Professional Documents
Culture Documents
Genival Da Silva JR.: Q P P, P 2p
Genival Da Silva JR.: Q P P, P 2p
VARIETIES
1. Introduction
The Hodge conjecture is major open problem in Complex Algebraic Geometry
that has been puzzling mathematician for decades now. The modern statement
is the following: Let X be smooth complex projective variety, then the (rational)
cycle class map is surjective:
cl⊗Q : CH p (X) ⊗ Q → H p,p ∩ H 2p (X, Q)
P P
where cl⊗Q ( ai Xi ) = ai [Xi ], ai ∈ Q and [Xi ] is the class of the subvariety Xi .
The case p = 1 is the only case that it is known to hold in general, which follows
from Lefschetz’s theorem on (1, 1)-classes. Special cases have emerged during the
years but all of them were specific for certain classes of varieties. For example,
Abelian varieties of prime dimension, unirational and uniruled fourfolds, hypersur-
faces of degree less than 6, and some others [3].
Using hard Lefschetz theorem, Lefschetz hyperplane theorem and some Hilbert
scheme arguments, we can reduce the Hodge conjecture to the case of an even
dimensional (> 2) variety and primitive middle cohomology classes.
Shioda [5] gave an interesting characterization of the Hodge conjecture for Fermat
varieties, which we now review.
2. Shioda’s work
n
Let Xm ∈ Pn+1 denote the Fermat variety of dimension n and degree m, i.e. the
solution to the equation:
xm m m
0 + x1 + . . . + xn+1 = 0
and µm the group of m-th roots of unity. Let Gnm be quotient of the group
n+2
z }| {
µm × . . . × µm by the subgroup of diagonal elements.
The group Gnm acts naturally on Xmn
by coordinatewise multiplication, moreover,
ˆn n
the character group Gm of Gm can be identified with the group:
Gˆnm = {(a0 , . . . , an+1 )|ai ∈ Zm , a0 + . . . + an+1 = 0}
a
via (ζ0 , . . . , ζn+1 ) 7→ ζ0a0 . . . ζn+1
n+1
, where (ζ0 , . . . , ζn+1 ) ∈ Gnm .
By the previous section, in order to prove the Hodge conjecture, it’s enough
to prove it for primitive classes, therefore in this paper we will focus on primitive
1
2 GENIVAL DA SILVA JR.
cohomology. The action of Gnm extends to the primitive cohomology and makes
i
Hprim n
(Xm i
, Q) and Hprim n
(Xm , C) a Gnm -module. For α ∈ Gˆnm , we set:
n
V (α) = {ξ ∈ Hprim n
(Xm , C)|g ∗ (ξ) = α(g)ξ for all g ∈ Gnm }
Before stating the characterization of Hodge classes we need a few notation. Let
Unm := {α = (a0 , . . . , an+1 ) ∈ Gˆnm |ai 6= 0 for all i}
For α ∈ Unm we set |α| = i <ami > , where < ai > is the representative of ai ∈ Zm
P
between 1 and m − 1. Suppose n = 2p, then we set
Bnm := {α ∈ Unm ||tα| = p + 1 for all t ∈ Z∗m }
2p
Theorem 2.1. [4, 5] Let Hdg p := H p,p ∩ Hprim (X, Q) be the group of primitive
Hodge cycles. Then:
(a) dim V (α)L= 0 or 1, and V (α) 6= 0 ⇐⇒ α ∈ Unm
p
(b) Hdg = α∈Bnm V (α)
n
Now let C(Xm ) denote the subspace of Hdg p which are classes of algebraic
n
cycles. Then C(Xm ) is a Gnm -submodule and by the theorem above there is a
n n
subset Cm ⊂ Bm such that:
M
n
C(Xm )= V (α)
α∈Cn
m
β 0 ∗ γ 0 = (b0 , . . . , br , c0 , . . . , cs ) ∈ Ur+s
m
n
By the above corollary, the Hodge conjecture can be proven for the Fermat Xm
n
if the following conditions are true for every α ∈ Bm :
(P1) α ∼ β 0 ∗ γ 0 for some (β 0 , γ 0 ) ∈ Br−1 s−1
m × Bm , (r, s odd).
(P2) α ∼ β#γ for some