P(3; 4; 10; 17) A2; A3; A1 ; A9

No. 81 X34  P(4; 6; 7; 17) A3; 2  A1 ; A5; A6
No. 82 X36  P(1; 5; 12; 18) A4; A5
No. 83 X36  P(3; 4; 11; 18) 2  A2 ; A1; A10
No. 84 X36  P(7; 8; 9; 12) A6; A7; A3 ; A2
No. 85 X38  P(3; 5; 11; 19) A2; A4; A10
No. 86 X38  P(5; 6; 8; 19) A4; A5; A1 ; A7
No. 87 X40  P(5; 7; 8; 20) 2  A4 ; A6; A3
No. 88 X42  P(1; 6; 14; 21) A1; A2; A6
No. 89 X42  P(2; 5; 14; 21) 3  A1 ; A4; A6
No. 90 X42  P(3; 4; 14; 21) 2  A2 ; A3; A1; A6
No. 91 X44  P(4; 5; 13; 22) A1; A4; A12
No. 92 X48  P(3; 5; 16; 24) 2  A2 ; A4; A7
No. 93 X50  P(7; 8; 10; 25) A6; A7; A1 ; A4
No. 94 X54  P(4; 5; 18; 27) A3; A1; A4 ; A8
No. 95 X66  P(5; 6; 22; 33) A4; A1; A2 ; A10
Table 1: Reid's 95 codimension 1 K3 surfaces
However there are not so many dimension 2 weighted hypersurfaces with
!S ' OS (1):
13.4 Theorem There are exactly three families of dimension 2 weighted
hypersurfaces with at worst canonical singularities and !S ' OS (1), and
exactly three families with !S ' OS ( 1),
=1 = 1
S5  P(1; 1; 1; 1) S3  P(1; 1; 1; 1)
S6  P(1; 1; 1; 2) S4  P(1; 1; 1; 2)
S8  P(1; 1; 1; 4) S6  P(1; 1; 2; 3)
13.5 Note These families are all nonsingular.
Proof Condition (2ii) of Theorem 13.1 is very strong when  = 1 and
forces the ai to be pairwise coprime. Similarly condition (1iv) forces ai j d for
each i. So a0 a1 a2a3 j d and d = a0 +


+ a3 + . Order a3  a2  a1  a0  1
and let d = a3 . Thus a0 a1 a2 j  and ( 1)a3 = a0 +


+ a2 + .
Suppose  = 1. Then 2a3  a3 = a0 +


+ a3 + 1  5a3. So 2    5.
Running through the possible values of :
(i)  = 5. So a0a1 a2 j 5. If a2 = 1 then a4 = 1 (that is, S5 in P(1; 1; 1; 1)).
If a2 = 5 then a3 = 2, a contradiction.
 A. R. Iano-Fletcher 141

(ii)  = 4. So a0 a1a2 j 4. If a2 = 1 then a4 = 34 , a contradiction. If a2 = 2

then a4 = 53 , a contradiction. If a2 = 4 then a4 = 73 , a contradiction.
(iii)  = 3. So a0 a1a2 j 3. If a2 = 1 then a4 = 2 (that is, S6 in P(1; 1; 1; 2)).
If a2 = 3 then a4 = 3, a contradiction.
(iv)  = 2. So a0 a1a2 j 2. If a2 = 1 then a4 = 4 (that is, S8 in P(1; 1; 1; 4)).
If a2 = 2 then a4 = 25 , a contradiction.
So there are exactly three families.
Suppose that  = 1. Then 2a3  a3 = a0 +


+ a3 1  6a3 .
Thus 2    6. As above this gives rise to the following families: S3 in
P(1; 1; 1; 1) in the case  = 3, S4 in P(1; 1; 1; 2) and S6 in P(1; 1; 2; 3) in the
case  = 2. 
Consider the case of codimension 2 complete intersections.
13.6 Theorem Suppose S = Sd ;d in P = P(a0 ; : : : ; a4 ) is quasismooth and
1 2
isPnot theP aintersection of a linear cone with another hypersurface. Let  =
d .i S is well formed and has at worst canonical singularities if and
only if the following hold:
(1) for all i, if ai - d1 and ai - d2 then ai j .
(2) for all distinct i and j , with h = hcf(ai; aj ), one of the following occurs:
(i) h j d1 and h j d2,
(ii) h j d1, h - d2 and h j , or
(iii) h - d1, h j d2 and h j .
(3) for all distinct i, j and k, with h = hcf(ai; aj ; ak ), h j d1 , h j d2 and
h j .
(4) for all distinct i, j , k and l, h = hcf(ai; aj ; ak ; al ) = 1
13.7 Note Since the hypersurface is well formed we have that !S = OS ().
Proof Let f1 and f2 be suciently general homogeneous polynomials of
degrees d1 and d2 respectively, in the variables x0 ; : : : ; x4 with respect to the
weights a0; : : : ; a4. Dene S : (f1 = 0; f2 = 0)  P.
Since S is quasismooth the only singularities are due to the K  -action
and hence are all cyclic quotient singularities occurring on the fundamental
simplex
.
Assume conditions (1); : : : ; (4) hold. By conditions (2), (3) and (4) S is
well formed. By condition (4) only the vertices, edges and faces of
need be
considered.
142 Working with weighted complete intersections

Suppose Pi 2 S . By renumbering we can assume that i = 0. So a0 - d1

and a0 - d2. As S is quasismooth (and using I = f0g in Theorem 8.7) there
exist monomials xn0 xe1 and xm0 xe2 of degrees d1 and d2, where e1 6= e2. By
renumbering we can write e1 = 1 and e2 = 2. So f1 and f2 are of the form:
f1 = xn0 x1 +



f2 = xm0 x2 +



Thus @f1 =@x1 and @f2 =@x2 are nonzero at P0 . By the Inverse Function The-
orem, x3 and x4 are local coordinates around P0. Hence P0 2 S is of type
a0 (a3 ; a4 ). As d1 + d2 = a0 +


+ a4 +  and a0 j  then a0 j a3 n+ a4 . Let
1
h = hcf(a0 ; a3). So h j a4 and, by condition (3), h j d1. Since deg(x0 x1 ) = d1,
h j a1 and so, by condition (4), h = 1. Thus P0 2 S is canonical.
Consider the edge PiPj . By renumbering we can assume that i = 0 and
j = 1. Let h = hcf(a0; a1 ). Notice that P0P1  Xd if and only if h - d
for  = 0; 1. By condition (2), h j d for some . Without loss of generality
assume that h j d1. There are 2 cases:
(a) h j d2. P0P1 \ (f = 0) is a nite set of points for  = 0; 1. Thus
P0P1 \ S = ;.
(b) h - d2. In this case no monomial of the form xn0 xm1 of degree d2 exists
(or else h j d2 ). From Theorem 8.7 (with I = f0; 1g) there exists a
monomial xn0 xm1 xe of degree d2, where e 6= 0; 1. By renumbering we can
assume that e = 2. Thus f2 is of the form:
f2 = xn0 xm1 x2 +



and @f2 =@x2 is nonzero on P0P1 \ S . By the Inverse Function Theorem,
x3 and x4 are local coordinates around each point of P0 P1 \ S and so
each is of type h1 (a3 ; a4). Condition (2b) gives h j  and so h j a3 + a4.
Let h0 = hcf(h; a3). So h0 j a4 and thus by condition (4) h0 = 1. Thus
these points are canonical.
Therefore S has at worst canonical points along P0P1.
Consider the face PiPj Pk . As before assume i = 0, j = 1 and k = 2.
By condition (3) h = hcf(a0 ; a1; a2) j d1 and h j d2. So P0 P1P2 intersects S
transversally. Each point in the intersection is of type h1 (a3; a4). As h j ,
h j a3 + a4 . By condition (4) hcf(h; a3 ) = 1. Thus these points are canonical.
Therefore conditions (1); : : : ; (4) are sucient.
Conversely assume that S is well formed and has at worst canonical sin-
gularities. Suppose ai - d1 and ai - d2. By renumbering assume i = 0. Thus
P0 2 S . Since S is quasismooth there exist 2 monomials xn0 xe1 and xm0 xe2 of
degrees d1 and d2, where e1 6= e2. Without loss of generality we can assume
 A. R. Iano-Fletcher 143

that e1 = 1 and e2 = 2. As before we nd that P0 2 S is of type a10 (a3 ; a4).

As this is canonical a0 j a3 + a4 and so a0 j . This is condition (1).
Suppose h = hcf(ai; aj ) for distinct i and j . As usual we can renumber
such that i = 0 and j = 1. As S is well formed then h j d for some . Suppose
h j d1 . If h j d2 then this is condition (2a). So assume that h - d2. As above
each point of P0P1 \ S is isolated and of type h1 (a3 ; a4). Thus h j a3 + a4 and
so h j . This is condition (2b). Likewise for the case when h j d2 but h - d1.
This gives condition (2c).
Suppose h = hcf(ai; aj ; ak ) for distinct i, j and k. Renumber such that
i = 0, j = 1 and k = 2. As S is well formed then h j d1 and h j d2. Thus
P0P1 P2 \ S is a nite number of points, all of type h1 (a3 ; a4). As these are
canonical h j a3 + a4 and so h j . This is condition (3). Also hcf(h; a3 ) =
hcf(h; a4 ) = 1, which is condition (4).
Therefore these conditions are necessary. 

13.8 Codimension 2 Weighted K3 Surfaces There are 84 families of

codimension 2 P
quasismooth, well-formed K3 surfaces with only canonical sin-
gularities and ai  100. These were found by means of a computer search
program.
Weighted K3 Singularities
No. 1 X2;3  P(1; 1; 1; 1; 1)
No. 2 X3;3  P(1; 1; 1; 1; 2) A1
No. 3 X3;4  P(1; 1; 1; 2; 2) 2  A1
No. 4 X4;4  P(1; 1; 1; 2; 3) A2
No. 5 X4;4  P(1; 1; 2; 2; 2) 4  A1
No. 6 X4;5  P(1; 1; 2; 2; 3) 2  A1 ; A2
No. 7 X4;6  P(1; 1; 2; 3; 3) 2  A2
No. 8 X4;6  P(1; 2; 2; 2; 3) 6  A1
No. 9 X5;6  P(1; 1; 2; 3; 4) A1; A3
No. 10 X5;6  P(1; 2; 2; 3; 3) 3  A1 ; 2  A2
No. 11 X6;6  P(1; 1; 2; 3; 5) A4
No. 12 X6;6  P(1; 2; 2; 3; 4) 4  A1 ; A3
No. 13 X6;6  P(1; 2; 3; 3; 3) 4  A2
No. 14 X6;6  P(2; 2; 2; 3; 3) 9  A1
No. 15 X6;7  P(1; 2; 2; 3; 5) 3  A1 ; A4
No. 16 X6;7  P(1; 2; 3; 3; 4) A1; 2  A2 ; A3
No. 17 X6;8  P(1; 1; 3; 4; 5) A4
No. 18 X6;8  P(1; 2; 3; 3; 5) 2  A2 ; A4
No. 19 X6;8  P(1; 2; 3; 4; 4) 2  A1 ; 2  A3
No. 20 X6;8  P(2; 2; 3; 3; 4) 6  A1 ; 2  A2
No. 21 X6;9  P(1; 2; 3; 4; 5) A1; A3 ; A4
144 Working with weighted complete intersections

No. 22 X7;8  P(1; 2; 3; 4; 5) 2  A1; A2 ; A4

No. 23 X6;10  P(1; 2; 3; 5; 5) 2  A4
No. 24 X6;10  P(2; 2; 3; 4; 5) 7  A1; A3
No. 25 X8;9  P(1; 2; 3; 4; 7) 2  A1; A6
No. 26 X8;9  P(1; 3; 4; 4; 5) 2  A3; A4
No. 27 X8;9  P(2; 3; 3; 4; 5) 2  A1; 3  A2; A4
No. 28 X8;10  P(1; 2; 3; 5; 7) A2 ; A6
No. 29 X8;10  P(1; 2; 4; 5; 6) 3  A1; A5
No. 30 X8;10  P(1; 3; 4; 5; 5) A2 ; 2  A4
No. 31 X8;10  P(2; 3; 4; 4; 5) 4  A1; A2 ; 2  A3
No. 32 X9;10  P(1; 2; 3; 5; 8) A1 ; A7
No. 33 X9;10  P(1; 3; 4; 5; 6) A2 ; A3; A5
No. 34 X9;10  P(2; 2; 3; 5; 7) 5  A1; A6
No. 35 X9;10  P(2; 3; 4; 5; 5) 2  A1; A3 ; 2  A4
No. 36 X8;12  P(1; 3; 4; 5; 7) A4 ; A6
No. 37 X8;12  P(2; 3; 4; 5; 6) 4  A1; 2  A2; A4
No. 38 X9;12  P(2; 3; 4; 5; 7) 3  A1; A4 ; A6
No. 39 X10;11  P(2; 3; 4; 5; 7) 2  A1; A2 ; A3; A6
No. 40 X10;12  P(1; 3; 4; 5; 9) A2 ; A8
No. 41 X10;12  P(1; 3; 5; 6; 7) 2  A2; A6
No. 42 X10;12  P(1; 4; 5; 6; 6) A1 ; 2  A5
No. 43 X10;12  P(2; 3; 4; 5; 8) 3  A1; A3 ; A7
No. 44 X10;12  P(2; 3; 5; 5; 7) 2  A4; A6
No. 45 X10;12  P(2; 4; 5; 5; 6) 5  A1; 2  A4
No. 46 X10;12  P(3; 3; 4; 5; 7) 4  A2; A6
No. 47 X10;12  P(3; 4; 4; 5; 6) 2  A2; 3  A3; A1
No. 48 X11;12  P(1; 4; 5; 6; 7) A1 ; A4; A6
No. 49 X10;14  P(1; 2; 5; 7; 9) A8
No. 50 X10;14  P(2; 3; 5; 7; 7) A2 ; 2  A6
No. 51 X10;14  P(2; 4; 5; 6; 7) 5  A1; A3 ; A5
No. 52 X10;15  P(2; 3; 5; 7; 8) A1 ; A6; A7
No. 53 X12;13  P(3; 4; 5; 6; 7) 2  A2; A1 ; A4; A6
No. 54 X12;14  P(1; 3; 4; 7; 11) A10
No. 55 X12;14  P(1; 4; 6; 7; 8) A1 ; A3; A7
No. 56 X12;14  P(2; 3; 4; 7; 10) 4  A1; A9
No. 57 X12;14  P(2; 3; 5; 7; 9) A2 ; A4; A8
No. 58 X12;14  P(3; 4; 5; 7; 7) A4 ; 2  A6
No. 59 X12;14  P(4; 4; 5; 6; 7) 3  A3; 2  A1; A4
No. 60 X12;15  P(1; 4; 5; 6; 11) A1 ; A10
No. 61 X12;15  P(3; 4; 5; 6; 9) 3  A2; A1 ; A8
No. 62 X12;15  P(3; 4; 5; 7; 8) A3 ; A6; A7
No. 63 X12;16  P(2; 5; 6; 7; 8) 4  A1; A4 ; A6
 A. R. Iano-Fletcher 145

No. 64 X14;15  P(2; 3; 5; 7; 12) A1; A2 ; A11

No. 65 X14;15  P(2; 5; 6; 7; 9) 2  A1 ; A5; A8
No. 66 X14;15  P(3; 4; 5; 7; 10) A3; A4 ; A9
No. 67 X14;15  P(3; 5; 6; 7; 8) 2  A2 ; A5; A7
No. 68 X14;16  P(1; 5; 7; 8; 9) A4; A8
No. 69 X14;16  P(3; 4; 5; 7; 11) A2; A4 ; A10
No. 70 X14;16  P(4; 5; 6; 7; 8) A1; 2  A3 ; A4; A5
No. 71 X15;16  P(2; 3; 5; 8; 13) 2  A1 ; A12
No. 72 X15;16  P(3; 4; 5; 8; 11) 2  A3 ; A10
No. 73 X14;18  P(2; 3; 7; 9; 11) 2  A2 ; A10
No. 74 X14;18  P(2; 6; 7; 8; 9) 5  A1 ; A2; A7
No. 75 X12;20  P(4; 5; 6; 7; 10) 2  A1 ; 2  A4; A6
No. 76 X16;18  P(1; 6; 8; 9; 10) A1; A2 ; A9
No. 77 X16;18  P(4; 6; 7; 8; 9) 2  A1 ; 2  A3; A2 ; A6
No. 78 X18;20  P(4; 5; 6; 9; 14) 2  A1 ; A2; A13
No. 79 X18;20  P(4; 5; 7; 9; 13) A6; A12
No. 80 X18;20  P(5; 6; 7; 9; 11) A2; A6 ; A10
No. 81 X18;22  P(2; 5; 9; 11; 13) A4; A12
No. 82 X20;21  P(3; 4; 7; 10; 17) A1; A16
No. 83 X18;30  P(6; 8; 9; 10; 15) 2  A1 ; 2  A2; A7 ; A4
No. 84 X24;30  P(8; 9; 10; 12; 15) A1; A3 ; A8; A2; A4
Table 2: The codimension 2 weighted K3 surfaces

14 Weighted 3-fold complete intersections

This section gives the corresponding conditions and lists for 3-folds.
14.1 Theorem
P Let Xd be a general hypersurface in P = P(a0 ; : : : ; a4) and let
=d ai . Then Xd is quasismooth with only isolated terminal quotient
singularities and is not a linear cone if and only if all the following hold:
(1) For all i,
(i) d > ai.
(ii) there exists a monomial xmi xe of degree d (that is, there exists e
such that ai j d ae).
(iii) if ai - d, there exists an m 6= i; e such that ai j am + .
(2) For all distinct i, j , with h = hcf(ai ; aj ), then
(i) h j d.
146 Working with weighted complete intersections

(ii) there exists an m 6= i, j such that h j am + .

(iii) one of the following holds:
either there exists a monomial xmi xnj of degree d,
or there exist monomials xni 1 xmj 1 xe1 and xni 2 xmj 2 xe2 of degree d
such that e1 and e2 are distinct.
(iv) there exists a monomial of degree d which does not involve xi or
xj .
(3) For all distinct i, j , k, hcf(ai; aj ; ak ) = 1.
14.2 Note Since the hypersurface is quasismooth and of dimension 3 then
it is well formed, and so !X = OX ().
Proof Let f be a general homogeneous polynomial of degree d in variables
x0 ; : : : ; x3 ; dene Xd : (f = 0)  P.
Xd is quasismooth and not a linear cone (and therefore well-formed) if
and only if conditions (1i), (1ii), (2i), (2iii), (2iv) and (3) hold (see Corollary
8.6). By calculating the types of the singularities on Xd we can show that
conditions (1iii), (2i), (2ii) and (3) are equivalent to these singularities being
terminal; the combinatorial conditions for which are found in Lemma 9.3.
Suppose furthermore that conditions (1iii), (2i), (2ii) and (3) hold. As Xd
is quasismooth the only singularities are due to the K  -action and hence are
cyclic quotient singularities on the fundamental simplex
 P. By condition
(3) only vertices and edges need be checked.
Consider Pi 2 Xd . By renumbering we can assume that i = 0. So a0 - d.
By condition (1ii) there exists an e 6= 0 such that a0 j d ae. Without loss of
generality we can assume that e = 1. So f is of the form f = xn0 x1 +


. Thus
@f=@x1 is nonzero at P0. By the Inverse Function Theorem x2 , x3 and x4 are
local coordinates around P0 . So P0 2 Xd is of type a10 (a2 ; a3; a4). However
d = a0 +


+ a4 +  and so a0 j a2 + a4 + . By condition (1iv), a0 j  + am
for some m = 2; 3; 4. Without loss of generality assume m = 2. By condition
(1iv), a0 j a3 + a4. Let h = hcf(a0; a3 ). So h j a3 and hence, by condition (3),
h = 1. Therefore P0 2 Xd is a terminal singularity.
Consider the edge PiPj . Again by renumbering assume that i = 0 and
j = 1. f restricted to P0P1 is:
X
f = xn0 xm1 ;
where the sum is taken over the set f(n; m) : na0 + ma1 = dg. If a0 - d
then a0 j d ae for some e 6= 0. If e 6= 1 then h = hcf(a0 ; a1) j ae and by
condition (4) h = 1. Then P0P1 is nonsingular. So assume that either a0 j d
or a0 j d a1 . Hence f is not identically zero on P0 P1, and so Xd \ P0P1 is
 A. R. Iano-Fletcher 147

nite. Each point in this intersection is of type h1 (a2; a3 ; a4). By condition

(2ii) h j  + am for some m = 2; 3; 4. By renumbering assume m = 2. Since
d = a0 +


+ a4 + , then h j a3 + a4 . Also hcf(h; a3) = 1. Thus each point
is terminal.
Therefore Xd in P has at worst terminal singularities.
Conversely assume that Xd is quasismooth, not a linear cone and has at
worst only terminal singularities. Suppose ai - d. By renumbering we can
assume that i = 0. So P0 2 Xd and a0 j d ae for some e. Without loss of
generality assume that e = 1. As above the singularity at P0 2 Xd is of type
a0 (a2 ; a3 ; a4 ). Since this is terminal we have, after renumbering, a0 j a2 + a3
1
and so a0 j  + am for some m. This is condition (1iv).
Suppose h = hcf(ai; aj ). By renumbering assume that i = 0 and j = 1.
As Xd is well formed then h j d, which is condition (2i). So P0P1 \ Xd is a
nite intersection, where each point is of type h1 (a2 ; a3; a4 ). This is terminal
and so h j  + am for m = 2; 3; 4. This is condition (2ii).
Suppose h = hcf(ai; aj ; ak ). Without loss of generality assume that i = 0,
j = 1 and k = 2. Let h0 = hcf(a0; a1). So h0 j d. Hence the line P0 P1 contains
singularities of type h10 (a2; a3; a4 ). As these are terminal h = hcf(h0; a2) = 1.
This is condition (3). 
14.3 Theorem There are exactly 4 families of quasismooth 3-fold weighted
hypersurfaces with only terminal isolated quotient singularities and !X ' OX :
X5  P(1; 1; 1; 1; 1)
X6  P(1; 1; 1; 1; 2)
X8  P(1; 1; 1; 1; 4)
X10  P(1; 1; 1; 2; 5)
Notice that the above are all nonsingular.
Proof As KX ' OX then  = 0. Suppose h = hcf(ai ; aj ) 6= 1 for distinct
6 i; j such that h j am + .
i, j . By Theorem 14.1 (2ii) there exists an m =
However  = 0 and so h j am . By (3) h = 1, a contradiction. Hence ai and
aj are coprime for distinct i, j .
Suppose that ai - d. Then there exists an m 6= i; ei such that ai j am + .
Thus ai = hcf(ai; am ) = 1, contradicting ai - d. Thus ai j d for all i.
Order the faig such that a4 


 a0 . So 5a4  d  2a4 . Let d = a4 .
Thus  = 2; 3; 4 or 5. As the faig are pairwise coprime then a0 a1a2 a3 a4 j d
and so a0a1 a2 a3 j . Also a0 +


+ a3 = ( 1)a4 . There are four cases:
(i)  = 5. Either (a0; a1 ; a2; a3) = (1; 1; 1; 1) giving a4 = 1 (that is, X5 in
P(1; 1; 1; 1; 1)) or (a0 ; a1 ; a2 ; a3 ) = (1; 1; 1; 5) giving a4 = 2 < a3 .
148 Working with weighted complete intersections

(ii)  = 4. So  1 = 3 and divides a0 +


+ a3. There are three

possibilities:
(a) (a0 ; a1; a2; a3 ) = (1; 1; 1; 1), giving 3 j 4.
(b) (a0 ; a1; a2; a3 ) = (1; 1; 1; 2), giving 3 j 5,
(c) (a0 ; a1; a2; a3 ) = (1; 1; 1; 4), giving 3 j 7.
All of these possibilities give contradictions.
(iii)  = 3. Either (a0 ; a1; a2 ; a3) = (1; 1; 1; 1) giving a4 = 2 (that is, X6 in
P(1; 1; 1; 1; 2)), or (a0 ; a1 ; a2 ; a3 ) = (1; 1; 1; 3) giving a4 = 3, contradict-
ing the coprime condition.
(iv)  = 2. Either (a0 ; a1; a2 ; a3) = (1; 1; 1; 1) giving a4 = 4 (that is, X8 in
P(1; 1; 1; 1; 4)), or (a0 ; a1 ; a2 ; a3 ) = (1; 1; 1; 2) giving a4 = 5 (that is, X10
in P(1; 1; 1; 2; 5)). 

Consider the case of codimension 2 complete intersections.

14.4 Theorem Suppose X = Xd ;d in P = P(a0 ; : : : ; a5) is quasismooth
1 2
and not
P the P intersection of a linear cone with another hypersurface. Let
 = d ai. X has at worst terminal singularities if and only if the
following hold:
(1) for all i, if ai - d1 and ai - d2 then there exists e1, e2 and m such that
ai j d1 ae1 , ai j d2 ae2 and ai j  + am, where fi; e1; e2 ; mg are distinct.
(2) for all distinct i and j , with h = hcf(ai; aj ), at least one of the following
occurs:
(a) h j d1 and h j d2,
(b) h j d1 , h - d2 and h j  + am for some m 6= i; j , or
(c) h - d1 , h j d2 and h j  + am for some m 6= i; j .
(3) for all distinct i, j and k, with h = hcf(ai; aj ; ak ), h j d1, h j d2 and
h j  + am for some m 6= i; j; k.
(4) for all distinct i, j , k and l, h = hcf(ai ; aj ; ak ; al) = 1.
14.5 Note Since X is quasismooth, of dimension 3 and not the intersect of
a linear cone with other hypersurfaces then X is well formed. Thus !X =
OX ().
 A. R. Iano-Fletcher 149

Proof Let f1 and f2 be suciently general homogeneous polynomials of

degrees d1 and d2 respectively, in the variables x0 ; : : : ; x4 with respect to the
weights a0; : : : ; a4. Dene X : (f1 = 0; f2 = 0)  P.
Since X is quasismooth the only singularities are due to the K  -action
and hence are all cyclic quotient singularities occurring on the fundamental
simplex
.
Assume conditions (1); : : : ; (4) hold. By condition (4) only the vertices,
edges and faces of
need be considered.
Suppose Pi 2 X . By renumbering we can assume that i = 0. So a0 - d1
and a0 - d2. By condition (1), there exist monomials xn0 1 xe1 and xn0 2 xe2 of
degrees d1 and d2, where e1 6= e2 . Note that this is really quasismoothness.
By renumbering we can write e1 = 1 and e2 = 2. So f1 and f2 are of the
form:
f1 = xn0 1 x1 +


;
f2 = xn0 2 x2 +


:
Thus @f1 =@x1 and @f2 =@x2 are nonzero at P0 . By the Inverse Function Theo-
rem, x3 , x4 and x5 are local coordinates. Hence P0 2 X is of type a10 (a3; a4 ; a5).
By condition (1) a0 j  + am for some m 6= 0; 1; 2. Without loss of general-
ity assume m = 3. As d1 + d2 = a0 +


+ a5 +  then a0 j a4 + a5 . Let
h = hcf(a0; a4 ). So h j a5 and, by condition (3), h j d1. Since deg xn0 x1 = d1,
h j a1 and so, by condition (4), h = 1. Thus P0 2 X is terminal.
Consider the edge PiPj . By renumbering we can assume that i = 0 and
j = 1. Let h = hcf(a0 ; a1). Notice that P0 P1  Xd if and only if h - d
for  = 0; 1. By condition (2), h j d for some . Without loss of generality
assume that h j d1. There are two cases:
(a) h j d2. P0 P1 \ (f = 0) is a nite set of points for  = 0; 1. Thus
P0P1 \ X = ;.
(b) h - d2. In this case no monomial of the form xn0 0 xn1 1 of degree d2 exists
(or else h j d2). From Theorem 8.7 (with I = f0; 1g) there exists a
monomial xn0 0 xn1 1 xe of degree d2, where e 6= 0; 1. By renumbering we
can assume that e = 2. Thus f2 is of the form:
f2 = xn0 0 xn1 1 x2 +



and @f2 =@x2 is nonzero on P0P1 \ X . By the Inverse Function Theorem,
x3 , x4 and x5 are local coordinates and so each point of P0 P1 \ X is of
type h1 (a3; a4 ; a5). Condition (2b) gives h j  + am for some m 6= 0; 1; 2.
Assume that m = 3, and hence h j a4 + a5. Let h0 = hcf(h; a4). So
h j a4 and thus by condition (4) h = 1. Thus these points are terminal.
150 Working with weighted complete intersections

Therefore X has at worst terminal points along P0P1.

Consider the face PiPj Pk . As before assume i = 0, j = 1 and k = 2.
By condition (3) h = hcf(a0; a1 ; a2) j d1 and h j d2 . So P0P1 P2 intersects
X transversally. Each point in the intersection is of type h1 (a3 ; a4; a5). As
h j  + am for some m 6= 0; 1; 2, after renumbering, h j a3 + a4. By condition
(4) hcf(h; a3 ) = 1. Thus these points are terminal.
Therefore condition (1); : : : ; (4) are sucient.
Conversely assume that X has at worst terminal singularities. Suppose
ai - d1 and ai - d2. By renumbering assume i = 0. Thus P0 2 X . Since X
is quasismooth there exist 2 monomials xn0 xe1 and xm0 xe2 of degrees d1 and
d2, where e1 6= e2 . This gives the rst part of condition (1). Without loss
of generality we can assume that e1 = 1 and e2 = 2. As before we nd
that P0 2 X is of type a10 (a3 ; a4; a5). As this is terminal, after renumbering,
a0 j a3 + a4 and so a0 j  + a5 . This is condition (1).
Suppose h = hcf(ai; aj ) for distinct i and j . As usual we can renumber
such that i = 0 and j = 1. As X is well formed then h j d for some .
Suppose h j d1. If h j d2 then this is condition (2a). So assume that h - d2.
As above each point of P0P1 \ X is isolated and of type h1 (a3; a4; a5 ). After
renumbering, h j a3 + a4 and so h j  + a5 . This is condition (2b). Likewise
for the case when h j d2 but h - d1. This gives condition (2c).
Suppose h = hcf(ai ; aj ; ak ) for distinct i, j and k. Renumber such that
i = 0, j = 1 and k = 2. Since X is well formed h j d1 and h j d2. P0P1P2 \ X is
a nite number of points, all of type h1 (a3 ; a4; a5 ). As these are terminal, after
renumbering, h j a3 + a4 and so h j  + a5 . This is condition (3). Condition
(4) follows from the fact that hcf(h; a3) = hcf(h; a4 ) = 1.
Therefore these conditions are necessary. 
14.6 Codimension 2 weighted 3-fold complete intersection with
trivial canonical bundle The four families of 3-fold codimension 2 quasi-
and
P complete intersections with at worst terminal singularities, !X ' OX
smooth
ai < 100 are:
X2;4  P(1; 1; 1; 1; 1; 1) X3;3  P(1; 1; 1; 1; 1; 1)
X3;4  P(1; 1; 1; 1; 1; 2) X4;4  P(1; 1; 1; 1; 2; 2)
Again the above are all nonsingular and were found using a computer search
based on the conditions of Theorem 14.4.

15 Canonically embedded weighted 3-folds

15.1 Canonically embedded 3-fold weighted hypersurfaces There
are 23 families of 3-fold quasismooth weighted hypersurfaces with only ter-
 A. R. Iano-Fletcher 151
P
minal isolated quotient singularities with !X ' OX (1) and ai  100.
Hypersurface KX3 pg Singularities
No. 1 X6  P(1; 1; 1; 1; 1) 6 5
No. 2 X7  P(1; 1; 1; 1; 2) 7=2 4 21 (1; 1; 1)
No. 3 X8  P(1; 1; 1; 2; 2) 2 3 4  12 (1; 1; 1)
No. 4 X9  P(1; 1; 1; 2; 3) 3=2 3 12 (1; 1; 1)
No. 5 X10  P(1; 1; 1; 1; 5) 2 4
No. 6 X10  P(1; 1; 2; 2; 3) 5=6 2 5  12 (1; 1; 1); 13 (1; 1; 1)
No. 7 X12  P(1; 1; 1; 2; 6) 1 3 2  12 (1; 1; 1)
No. 8 X12  P(1; 1; 2; 3; 4) 1=2 2 3  12 (1; 1; 1)
No. 9 X12  P(1; 2; 2; 3; 3) 1=3 1 6  12 (1; 1; 1); 4  13 (1; 1; 1)
No. 10 X14  P(1; 1; 2; 2; 7) 1=2 2 7  12 (1; 1; 1)
No. 11 X15  P(1; 2; 3; 3; 5) 1=6 1 12 (1; 1; 1); 5  13 (1; 1; 1)
No. 12 X16  P(1; 1; 2; 3; 8) 1=3 2 2  12 (1; 1; 1); 13 (1; 1; 1)
No. 13 X16  P(1; 2; 3; 4; 5) 2=15 1 4  12 (1; 1; 1); 13 (1; 1; 1);
1 (1; 1; 2)
5
No. 14 X18  P(1; 2; 2; 3; 9) 1=6 1 9  12 (1; 1; 1); 2  13 (1; 1; 1)
No. 15 X18  P(2; 3; 3; 4; 5) 1=20 0 4  12 (1; 1; 1); 6  13 (1; 1; 1);
1 (1; 1; 1); 1 (1; 1; 2)
4 5
No. 16 X20  P(2; 3; 4; 5; 5) 1=30 0 5  12 (1; 1; 1); 13 (1; 1; 1);
4  15 (1; 1; 2)
No. 17 X21  P(1; 3; 4; 5; 7) 1=20 1 14 (1; 1; 1); 15 (1; 1; 2)
No. 18 X22  P(1; 2; 3; 4; 11) 1=12 1 5  12 (1; 1; 1); 13 (1; 1; 1);
1 (1; 1; 1)
4
No. 19 X28  P(1; 3; 4; 5; 14) 1=30 1 13 (1; 1; 1); 12 (1; 1; 1);
1 (1; 1; 1)
5
No. 20 X28  P(3; 4; 5; 7; 8) 1=120 0 13 (1; 1; 1); 3  14 (1; 1; 1);
1 (1; 1; 2); 1 (1; 1; 3)
5 8
No. 21 X30  P(2; 3; 4; 5; 15) 1=60 0 7  12 (1; 1; 1); 2  13 (1; 1; 1);
1 (1; 1; 1); 2  1 (1; 1; 2)
4 5
No. 22 X40  P(3; 4; 5; 7; 20) 1=210 0 13 (1; 1; 1); 2  14 (1; 1; 1);
2  15 (1; 1; 2); 17 (1; 1; 2)
No. 23 X46  P(4; 5; 6; 7; 23) 1=420 0 14 (1; 1; 1); 3  12 (1; 1; 1);
1 (1; 1; 2); 1 (1; 1; 1);
15 (1; 1; 3) 6
7
Table 3: 3-fold canonical hypersurfaces

15.2 Conjecture This list was produced using a computer program. In

fact the program was run much further but produced no more examples. I
152 Working with weighted complete intersections

conjecture that the lists in this sectionPand in Sections 15.4, 16.6, and 16.7
are complete lists, and not limited by ai  100.
15.3 Interesting example The family X46 in P(4; 5; 6; 7; 23) has pg , P2
and P3 all zero. It is interesting to nd canonical 3-folds with as many of
their rst plurigenera equal to zero as possible (see also [F1], Section 4.9).
This is the best such weighted complete intersections example found in these
lists.
15.4 Canonically embedded codimension 2 weighted 3-folds There
are 59 families of 3-fold codimension 2 weighted complete intersections
P satis-
fying the conditions of Theorem 14.4 with !X ' OX (1) and ai  100.
Complete Intersection KX3 pg Singularities
No. 1 X2;5  P(1; 1; 1; 1; 1; 1) 10 6
No. 2 X3;4  P(1; 1; 1; 1; 1; 1) 12 6
No. 3 X3;5  P(1; 1; 1; 1; 1; 2) 15=2 5 12 (1; 1; 1)
No. 4 X4;4  P(1; 1; 1; 1; 1; 2) 8 5
No. 5 X3;6  P(1; 1; 1; 1; 2; 2) 9=2 4 3  12 (1; 1; 1)
No. 6 X4;5  P(1; 1; 1; 1; 2; 2) 5 4 2  12 (1; 1; 1)
No. 7 X2;8  P(1; 1; 1; 1; 1; 4) 4 5
No. 8 X4;6  P(1; 1; 1; 1; 2; 3) 4 4
No. 9 X4;6  P(1; 1; 1; 2; 2; 2) 3 3 6  12 (1; 1; 1)
No. 10 X3;8  P(1; 1; 1; 1; 2; 4) 3 4 2  12 (1; 1; 1)
No. 11 X4;7  P(1; 1; 1; 2; 2; 3) 7=3 3 13 (1; 1; 1); 2  12 (1; 1; 1)
No. 12 X5;6  P(1; 1; 1; 2; 2; 3) 5=2 3 3  12 (1; 1; 1)
No. 13 X6;6  P(1; 1; 1; 2; 3; 3) 2 3
No. 14 X4;8  P(1; 1; 2; 2; 2; 3) 4=3 2 13 (1; 1; 1); 8  12 (1; 1; 1)
No. 15 X6;6  P(1; 1; 2; 2; 2; 3) 3=2 2 9  12 (1; 1; 1)
No. 16 X3;10  P(1; 1; 1; 2; 2; 5) 3=2 3 5  12 (1; 1; 1)
No. 17 X4;9  P(1; 1; 2; 2; 3; 3) 1 2 2  12 (1; 1; 1); 3  13 (1; 1; 1)
No. 18 X6;7  P(1; 1; 2; 2; 3; 3) 7=6 2 3  12 (1; 1; 1); 2  13 (1; 1; 1)
No. 19 X4;10  P(1; 1; 1; 2; 3; 5) 4=3 3 13 (1; 1; 1)
No. 20 X4;10  P(1; 1; 2; 2; 2; 5) 1 2 10  12 (1; 1; 1)
No. 21 X6;8  P(1; 1; 2; 2; 3; 4) 1 2 6  12 (1; 1; 1)
No. 22 X6;8  P(1; 2; 2; 2; 3; 3) 2=3 1 12  12 (1; 1; 1); 2  13 (1; 1; 1)
No. 23 X6;9  P(1; 1; 2; 3; 3; 4) 3=4 2 14 (1; 1; 1); 12 (1; 1; 1)
No. 24 X6;9  P(1; 2; 2; 3; 3; 3) 1=2 1 3  12 (1; 1; 1); 6  13 (1; 1; 1)
No. 25 X4;12  P(1; 1; 2; 2; 3; 6) 2=3 2 4  12 (1; 1; 1); 2  13 (1; 1; 1)
No. 26 X6;10  P(1; 1; 2; 3; 3; 5) 2=3 2 2  13 (1; 1; 1)
No. 27 X6;10  P(1; 2; 2; 2; 3; 5) 1=2 1 15  12 (1; 1; 1)
No. 28 X6;10  P(1; 2; 2; 3; 3; 4) 5=12 1 14 (1; 1; 1); 7  12 (1; 1; 1);
2  13 (1; 1; 1)
 A. R. Iano-Fletcher 153

No. 29 X4;14  P(1; 2; 2; 2; 3; 7) 1=3 1 31 (1; 1; 1); 14 1 2 (1; 1; 1)

1 1
No. 30 X6;12  P(1; 2; 2; 3; 4; 5) 3=10 1 (1; 1; 2); 9  (1; 1; 1)
No. 31 X8;10  P(1; 2; 2; 3; 4; 5) 1=3 1 15 (1; 1; 1); 10 2 1 (1; 1; 1)
3 2
No. 32 X6;12  P(1; 2; 3; 3; 4; 4) 1=4 1 3  12 (1; 1; 1); 3  14 (1; 1; 1)
No. 33 X6;12  P(2; 2; 3; 3; 3; 4) 1=6 0 9  12 (1; 1; 1); 8  13 (1; 1; 1)
No. 34 X6;13  P(1; 2; 3; 3; 4; 5) 13=60 1 1 (1; 1; 1); 1 (1; 1; 2);
14 (1; 1; 1); 25  1 (1; 1; 1)
2 3
No. 35 X9;10  P(1; 2; 3; 3; 4; 5) 1=4 1 1 (1; 1; 1); 2  1 (1; 1; 1);
4 1 2
3  3 (1; 1; 1)
No. 36 X6;14  P(1; 2; 2; 3; 4; 7) 1=4 1 1 (1; 1; 1); 10  1 (1; 1; 1)
X8;12  P(1; 2; 3; 4; 4; 5) 1=5 14 (1; 1; 1); 6  1 (1
2
No. 37 1 5 2 ; 1 1; 1)
No. 38 X6;14  P(2; 2; 2; 3; 3; 7) 1=6 0 21  2 (1; 1; 1); 2  3 (1; 1; 1)
1
No. 39 X8;12  P(2; 2; 3; 3; 4; 5) 2=15 0 1 (1; 1; 2); 12  1 (1; 1; 1);
5 2
4  13 (1; 1; 1)
No. 40 X6;15  P(2; 3; 3; 3; 4; 5) 1=12 0 1 (1; 1; 1); 1 (1; 1; 1);
4 2
10  13 (1; 1; 1)
No. 41 X6;16  P(1; 2; 3; 3; 4; 8) 1=6 1 2  12 (1; 1; 1); 2  13 (1; 1; 1);
2  14 (1; 1; 1)
No. 42 X10;12  P(1; 2; 3; 4; 5; 6) 1=6 1 5  12 (1; 1; 1); 2  13 (1; 1; 1)
No. 43 X10;12  P(2; 2; 3; 4; 5; 5) 1=10 0 15  21 (1; 1; 1); 2  15 (1; 1; 2)
No. 44 X10;12  P(2; 3; 3; 4; 4; 5) 1=12 0 6  12 (1; 1; 1); 4  13 (1; 1; 1);
3  14 (1; 1; 1)
No. 45 X8;15  P(2; 3; 3; 4; 5; 5) 1=15 0 2  12 (1; 1; 1); 5  13 (1; 1; 1);
3  15 (1; 1; 2)
No. 46 X6;18  P(1; 2; 3; 3; 5; 9) 2=15 1 1 (1; 1; 2); 4  1 (1; 1; 1)
5 3
No. 47 X6;18  P(2; 2; 3; 3; 4; 9) 1=12 0 1 (1; 1; 1); 13  1 (1; 1; 1);
4 2
4  13 (1; 1; 1)
No. 48 X10;14  P(2; 2; 3; 4; 5; 7) 1=12 0 1 (1; 1; 1); 1 (1; 1; 1);
3 4
17  12 (1; 1; 1)
No. 49 X6;20  P(1; 2; 3; 4; 5; 10) 1=10 1 3  12 (1; 1; 1); 2  15 (1; 1; 2)
No. 50 X12;14  P(2; 3; 4; 4; 5; 7) 1=20 0 1 (1; 1; 2); 9  1 (1; 1; 1);
5 2
3  14 (1; 1; 1)
No. 51 X12;15  P(1; 3; 4; 5; 6; 7) 1=14 1 1 (1; 1; 2); 1 (1; 1; 1)
7 2
No. 52 X10;18  P(2; 3; 4; 5; 6; 7) 1=28 0 1 (1; 1; 1); 1 (1; 1; 3);
4 7
7  12 (1; 1; 1); 3  13 (1; 1; 1)
No. 53 X12;16  P(2; 3; 4; 5; 6; 7) 4=105 0 1 (1; 1; 2); 1 (1; 1; 2);
5 7
8  12 (1; 1; 1); 2  13 (1; 1; 1)
No. 54 X8;22  P(2; 3; 4; 4; 5; 11) 1=30 0 1 (1; 1; 1); 1 (1; 1; 1);
3 5
10  12 (1; 1; 1); 2  14 (1; 1; 1)
No. 55 X12;18  P(2; 3; 4; 5; 6; 9) 1=30 0 1 (1; 1; 1); 9  1 (1; 1; 1);
5 2
4  13 (1; 1; 1)
154 Working with weighted complete intersections

No. 56 X12;18  P(3; 4; 4; 5; 6; 7) 3=140 0 15 (1; 1; 1); 17 (1; 1; 2);

3  14 (1; 1; 1); 3  12 (1; 1; 1)
No. 57 X10;21  P(3; 4; 5; 5; 6; 7) 1=60 0 14 (1; 1; 1); 16 (1; 1; 1);
3  13 (1; 1; 1); 2  15 (1; 1; 2)
No. 58 X12;21  P(3; 4; 5; 6; 7; 7) 1=70 0 15 (1; 1; 2); 12 (1; 1; 1);
3  17 (1; 1; 2)
No. 59 X12;28  P(3; 4; 5; 6; 7; 14) 1=105 0 15 (1; 1; 1); 2  13 (1; 1; 1);
2  12 (1; 1; 1); 2  17 (1; 1; 2)
Table 4: Canonical codimension 2 complete intersections

16 Q-Fano 3-folds
In [R4], Section 4.3 Reid conjectures that if X is a Q -Fano 3-fold then
OX ( KX ) has a global section. This is false as shown by the following ex-
ample:
16.1 Example The family X12;14 in P(2; 3; 4; 5; 6; 7) is an anticanonically
embedded Fano 3-fold with only the following isolated terminal singularities:
1 of type 15 (4; 1; 2), 2 of type 13 (2; 1; 1) and 7 of type 12 (1; 1; 1). These singu-
larities were determined in 10.3 above.
Since it is quasismooth and of dimension 3, !X ' OX ( 1) and KX3 = 301 .
By an unpublished result of Barlow (see [R4], Corollary 10.3) we have
X rQ2 1
KX
c2(X ) = 24(OX )
singularities Q rQ

where rQ is the index of the singularity Q of type r1Q (1; 1; bQ). So KX
c2 =

101 < 0. However O ( K ) ' O (1) has no global sections.
30 X X X
Experimentation leads to the following:
16.2 Conjecture Every weighted hypersurface Q -Fano 3-fold X , with can-
onical singularities, has a global section of !X1.
This is clear in one particular case.
16.3 Lemma Consider Xd in P(a0 ; : : : ; a4 ) be a family of Q -Fano 3-folds
with only isolated terminal singularities. Suppose also that a0 


 a4
and a4 - d. Then !X1 has a global section.
 A. R. Iano-Fletcher 155

Proof As a4 - d, the vertex P4 is contained in X . The condition for a

terminal singularity at Pi gives that there exists an am such that a4 j am + .
So am = a4 +( ) for some integer . Since  < 0 and a4  am , then   0.
Thus deg(x4( ) xm ) =  and so dim H0 (OX ( ))  1. But H0(!X1) '
H0 (OX ( )), and so !X1 has a global section. 
Notice that when  = 1, there exists a generator xi with deg (xi ) = 1,
that is, a0 = 1.
16.4 Lemma There is a bijection between the following:
(i) the set of families of quasismooth, well-formed weighted surface hyper-
surfaces Sd in P(a1 ; : : : ; a4) with only canonical singularities and trivial
canonical class.
(ii) the set of families of quasismooth weighted 3-folds hypersurfaces Xd in
P(1; a1 ; : : : ; a4 ) with only terminal singularities and !X ' OX ( 1).

Proof Suppose that Sd in P = P(a1 ; : : : ; a4) is a K3 surface, with at worst

canonical singularities. By comparing the conditions in Theorems 13.1 and
14.1 it is clear that the conditions of the latter are satised for X = Xd in
P(1; a1 ; : : : ; an ). Thus X is quasismooth with at worst terminal singularities.
Conversely suppose Xd in P(1; a1 ; : : : ; an) is quasismooth and has at worst
terminal singularities. It can be seen from Theorems 13.1 and 14.1 that only
condition (1ii) of Theorem 13.1 needs proof (the others being either trivially
satised or equivalent in both the surface and the 3-fold case).
Set a0 = 1 and consider ai for i 6= 0. Suppose that condition (1ii) does not
hold. So ai - d ae for all e = 1; : : : ; 4. In particular ai - d. Thus ai j d a0 ,
that is, ai j d 1. Since ai - d then Theorem 14.1 (1iv) gives that there exists
an m 6= 0; i such that ai j am 1. Hence ai j (d 1) (am 1), that is,
ai j d am , a contradiction. So ai j d ae for some e 6= 0; i, which is condition
(1ii) of Theorem 13.1. 
16.5 Note Each singularity on the K3 surface is of type 1r (a; a) for some r
and a, with respect to some pair of the coordinates x1 ; : : : ; x4. Forming the
corresponding Q -Fano 3-fold results in an extra local coordinate x0 at each
singularity, which is thus of type 1r (a; a; 1). A similar result holds for higher
codimensions.
16.6 List of anticanonically embedded (Q -Fano) weighted 3-folds
The previous lemma gives a bijection between Reid's list of 95 families of
weighted K3 surfaces (see Section 13.3 or [R4], Section 4.5) and the 95 families
of
P aquasismooth weighted hypersurface Q -Fano 3-folds, with  = 1 and
i  100. These were found by a computer search and are listed below.
156 Working with weighted complete intersections

Hypersurface KX3 Singularities

No. 1 X4  P(1; 1; 1; 1; 1) 4
No. 2 X5  P(1; 1; 1; 1; 2) 5=2 1 (1;
2 1; 1)
No. 3 X6  P(1; 1; 1; 1; 3) 2
No. 4 X6  P(1; 1; 1; 2; 2) 3=2 3  12 (1; 1; 1)
No. 5 X7  P(1; 1; 1; 2; 3) 7=6 1 (1; 1; 1); 1 (1; 1; 1)
2 3
No. 6 X8  P(1; 1; 1; 2; 4) 1 2  12 (1; 1; 1)
No. 7 X8  P(1; 1; 2; 2; 3) 2=3 4  12 (1; 1; 1); 13 (1; 1; 1)
No. 8 X9  P(1; 1; 1; 3; 4) 3=4 1 (1; 1; 1)
No. 9 X9  P(1; 1; 2; 3; 3) 1=2 14 (1; 1; 1); 3  1 (1; 1; 1)
No. 10 X10  P(1; 1; 1; 3; 5) 2=3 12 (1; 1; 1) 3
3
No. 11 X10  P(1; 1; 2; 2; 5) 1=2 5  12 (1; 1; 1)
No. 12 X10  P(1; 1; 2; 3; 4) 5=12 2  12 (1; 1; 1); 13 (1; 1; 1); 14 (1; 1; 1)
No. 13 X11  P(1; 1; 2; 3; 5) 11=30 1 (1; 1; 1); 1 (1; 1; 1); 1 (1; 1; 2)
No. 14 X12  P(1; 1; 1; 4; 6) 1=2 12 (1; 1; 1) 3 5
2
No. 15 X12  P(1; 1; 2; 3; 6) 1=3 2  12 (1; 1; 1); 2  13 (1; 1; 1)
No. 16 X12  P(1; 1; 2; 4; 5) 3=10 3  12 (1; 1; 1); 15 (1; 1; 1)
No. 17 X12  P(1; 1; 3; 4; 4) 1=4 3  14 (1; 1; 1)
No. 18 X12  P(1; 2; 2; 3; 5) 1=5 6  12 (1; 1; 1); 15 (1; 1; 2)
No. 19 X12  P(1; 2; 3; 3; 4) 1=6 3  12 (1; 1; 1); 4  13 (1; 1; 1)
No. 20 X13  P(1; 1; 3; 4; 5) 13=60 1 (1; 1; 1); 1 (1; 1; 1); 1 (1; 1; 1)
3 4 5
No. 21 X14  P(1; 1; 2; 4; 7) 1=4 3  12 (1; 1; 1); 14 (1; 1; 1)
No. 22 X14  P(1; 2; 2; 3; 7) 1=6 7  12 (1; 1; 1); 13 (1; 1; 1)
No. 23 X14  P(1; 2; 3; 4; 5) 7=60 3  12 (1; 1; 1); 13 (1; 1; 1);
1 (1; 1; 1); 1 (1; 1; 2)
4 5
No. 24 X15  P(1; 1; 2; 5; 7) 3=14 1 (1; 1; 1); 1 (1; 1; 3)
21
No. 25 X15  P(1; 1; 3; 4; 7) 5=28 (1; 1 ; 1); 17 (1; 1; 2)
4 7
No. 26 X15  P(1; 1; 3; 5; 6) 1=6 2  13 (1; 1; 1); 16 (1; 1; 1)
No. 27 X15  P(1; 2; 3; 5; 5) 1=10 1 (1; 1; 1); 3  1 (1; 1; 2)
2 5
No. 28 X15  P(1; 3; 3; 4; 5) 1=12 5  13 (1; 1; 1); 14 (1; 1; 1)
No. 29 X16  P(1; 1; 2; 5; 8) 1=5 2  12 (1; 1; 1); 15 (1; 1; 2)
No. 30 X16  P(1; 1; 3; 4; 8) 1=6 1 (1; 1; 1); 2  1 (1; 1; 1)
No. 31 X16  P(1; 1; 4; 5; 6) 2=15 13 (1; 1; 1); 1 (1;4 1; 1); 1 (1; 1; 1)
2 5 6
No. 32 X16  P(1; 2; 3; 4; 7) 2=21 4  12 (1; 1; 1); 13 (1; 1; 1); 17 (1; 1; 2)
No. 33 X17  P(1; 2; 3; 5; 7) 17
210
1 (1; 1; 1); 1 (1; 1; 1); 1 (1; 1; 2);
12 (1; 1; 3) 3 5
7
No. 34 X18  P(1; 1; 2; 6; 9) 1=6 3  12 (1; 1; 1); 13 (1; 1; 1)
No. 35 X18  P(1; 1; 3; 5; 9) 2=15 2  13 (1; 1; 1); 15 (1; 1; 1)
No. 36 X18  P(1; 1; 4; 6; 7) 3=28 1 (1; 1; 1); 1 (1; 1; 1); 1 (1; 1; 1)
4 2 7
No. 37 X18  P(1; 2; 3; 4; 9) 1=12 4  12 (1; 1; 1); 2  13 (1; 1; 1);
1 (1; 1; 1)
4
 A. R. Iano-Fletcher 157

No. 38 X18  P(1; 2; 3; 5; 8) 3=40 2  21 (1; 1; 1); 15 (1; 1; 2); 81 (1; 1; 3)

No. 39 X18  P(1; 3; 4; 5; 6) 1=20 3  13 (1; 1; 1); 14 (1; 1; 1);
1 (1; 1; 1); 1 (1; 1; 1)
2 5
No. 40 X19  P(1; 3; 4; 5; 7) 19
420
1 (1; 1; 1); 1 (1; 1; 1); 1 (1; 1; 2);
13 (1; 1; 2) 4 5
7
No. 41 X20  P(1; 1; 4; 5; 10) 1=10 12 (1; 1; 1); 2  15 (1; 1; 1)
No. 42 X20  P(1; 2; 3; 5; 10) 1=15 2  12 (1; 1; 1); 13 (1; 1; 1);
2  15 (1; 1; 2)
No. 43 X20  P(1; 2; 4; 5; 9) 1=18 5  12 (1; 1; 1); 19 (1; 1; 2)
No. 44 X20  P(1; 2; 5; 6; 7) 1=21 3  12 (1; 1; 1); 16 (1; 1; 1); 71 (1; 1; 3)
No. 45 X20  P(1; 3; 4; 5; 8) 1=24 13 (1; 1; 1); 2  14 (1; 1; 1); 81 (1; 1; 3)
No. 46 X21  P(1; 1; 3; 7; 10) 1=10 101 (1; 1; 3)
No. 47 X21  P(1; 1; 5; 7; 8) 3=40 15 (1; 1; 2); 18 (1; 1; 1)
No. 48 X21  P(1; 2; 3; 7; 9) 1=18 21 (1; 1; 1); 2  13 (1; 1; 1); 91 (1; 1; 4)
No. 49 X21  P(1; 3; 5; 6; 7) 1=30 3  13 (1; 1; 1); 15 (1; 1; 2); 61 (1; 1; 1)
No. 50 X22  P(1; 1; 3; 7; 11) 2=21 13 (1; 1; 1); 17 (1; 1; 2)
No. 51 X22  P(1; 1; 4; 6; 11) 1=12 41 (1; 1; 1); 12 (1; 1; 1); 61 (1; 1; 1)
No. 52 X22  P(1; 2; 4; 5; 11) 1=20 5  12 (1; 1; 1); 14 (1; 1; 1); 51 (1; 1; 1)
No. 53 X24  P(1; 1; 3; 8; 12) 1=12 2  13 (1; 1; 1); 14 (1; 1; 1)
No. 54 X24  P(1; 1; 6; 8; 9) 1=18 12 (1; 1; 1); 13 (1; 1; 1); 91 (1; 1; 1)
No. 55 X24  P(1; 2; 3; 7; 12) 1=21 2  12 (1; 1; 1); 2  13 (1; 1; 1);
1 (1; 1; 3)
7
No. 56 X24  P(1; 2; 3; 8; 11) 1=22 3  12 (1; 1; 1); 111 (1; 1; 4)
No. 57 X24  P(1; 3; 4; 5; 12) 1=30 2  13 (1; 1; 1); 2  14 (1; 1; 1);
1 (1; 1; 2)
5
No. 58 X24  P(1; 3; 4; 7; 10) 1=35 12 (1; 1; 1); 17 (1; 1; 2); 101 (1; 1; 3)
No. 59 X24  P(1; 3; 6; 7; 8) 1=42 4  13 (1; 1; 1); 12 (1; 1; 1); 71 (1; 1; 1)
No. 60 X24  P(1; 4; 5; 6; 9) 1=45 2  12 (1; 1; 1); 15 (1; 1; 1);
1 (1; 1; 1); 1 (1; 1; 2)
3 9
No. 61 X25  P(1; 4; 5; 7; 9) 5
252
1 (1; 1; 1); 1 (1; 1; 3); 1 (1; 1; 2)
4 7 9
No. 62 X26  P(1; 1; 5; 7; 13) 2=35 15 (1; 1; 2); 17 (1; 1; 1)
No. 63 X26  P(1; 2; 3; 8; 13) 1=24 3  21 (1; 1; 1); 13 (1; 1; 1); 81 (1; 1; 3)
No. 64 X26  P(1; 2; 5; 6; 13) 1=30 4  12 (1; 1; 1); 15 (1; 1; 2); 61 (1; 1; 1)
No. 65 X27  P(1; 2; 5; 9; 11) 3
110
1 (1; 1; 1); 1 (1; 1; 1); 1 (1; 1; 5)
2 5 11
No. 66 X27  P(1; 5; 6; 7; 9) 1=70 15 (1; 1; 1); 16 (1; 1; 1); 31 (1; 1; 1);
1 (1; 1; 3)
7
No. 67 X28  P(1; 1; 4; 9; 14) 1=18 12 (1; 1; 1); 19 (1; 1; 2)
No. 68 X28  P(1; 3; 4; 7; 14) 1=42 31 (1; 1; 1); 12 (1; 1; 1); 2  71 (1; 1; 2)
No. 69 X28  P(1; 4; 6; 7; 11) 1=66 2  12 (1; 1; 1); 16 (1; 1; 1);
1
11 (1; 1; 3)
No. 70 X30  P(1; 1; 4; 10; 15) 1=20 14 (1; 1; 1); 12 (1; 1; 1); 51 (1; 1; 1)
158 Working with weighted complete intersections

No. 71 X30  P(1; 1; 6; 8; 15) 1=24 21 (1; 1; 1); 13 (1; 1; 1); 18 (1; 1; 1)
No. 72 X30  P(1; 2; 3; 10; 15) 1=30 3  12 (1; 1; 1); 2  13 (1; 1; 1);
1 (1; 1; 2)
5
No. 73 X30  P(1; 2; 6; 7; 15) 1=42 5  12 (1; 1; 1); 13 (1; 1; 1); 17 (1; 1; 1)
No. 74 X30  P(1; 3; 4; 10; 13) 1=52 14 (1; 1; 1); 12 (1; 1; 1); 131 (1; 1; 4)
No. 75 X30  P(1; 4; 5; 6; 15) 1=60 41 (1; 1; 1); 2  12 (1; 1; 1);
2  15 (1; 1; 1); 13 (1; 1; 1)
No. 76 X30  P(1; 5; 6; 8; 11) 1=88 12 (1; 1; 1); 18 (1; 1; 3); 111 (1; 1; 2)
No. 77 X32  P(1; 2; 5; 9; 16) 1=45 2  21 (1; 1; 1); 15 (1; 1; 1); 19 (1; 1; 4)
No. 78 X32  P(1; 4; 5; 7; 16) 1=70 2  14 (1; 1; 1); 15 (1; 1; 1); 17 (1; 1; 3)
No. 79 X33  P(1; 3; 5; 11; 14) 1=70 15 (1; 1; 1); 141 (1; 1; 5)
No. 80 X34  P(1; 3; 4; 10; 17) 1=60 13 (1; 1; 1); 14 (1; 1; 1); 12 (1; 1; 1);
1
10 (1; 1; 3)
No. 81 X34  P(1; 4; 6; 7; 17) 1=84 14 (1; 1; 1); 2  12 (1; 1; 1);
1 (1; 1; 1); 1 (1; 1; 2)
6 7
No. 82 X36  P(1; 1; 5; 12; 18) 1=30 15 (1; 1; 2); 16 (1; 1; 1)
No. 83 X36  P(1; 3; 4; 11; 18) 1=66 2  13 (1; 1; 1); 12 (1; 1; 1);
1
11 (1; 1; 3)
No. 84 X36  P(1; 7; 8; 9; 12) 168
1 1 (1; 1; 3); 1 (1; 1; 1); 1 (1; 1; 1);
17 (1; 1; 1) 8 4
3
No. 85 X38  P(1; 3; 5; 11; 19) 165
2 1 1 1
3 (1; 1; 1); 5 (1; 1; 1); 11 (1; 1; 4)
No. 86 X38  P(1; 5; 6; 8; 19) 120
1 1 1 1
51 (1; 1; 1); 6 (1; 1; 1); 2 (1; 1; 1);
8 (1; 1; 3)
No. 87 X40  P(1; 5; 7; 8; 20) 140 2  51 (1; 1; 2); 17 (1; 1; 1); 14 (1; 1; 1)
1
No. 88 X42  P(1; 1; 6; 14; 21) 1=42 12 (1; 1; 1); 13 (1; 1; 1); 17 (1; 1; 1)
No. 89 X42  P(1; 2; 5; 14; 21) 1=70 3  12 (1; 1; 1); 15 (1; 1; 1); 17 (1; 1; 3)
No. 90 X42  P(1; 3; 4; 14; 21) 1=84 2  13 (1; 1; 1); 14 (1; 1; 1);
1 (1; 1; 1); 1 (1; 1; 2)
2 7
No. 91 X44  P(1; 4; 5; 13; 22) 130
1 1 (1; 1; 1); 1 (1; 1; 2); 1 (1; 1; 3)
2 5 13
No. 92 X48  P(1; 3; 5; 16; 24) 120
1 2  31 (1; 1; 1); 15 (1; 1; 1); 18 (1; 1; 3)
No. 93 X50  P(1; 7; 8; 10; 25) 280
1 1 1 1
71 (1; 1; 2); 8 (1; 1; 1); 2 (1; 1; 1);
5 (1; 1; 2)
No. 94 X54  P(1; 4; 5; 18; 27) 180 41 (1; 1; 1); 12 (1; 1; 1); 15 (1; 1; 2);
1
1 (1; 1; 2)
9
No. 95 X66  P(1; 5; 6; 22; 33) 330 51 (1; 1; 2); 12 (1; 1; 1); 13 (1; 1; 1);
1
1
11 (1; 1; 2)
Table 5: The 95 anticanonical 3-folds hypersurfaces

16.7 Codimension 2 Q -Fano weighted complete intersections There

are 85 codimension 2 quasismooth Q -Fano weighted complete intersections
 A. R. Iano-Fletcher 159
P
which satisfy the conditions of Theorem 14.4,  = 1 and ai  100.
Complete intersection KX3 Singularities
No. 1 X2;3  P(1; 1; 1; 1; 1; 1) 6
No. 2 X3;3  P(1; 1; 1; 1; 1; 2) 9=2 21 (1; 1; 1)
No. 3 X3;4  P(1; 1; 1; 1; 2; 2) 3 2  21 (1; 1; 1)
No. 4 X4;4  P(1; 1; 1; 1; 2; 3) 8=3 13 (1; 1; 1)
No. 5 X4;4  P(1; 1; 1; 2; 2; 2) 2 4  21 (1; 1; 1)
No. 6 X4;5  P(1; 1; 1; 2; 2; 3) 5=3 13 (1; 1; 1); 2  12 (1; 1; 1)
No. 7 X4;6  P(1; 1; 1; 2; 3; 3) 4=3 2  31 (1; 1; 1)
No. 8 X4;6  P(1; 1; 2; 2; 2; 3) 1 6  12 (1; 1; 1)
No. 9 X5;6  P(1; 1; 1; 2; 3; 4) 5=4 14 (1; 1; 1); 12 (1; 1; 1)
No. 10 X5;6  P(1; 1; 2; 2; 3; 3) 5=6 3  21 (1; 1; 1); 2  13 (1; 1; 1)
No. 11 X6;6  P(1; 1; 1; 2; 3; 5) 6=5 15 (1; 1; 2)
No. 12 X6;6  P(1; 1; 2; 2; 3; 4) 3=4 14 (1; 1; 1); 4  12 (1; 1; 1)
No. 13 X6;6  P(1; 1; 2; 3; 3; 3) 2=3 4  31 (1; 1; 1)
No. 14 X6;6  P(1; 2; 2; 2; 3; 3) 1=2 9  21 (1; 1; 1)
No. 15 X6;7  P(1; 1; 2; 2; 3; 5) 7=10 15 (1; 1; 2); 3  12 (1; 1; 1)
No. 16 X6;7  P(1; 1; 2; 3; 3; 4) 7=12 14 (1; 1; 1); 12 (1; 1; 1);
2  13 (1; 1; 1)
No. 17 X6;8  P(1; 1; 1; 3; 4; 5) 4=5 15 (1; 1; 1)
No. 18 X6;8  P(1; 1; 2; 3; 3; 5) 8=15 15 (1; 1; 2); 2  13 (1; 1; 1)
No. 19 X6;8  P(1; 1; 2; 3; 4; 4) 1=2 2  21 (1; 1; 1); 2  14 (1; 1; 1)
No. 20 X6;8  P(1; 2; 2; 3; 3; 4) 1=3 6  21 (1; 1; 1); 2  13 (1; 1; 1)
No. 21 X6;9  P(1; 1; 2; 3; 4; 5) 9=20 14 (1; 1; 1); 15 (1; 1; 2);
1 (1; 1; 1)
2
No. 22 X7;8  P(1; 1; 2; 3; 4; 5) 7=15 13 (1; 1; 1); 15 (1; 1; 1);
2  12 (1; 1; 1)
No. 23 X6;10  P(1; 1; 2; 3; 5; 5) 2=5 2  15 (1; 1; 2)
No. 24 X6;10  P(1; 2; 2; 3; 4; 5) 1=4 14 (1; 1; 1); 7  12 (1; 1; 1)
No. 25 X8;9  P(1; 1; 2; 3; 4; 7) 3=7 17 (1; 1; 2); 2  12 (1; 1; 1)
No. 26 X8;9  P(1; 1; 3; 4; 4; 5) 3=10 15 (1; 1; 1); 2  14 (1; 1; 1)
No. 27 X8;9  P(1; 2; 3; 3; 4; 5) 1=5 15 (1; 1; 2); 2  12 (1; 1; 1);
3  13 (1; 1; 1)
No. 28 X8;10  P(1; 1; 2; 3; 5; 7) 8=21 13 (1; 1; 1); 17 (1; 1; 3)
No. 29 X8;10  P(1; 1; 2; 4; 5; 6) 1=3 16 (1; 1; 1); 3  12 (1; 1; 1)
No. 30 X8;10  P(1; 1; 3; 4; 5; 5) 4=15 13 (1; 1; 1); 2  15 (1; 1; 1)
No. 31 X8;10  P(1; 2; 3; 4; 4; 5) 1=6 13 (1; 1; 1); 4  12 (1; 1; 1);
2  14 (1; 1; 1)
No. 32 X9;10  P(1; 1; 2; 3; 5; 8) 3=8 18 (1; 1; 3); 12 (1; 1; 1)
No. 33 X9;10  P(1; 1; 3; 4; 5; 6) 1=4 14 (1; 1; 1); 16 (1; 1; 1);
1 (1; 1; 1)
3
160 Working with weighted complete intersections

No. 34 X9;10  P(1; 2; 2; 3; 5; 7) 3=14 71 (1; 1; 3); 5  12 (1; 1; 1)

No. 35 X9;10  P(1; 2; 3; 4; 5; 5) 3=20 14 (1; 1; 1); 2  12 (1; 1; 1);
2  15 (1; 1; 2)
No. 36 X8;12  P(1; 1; 3; 4; 5; 7) 8=35 15 (1; 1; 1); 17 (1; 1; 2)
No. 37 X8;12  P(1; 2; 3; 4; 5; 6) 2=15 15 (1; 1; 1); 4  12 (1; 1; 1);
2  13 (1; 1; 1)
No. 38 X9;12  P(1; 2; 3; 4; 5; 7) 9=70 15 (1; 1; 2); 17 (1; 1; 2);
3  12 (1; 1; 1)
No. 39 X10;11  P(1; 2; 3; 4; 5; 7) 11
84
1 1
31 (1; 1; 1); 4 (1;1 1; 1);
7 (1; 1; 3); 2  2 (1; 1; 1)
No. 40 X10;12  P(1; 1; 3; 4; 5; 9) 2=9 9 (1; 1; 2); 13 (1; 1; 1)
1
No. 41 X10;12  P(1; 1; 3; 5; 6; 7) 4=21 17 (1; 1; 1); 2  13 (1; 1; 1)
No. 42 X10;12  P(1; 1; 4; 5; 6; 6) 1=6 12 (1; 1; 1); 2  16 (1; 1; 1)
No. 43 X10;12  P(1; 2; 3; 4; 5; 8) 1=8 18 (1; 1; 3); 3  12 (1; 1; 1);
1 (1; 1; 1)
4
No. 44 X10;12  P(1; 2; 3; 5; 5; 7) 4=35 17 (1; 1; 3); 2  15 (1; 1; 2)
No. 45 X10;12  P(1; 2; 4; 5; 5; 6) 1=10 5  12 (1; 1; 1); 2  15 (1; 1; 1)
No. 46 X10;12  P(1; 3; 3; 4; 5; 7) 2=21 17 (1; 1; 2); 4  13 (1; 1; 1)
No. 47 X10;12  P(1; 3; 4; 4; 5; 6) 1=12 2  13 (1; 1; 1); 3  14 (1; 1; 1);
1 (1; 1; 1)
2
No. 48 X11;12  P(1; 1; 4; 5; 6; 7) 11
70
1 1
51 (1; 1; 1); 7 (1; 1; 1);
2 (1; 1; 1)
No. 49 X10;14  P(1; 1; 2; 5; 7; 9) 1
2=9 9 (1; 1; 4)
No. 50 X10;14  P(1; 2; 3; 5; 7; 7) 2=21 13 (1; 1; 1); 2  17 (1; 1; 3)
No. 51 X10;14  P(1; 2; 4; 5; 6; 7) 1=12 14 (1; 1; 1); 16 (1; 1; 1);
5  12 (1; 1; 1)
No. 52 X10;15  P(1; 2; 3; 5; 7; 8) 5=56 17 (1; 1; 3); 18 (1; 1; 3);
1 (1; 1; 1)
2
No. 53 X12;13  P(1; 3; 4; 5; 6; 7) 13
210
1 1
5 (1;1 1; 1); 7 (1;1 1; 2);
2  3 (1; 1; 1); 2 (1; 1; 1)
No. 54 X12;14  P(1; 1; 3; 4; 7; 11) 2=11 111 (1; 1; 3)
No. 55 X12;14  P(1; 1; 4; 6; 7; 8) 1=8 18 (1; 1; 1); 12 (1; 1; 1);
1 (1; 1; 1)
4
No. 56 X12;14  P(1; 2; 3; 4; 7; 10) 1=10 101 (1; 1; 3); 4  12 (1; 1; 1)
No. 57 X12;14  P(1; 2; 3; 5; 7; 9) 4=45 15 (1; 1; 2); 19 (1; 1; 4);
1 (1; 1; 1)
3
No. 58 X12;14  P(1; 3; 4; 5; 7; 7) 2=35 15 (1; 1; 2); 2  17 (1; 1; 2)
No. 59 X12;14  P(1; 4; 4; 5; 6; 7) 1=20 15 (1; 1; 1); 3  14 (1; 1; 1);
2  12 (1; 1; 1)
No. 60 X12;14  P(2; 3; 4; 5; 6; 7) 1=30 15 (1; 1; 2); 7  12 (1; 1; 1);
2  13 (1; 1; 1)
 A. R. Iano-Fletcher 161

No. 61 X12;15  P(1; 1; 4; 5; 6; 11) 3=22 111 (1; 1; 2); 12 (1; 1; 1)

No. 62 X12;15  P(1; 3; 4; 5; 6; 9) 1=18 19 (1; 1; 2); 3  13 (1; 1; 1);
1 (1; 1; 1)
2
No. 63 X12;15  P(1; 3; 4; 5; 7; 8) 3=56 17 (1; 1; 2); 18 (1; 1; 3);
1 (1; 1; 1)
4
No. 64 X12;16  P(1; 2; 5; 6; 7; 8) 2=35 15 (1; 1; 2); 17 (1; 1; 1);
4  12 (1; 1; 1)
No. 65 X14;15  P(1; 2; 3; 5; 7; 12) 1=12 121 (1; 1; 5); 12 (1; 1; 1);
1 (1; 1; 1)
3
No. 66 X14;15  P(1; 2; 5; 6; 7; 9) 1=18 16 (1; 1; 1); 19 (1; 1; 4);
2  12 (1; 1; 1)
No. 67 X14;15  P(1; 3; 4; 5; 7; 10) 1=20 14 (1; 1; 1); 101 (1; 1; 3);
1 (1; 1; 2)
5
No. 68 X14;15  P(1; 3; 5; 6; 7; 8) 1=24 16 (1; 1; 1); 18 (1; 1; 3);
2  13 (1; 1; 1)
No. 69 X14;16  P(1; 1; 5; 7; 8; 9) 4=45 15 (1; 1; 2); 19 (1; 1; 1)
No. 70 X14;16  P(1; 3; 4; 5; 7; 11) 1658 13 (1; 1; 1); 15 (1; 1; 2);
1
11 (1; 1; 3)
No. 71 X14;16  P(1; 4; 5; 6; 7; 8) 1=30 15 (1; 1; 2); 16 (1; 1; 1);
1 (1; 1; 1); 2  1 (1; 1; 1)
2 4
No. 72 X15;16  P(1; 2; 3; 5; 8; 13) 1=13 131 (1; 1; 5); 2  12 (1; 1; 1)
No. 73 X15;16  P(1; 3; 4; 5; 8; 11) 1=22 111 (1; 1; 4); 2  14 (1; 1; 1)
No. 74 X14;18  P(1; 2; 3; 7; 9; 11) 2=33 111 (1; 1; 5); 2  13 (1; 1; 1)
No. 75 X14;18  P(1; 2; 6; 7; 8; 9) 1=24 18 (1; 1; 1); 5  12 (1; 1; 1);
1 (1; 1; 1)
3
No. 76 X12;20  P(1; 4; 5; 6; 7; 10) 1=35 17 (1; 1; 2); 2  12 (1; 1; 1);
2  15 (1; 1; 1)
No. 77 X16;18  P(1; 1; 6; 8; 9; 10) 1=15 101 (1; 1; 1); 12 (1; 1; 1);
1 (1; 1; 1)
3
No. 78 X16;18  P(1; 4; 6; 7; 8; 9) 1=42 17 (1; 1; 1); 2  12 (1; 1; 1);
2  14 (1; 1; 1); 13 (1; 1; 1)
No. 79 X18;20  P(1; 4; 5; 6; 9; 14) 1=42 141 (1; 1; 3); 2  12 (1; 1; 1);
1 (1; 1; 1)
3
No. 80 X18;20  P(1; 4; 5; 7; 9; 13) 2=91 17 (1; 1; 3); 131 (1; 1; 3)
No. 81 X18;20  P(1; 5; 6; 7; 9; 11) 2314 17 (1; 1; 3); 111 (1; 1; 2);
1 (1; 1; 1)
3
No. 82 X18;22  P(1; 2; 5; 9; 11; 13) 2=65 51 (1; 1; 1); 131 (1; 1; 6)
No. 83 X20;21  P(1; 3; 4; 7; 10; 17) 1=34 171 (1; 1; 5); 12 (1; 1; 1)
No. 84 X18;30  P(1; 6; 8; 9; 10; 15) 1201 18 (1; 1; 1); 2  12 (1; 1; 1);
2  13 (1; 1; 1); 15 (1; 1; 1)
162 Working with weighted complete intersections

No. 85 X24;30  P(1; 8; 9; 10; 12; 15) 1

180
1 (1;
1; 1); 12 (1; 1; 1);
19 (1;
1; 1); 13 (1; 1; 1);
14 (1;
1; 2)
5
Table 6: Anticanonical 3-fold codimension 2 complete intersections

16.8 Note X12;14 in P(2; 3; 4; 5; 6; 7) is the only element in the above list with
ai  2 for all i (see Example 16.1).

17 The plurigenus formula

Before we describe Reid's table method for producing examples of weighted
complete intersection we must state the plurigenus formula for canonical and
Q -Fano 3-folds.

17.1 Denition For a singularity Q of type 1r (1; 1; b) dene:

80
< if n = 0; 1;
l(Q; n) = :nP1 bk(r bk)
2r if n  2;
k=1
where x denotes the smallest nonnegative residue of x modulo r. This is
extended to negative integers via:
l( n) = l(n + 1)
for all n  0. This is for consistency with Serre duality. For a collection (or
basket ) B of singularities dene:
X
l (n) = l(q; n)
Q2B
for all n 2 Z.
From [F1], Theorem 2.5, equation (4) (see also [R4], Chapter III) we have
the following:
17.2 Theorem For any projective 3-fold X , with at worst canonical singu-
larities, there exists a basket B of singularities such that

(OX (nKX )) = (2n 1)12nr(n 1) KX3 (2n 1)(OX ) + l(n)

for all n 2 Z.
 A. R. Iano-Fletcher 163

17.3 Canonical 3-folds Let X be a canonical 3-fold. Then KX is ample

and we have:
Pn = (OX (nKX )) = (2n 1)12nr(n 1) KX3 (2n 1)(OX ) + l(n)
for all n  2. This formula is Reid's exact plurigenus formula.
17.4 -Fano 3-fold complete intersections If X is a Q -Fano 3-fold then
Q
KX is ample. Moreover if X is also a complete intersection then (OX ) = 1.
So:
3
P n = (OX ( nKX )) = (2n + 1)12nr(n + 1) ( KX ) + (2n + 1) l(n + 1)
for all n  1.

18 The Reid table method

Consider a complete intersection Xd1 ;:::;dc in P(a0 ; : : : ; an). The Poincare se-
ries (see [WPS], Section 3.4 and compare [AM], 11.1) corresponding to the
coordinate ring R of X is:
X
1
P (t) = h0(X; OX (n))tn
Q n=0
i=c (1 tdi ) :
=Q i=1
i=n (1
i=0 tai )
P
1
Moreover, if !X ' OX (1) then P (t) = Pn(X )tn, where Pn(X ) are the
n=0
plurigenera of X . In the case of a Q -Fano 3-fold with !X ' OX ( 1) then
P (t) = P P n(X )tn, where P n(X ) are the anti-plurigenera of X .
1
n=0
18.1 Example X6 in P4 has Poincare series
P (t) = (1 t6 ) = 1 + t + 5t2 + 15t3 +



(1 t)5
So pg = 1, P2 = 5, P3 = 15, etc.
18.2 Question Given a list of plurigenera (which could arise from a record
of pluridata) does there exist a complete intersection with !X ' OX (1)?
The following lemma due to Reid helps answer the above.
164 Working with weighted complete intersections

18.3 Lemma Given a sequence p0 = 1, p1, p2, : : : such that

X
1 Qi=c (1 tdi )
p i ti = Qii==1n
i=0 (1 tai )
i=0
for some fdi; ai g. Then these fdi; aig are unique up to ai 6= dj and are
determinable.
Proof The following is a constructive proof. Let qi0 = pi. So
X
1 Q(1 tdi )
qi0 ti =Q :
i=0 (1 tai )
Without loss of generality assume that dc 


 d1 and an 


 a0.
Clearly we may assume a0 6= d1 or else these two terms would cancel. There
are two cases:
(i) a0 < d1 . Let a0 occur with multiplicity . Then P (t) = 1+ ta0 +higher
order terms. So the rst nonzero qi0 is qa00 =  < 0. Dene qi1 =
qi0 qi0 a0 , where qi0 = 0 if i < 0. Then qa10 = qa00 1. Thus
X
1 X
1
qi1ti = (qi0 qi0 a0 )ti
i=0 i=0
X
1
= (1 ta0 ) qi0ti
Qc (1 i=0tdi )
= Qin=1 :
(1 tai )
i=1
This involves one less ai .
(ii) d1 < a0 . Let d1 occur with multiplicity . Then P (t) = 1 td1 +higher
order terms. So the rst nonzero qi0 is qd01 =  < 0. Dene qi1 =
qi0 + qi1 d1 , for i = 1; 2; : : : where qi1 = 0 if i<0. This corresponds to:
X
1 X
1
qi1ti = (qi0 + qi1 d1 )ti
i=0 i=0
X1
= (qi0 + qi0 d + qi0 2d +


)ti
1 1

Q
i=0
c
= Qni=2 (1 tai ) :
di
(1 t )
i=0
This involves one less di.
 A. R. Iano-Fletcher 165

Repetition of the above steps clearly terminates when

X
1
qibti = 1
i=0

By induction on the number of ai and dj it is clear that the process uniquely

determines the ai and dj . 

18.4 The table method The proof of the above lemma allows us to con-
struct a weighted complete intersection from a list of `plurigenera'. This
construction is easily set out in the form of a table. In the rst column write
down the integers f0; 1; 2; : : : g and in the second the list f1; P1; P2; : : : g. Let
the nth column be denoted by qin for i = 0; 1; : : : . Each successive column is
obtained as follows. Look down the list fqing of the nth column to nd the
position of the rst nonzero entry (disregard the initial 1 at the top of the
column). Suppose this is in row r. There are 2 cases:
(i) this entry is positive. First enter (r) at the head of this column. This
will keep a record of the degrees of the generators. The (n +1)th column
is obtained by the rule:
qin+1 = qin qin r ;
assuming that qin = 0 for all i < 0.
(ii) this entry is negative. First enter ( r) at the head of this column. This
will keep a record of the degrees of the relations. The (n + 1)st column
is obtained by the rule:
qin+1 = qin qin+1r ;
assuming that qin+1 = 0 and for all i < 0.
The process is clearly dened and the integers at the head of each column
keep track of the ai and di .

18.5 Example Consider the record of pluridata K 3 = 61 ,  = 1, pg = 0,

9 singularities of type 12 (1; 1; 1) and 8 singularities of type 31 (2; 1; 1). Using
Reid's plurigenus formula (see Section 17) the plurigenera Pn corresponding
to this record was calculated and is given below. The table obtained is the
166 Working with weighted complete intersections

following:
n Pn (2) (2) (3) (3) (3) (4) ( 6) ( 12)
0 1 1 1 1 1 1 1 1 1
1 0 0 0 0 0 0 0 0 0
2 2 1 0 0 0 0 0 0 0
3 3 3 3 2 1 0 0 0 0
4 4 2 1 1 1 1 0 0 0
5 6 3 0 0 0 0 0 0 0
6 11 7 5 2 0 1 1 0 0
7 12 6 3 2 1 0 0 0 0
8 19 8 1 1 1 1 0 0 0
9 25 13 7 2 0 0 0 0 0
10 32 13 5 2 0 1 0 0 0
11 41 16 3 2 1 0 0 0 0
12 54 22 9 2 0 0 1 1 0
13 64 23 7 2 0 0 0 0 0
14 81 27 5 2 0 1 0 0 0
15 98 34 11 2 0 0 0 0 0
16 117 36 9 2 0 0 0 0 0
17 139 41 7 2 0 0 0 0 0
18 166 49 13 2 0 0 1 0 0
19 191 52 11 2 0 0 0 0 0
20 224 58 9 2 0 0 0 0 0
This gives X6;12 in P(2; 2; 3; 3; 3; 4), which has the above record.
18.6 Note Of course this method cannot tell the dierence between X6 in
P(1; 1; 1; 2) and the example of V. Iliev X3;6 in P(1; 1; 1; 2; 3), in which the
cubic relation does not involve the degree 3 generator.
However in this section we are only interested in the general member of
a family of weighted complete intersections and so Iliev's example does not
occur.
18.7 Warning Although in general it is clear when this process stops, it
is not clear when it is worth continuing with a particular list of integers.
18.8 The analysis This process is basically the same as that in Sec-
tion 12.6 on the coordinate ring
M
R= Rm :
m0
 A. R. Iano-Fletcher 167

Starting from the dimensions of each Rm the degrees of the generators and
relations can be found. At each stage it is assumed that the monomials are
linearly independent unless
either there already exist some relations of a lower degree,
or a relation is forced by the dimension not being large enough.
For the above example we have the following analysis:
deg dim monomials
0 1 1
1 0 ;
2 2 x0 ; x1
3 3 y0; y1; y2
4 4 x20 ; x0 x1; x21 ; z
5 6 x0 y0; x0 y1; x0y2; x1 y0; x1y1; x1 y2
6 11 x30 ; x20 x1; x0 x21 ; x31; y02; y0y1; y0y2; y12; y1y2; y22; x0 z; x1 z
1 relation.
If this calculation is continued only one more relation is found, of degree 12
18.9 Canonical 3-fold complete intersections The formula:
P2 = 21 KX3 3(1 pg ) + l(2)
limits the value of pg (since KX3 > 0) and denes KX3 in terms of a particular
basket of singularities and P2.
18.10 Q -Fano complete intersections The formula:
P 1 = 21 KX3 + 3 l(2)
denes KX3 in terms of a particular basket of singularities and P 1.
18.11 The search The search through all combinations of P  0 (P2 = P
for canonical 3-folds and P 1 = P for the Fano case) and baskets will give
every possible list of plurigenera (respectively anti-plurigenera). Hence a list
of canonically (respectively anticanonically) embedded complete intersections
can be found. Of course this is not a nite search, and requires a computer
to make any reasonable progress.
The order of the search was as follows. Let Qi for i = 0; 1; : : : be a
list of the types of 3-fold cyclic quotient singularity 1r (1; 1; a) in order of
increasing index r and increasing a within each index. So Q0 = 21 (1; 1; 1),
Q1 = 31 (1; 1; 1), etc. The program took 2 integer arguments l and u, and
P
1
searched through all baskets fni  Qi g such that l  ni (i + 2) < u.
i=0
168 Working with weighted complete intersections

18.12 The raw list Here is the rst part of the list produced by the search
program (with arguments l = 0, u = 8).
X6  P(1; 1; 1; 1; 3) X12  P(1; 1; 1; 4; 6)
X4  P(1; 1; 1; 1; 1) X5  P(1; 1; 1; 1; 2)
X8  P(1; 1; 1; 2; 4) X10  P(1; 1; 1; 3; 5)
X2;3  P(1; 1; 1; 1; 1; 1) X3;3  P(1; 1; 1; 1; 1; 2)
X3;4  P(1; 1; 1; 1; 2; 2) X6  P(1; 1; 1; 2; 2)
X4;4  P(1; 1; 1; 1; 2; 3) X7  P(1; 1; 1; 2; 3)
X9  P(1; 1; 1; 3; 4) X2;2;2  P(1; 1; 1; 1; 1; 1; 1)
X6;6  P(1; 1; 1; 2; 3; 3) X12  P(1; 1; 2; 3; 4)
X4;4  P(1; 1; 1; 2; 2; 2) X10  P(1; 1; 2; 2; 5)
X4;5  P(1; 1; 1; 2; 2; 3) X18  P(1; 1; 2; 6; 9)
X4;6  P(1; 1; 1; 2; 3; 3) X5;6  P(1; 1; 1; 2; 3; 4)
X6;8  P(1; 1; 1; 3; 4; 5)
18.13 Renement Of course this list contains complete intersections al-
ready obtained in other ways (see Sections 15 and 16) and some intersections
which do not meet the requirements; that is,
(1) dimension 3,
(2) quasismooth but not the intersection of a linear cone with other hyper-
surfaces,
(3) canonically or anticanonically embedded,
(4) and have at worst terminal singularities.
The example X6;22 in P(2; 2; 3; 4; 5; 11) from the raw list is not quasismooth,
since the polynomial of degree 6 does not involve the generator of weight 5.
We use the following lemma to cut out a large number of elements from
the raw list produced by the search program.
18.14 Lemma Let Xd ;:::;dc in P(a0 ; : : : ; an) be quasismooth but not an inter-
1
section of a linear cone with other hypersurfaces. Suppose also that d1; : : : ; dc
and a0 ; : : : ; an are in increasing order. Then:
(i) dc > an, dc 1 > an 1, : : : , and d1 > an c+1 .

(ii) if dc 1 < an then an j dc.

 A. R. Iano-Fletcher 169

Proof (i). Suppose dc > an , : : : , dc k+1 > an k+1 and dc k < an k for some
k = 0; : : : ; c 1. So di < an k for all i  c k. Therefore the polynomials
f1 ; : : : ; fn k do not involve the variables xn k ; : : : ; xn.
Let  be the coordinate (k + 1)-plane in A n + 1 given by x0 =


=
xn k 1 = 0. So f1 ; : : : ; fn k are identically zero on . Dene Z = (fc k+1 =



= fc = 0) \ . Thus dim Z  1 and so Z 0 is nonempty. Let Q 2 Z 0.
Then @fi =@xj are zero at Q for all i  c k and for all j . Therefore
0 1
@f1 =@x0 (Q) : : : @f1 =@xn (Q)
rank B
@ ... ... CA  k c:
@fc=@x0 (Q) : : : @fc =@xn (Q)
Thus Q 2 CX is singular and so X is not quasismooth.
(ii) is treated likewise. 

18.15 Examples Therefore a codimension 2 complete intersection Xd ;d in 1 2

P(a1 ; : : : ; an), which is quasismooth and not the intersection of a linear cone
with another hypersurface, satises:
(i) d2 > an and d1 > an 1.
(ii) if d1 < an then an j d2.
So this lemma gives extra combinatoric conditions to help remove nasty com-
plete intersections.
18.16 The nal list The program was run between the limits 0 and 32
and gave the following list (after cutting out repetitions and nasty complete
intersections):
Complete intersection KX3 pg Singularities
1. X2;2;2  P(1; 1; 1; 1; 1; 1; 1) 8 0
2. X2;2;4  P(1; 1; 1; 1; 1; 1; 1) 16 7
3. X2;2;6  P(1; 1; 1; 1; 1; 1; 3) 8 6
4. X2;3;3  P(1; 1; 1; 1; 1; 1; 1) 18 7
5. X3;3;3  P(1; 1; 1; 1; 1; 1; 2) 27=2 6 12 (1; 1; 1)
6. X3;3;4  P(1; 1; 1; 1; 1; 2; 2) 9 5 2  21 (1; 1; 1)
7. X3;4;4  P(1; 1; 1; 1; 2; 2; 2) 6 4 4  12 (1; 1; 1)
8. X4;4;4  P(1; 1; 1; 1; 2; 2; 3) 16=3 4 13 (1; 1; 1)
9. X4;4;4  P(1; 1; 1; 2; 2; 2; 2) 4 3 8  21 (1; 1; 1)
10. X4;4;5  P(1; 1; 1; 2; 2; 2; 3) 10=3 3 13 (1; 1; 1); 4  21 (1; 1; 1)
11. X4;4;6  P(1; 1; 1; 2; 2; 3; 3) 8=3 3 2  13 (1; 1 ; 1)
12. X4;4;6  P(1; 1; 2; 2; 2; 2; 3) 2 2 12  12 (1; 1; 1)
170 Working with weighted complete intersections

13. X4;5;6  P(1; 1; 2; 2; 2; 3; 3) 5=3 2  13 (1; 1; 1); 6  12 (1; 1; 1)

2
14. X4;6;6  P(1; 1; 2; 2; 3; 3; 3) 4=3 4  13 (1; 1; 1)
2
15. X4;6;6  P(1; 2; 2; 2; 2; 3; 3) 1 18  12 (1; 1; 1)
1
16. X5;6;6  P(1; 1; 2; 2; 3; 3; 4) 5=4 21 (1; 1; 1); 4  1 (1; 1; 1)
4 2
17. X5;6;6  P(1; 2; 2; 2; 3; 3; 3) 5=6 4  13 (1; 1; 1); 9  12 (1; 1; 1)
1
18. X6;6;10  P(2; 2; 2; 3; 3; 4; 5) 1=4 1 (1; 1; 1); 22  1 (1; 1; 1)
0
19. X6;6;6  P(1; 2; 2; 2; 3; 3; 4) 3=4 114 (1; 1; 1); 13  12 (1; 1; 1)
4 2
20. X6;6;6  P(1; 2; 2; 3; 3; 3; 3) 2=3 8  13 (1; 1; 1)
1
21. X6;6;6  P(2; 2; 2; 2; 3; 3; 3) 1=2 27  12 (1; 1; 1)
0
22. X6;6;7  P(1; 2; 2; 3; 3; 3; 4) 7=12 11 (1; 1; 1); 4  1 (1; 1; 1);
4 3
4  12 (1; 1; 1)
23. X6;6;8  P(1; 1; 2; 3; 3; 4; 5) 4=5 2 15 (1; 1; 2)
24. X6;6;8  P(1; 2; 2; 3; 3; 4; 4) 1=2 1 14 (1; 1; 1); 8  12 (1; 1; 1)
25. X6;6;8  P(2; 2; 2; 3; 3; 3; 4) 1=3 0 18  12 (1; 1; 1); 4  13 (1; 1; 1)
26. X6;7;8  P(1; 2; 2; 3; 3; 4; 5) 7=15 1 15 (1; 1; 2); 2  13 (1; 1; 1);
6  12 (1; 1; 1)
27. X6;8;10  P(1; 2; 3; 3; 4; 5; 5) 4=15 1 2  15 (1; 1; 2); 2  13 (1; 1; 1)
28. X6;8;10  P(2; 2; 3; 3; 4; 4; 5) 1=6 0 2  14 (1; 1; 1); 2  13 (1; 1; 1);
14  12 (1; 1; 1)
29. X6;8;9  P(1; 2; 3; 3; 4; 4; 5) 3=10 1 15 (1; 1; 2); 2  14 (1; 1; 1);
2  12 (1; 1; 1)
30. X8;10;12  P(2; 3; 4; 4; 5; 5; 6) 1=15 0 2  15 (1; 1; 1); 2  13 (1; 1; 1);
10  12 (1; 1; 1)
31. X8;9;10  P(2; 3; 3; 4; 4; 5; 5) 1=10 0 2  15 (1; 1; 2); 2  14 (1; 1; 1);
3  13 (1; 1; 1); 4  12 (1; 1; 1)
32. X9;10;12  P(2; 3; 3; 4; 5; 6; 7) 1=14 0 17 (1; 1; 2); 6  13 (1; 1; 1);
5  12 (1; 1; 1)
33. X10;11;12  P(2; 3; 4; 5; 5; 6; 7) 210
11 0 5  21 (1; 1; 1); 2  31 (1; 1; 1);
2  15 (1; 1; 2); 17 (1; 1; 3)
34. X10;12;14  P(2; 3; 4; 5; 6; 7; 8) 1=24 0 18 (1; 1; 3); 14 (1; 1; 1);
2  13 (1; 1; 1); 8  12 (1; 1; 1)
35. X10;12;18  P(3; 4; 5; 5; 6; 7; 9) 1052 0 71 (1; 1; 1); 2  15 (1; 1; 1);
4  13 (1; 1; 1)
36. X12;14;15  P(3; 4; 5; 6; 7; 7; 8) 1=56 0 12 (1; 1; 1); 41 (1; 1; 1);
2  17 (1; 1; 2); 18 (1; 1; 3)
37. X12;15;16  P(3; 4; 5; 6; 7; 8; 9) 1=63 0 2  12 (1; 1; 1); 3  31 (1; 1; 1);
1 (1; 1; 2); 1 (1; 1; 2)
7 9
38. X12;16;18  P(4; 5; 6; 6; 7; 8; 9) 1051 0 71 (1; 1; 1); 15 (1; 1; 1);
2  13 (1; 1; 1); 6  12 (1; 1; 1)
Table 7: Some codimension 3 complete intersections
 A. R. Iano-Fletcher 171

18.17 Note After renement there are no codimension 2 or 1 complete in-

tersections left in the list.
18.18 Extra example The family of intersections X2;2;2;2;2 in P8 is smooth,
KX3 = 16, pg = 9 and (OX ) = 8. If the above search were continued
this family would eventually appear; however my implementation becomes
painfully slow.
18.19 Conjecture (1) There are no canonical complete intersections with
codimension greater than 5.
(2) There are no Q -Fano complete intersections with codimension greater
than 3.
18.20 K3 surfaces Reid has done a similar search to produce lists of K3
surface weighted complete intersections; using Riemann{Roch for OS (1) (see
[R4], Theorem 9.1).
This time the search is nite due to the following theorem pointed out by
Reid:
18.21 Theorem Let S be a K3 surface with canonicalP (Du Val) singularities
of types Ani , Dni or Eni for i = 1; : : : ; n. So ni  19. This limits the
singularities present on the K3 surface to a nite list.
Proof Let f : T ! S be a minimal resolution. T is still a K3 surface.
By [BPV], Proposition VIII.3.3, h1;1 = h1 (
1T ) = 20. By the Signature
Theorem [BPV], Theorem IV.2.13, the cup product restricted to H2(T; R ) is
nondegenerate of type (1; h1;1 1) = (1; 19). Via the Neron{Severi group, the
exceptional 2-curves of the resolution f are linearly independent in H1;1 ,
each with negative self-intersection.
It is well known that a Du ValP singularity of type An, Dn or En contributes
exactly n 2-curves to T . Thus ni  19. 

References
[ACGH] E. Arbarello, M. Cornalba, P. A. Griths, J. Harris: Geometry of
algebraic curves, Vol. 1, Comprehensive studies in Math, 267, (1985)
Springer Verlag.
[AM] M. F. Atiyah, I. G. MacDonald: Introduction to commutative algebra,
Addison Wesley Publishing Co. 1969.
172 Working with weighted complete intersections

[BPV] W. Barth, C. Peters, A. Van de Ven: Compact complex surfaces,

(1984) Springer Verlag.
[Be] D. N. Bernshtein: The number of roots of a system of equations, Funk.
Anal 9 no. 3 (1975) pp. 1{4.
[Da] V. I. Danilov: The geometry of toric varieties, Uspekhi Mat. Nauk.
33 no. 2 (1978) pp. 85{134, English transl. Russian Math Survey 33
no. 2 (1978) pp. 97{154.
[De] C. Delorme: Espaces projectifs anisotropes, Bull. Soc. math. France,
103 (1975) pp. 203{223.
[Di] A. Dimca: Singularities and coverings of weighted complete intersec-
tions, Journal reine u. angew. Math. 366 (1986) pp. 184{193.
[Du] A. Durfee: Fifteen characterizations of rational double points and sim-
ple critical points, Ens. math 25 no. 2 (1979) pp. 131{163.
[EGA] A. Grothendieck: Elements de geometrie algebrique Chap. II. Publ.
Math. de l'IHES 8 (1961).
[F1] A. R. Fletcher: Contributions to Riemann{Roch on projective 3-folds
with only canonical singularities and applications, Proc. Symposia in
Pure Math 46 (1987) Vol 1, pp. 221{231.
[F2] A. R. Fletcher: Plurigenera of 3-folds and weighted hypersurfaces,
Thesis submitted to the University of Warwick for the degree of Ph.D.
(1988).
[Hart] R. Hartshorne: Algebraic geometry, Grad. Texts in Mathematics 52,
Springer Verlag 1977.
[Ku] A. G. Kushnirenko: Newton polytopes and the Bezout's theorem, Funk.
Anal. 10 no. 3 (1976) pp. 82{83
[M] D. Mumford: Curves and their Jacobians, University of Michigan
Press, 1975.
[OW] P. Orlick, P. Wagreich: Equivariant resolution of singularities with
C  -actions, Proc. Second Conference on Compact Transformation
Groups, Part I, LNM 298 (1971), Springer Verlag.
[R1] M. Reid: Canonical 3-folds, Proc. Alg. Geom. Anger 1979. Sijtho
and Nordho, pp. 273{310.
 A. R. Iano-Fletcher 173

[R2] M. Reid: Minimal models of canonical 3-folds, Algebraic varieties and

analytic varieties, Adv. Stud. Pure Math., North Holland, Amsterdam
and Kinokuniya Book Co., Tokyo, Math vol. 1, 1983 pp. 131{180.
[R3] M. Reid: Tendencious survey of 3-folds, Proc. Symposia in Pure Math
46 (1987) Vol 1, pp. 333{344.
[R4] M. Reid: Young person's guide to canonical singularities, Proc. Sym-
posia in Pure Math 46 (1987) Vol 1, pp. 345{414.
[S] J. H. M. Steenbrink: Mixed Hodge structures and singularities: a
survey, Geometrie algebrique et applications, III (La Rabida, 1984),
Travaux en Cours, 24, Hermann, Paris, 1987, pp. 99{123
[WPS] I. Dolgachev: Weighted projective spaces, Group actions and vector
elds, Proc. Vancouver 1981, LNM 956, pp. 34{71 Springer Verlag.

Anthony Iano-Fletcher,
Computational Bioscience and Engineering Laboratory,
Centre for Information Technology, Building 12A, Room 2033,
National Institutes of Health, 9000 Rockville Pike,
Bethesda, MD 20892-5624, USA.
e-mail: [email protected]
web: cbel.cit.nih.gov/arif/cv.html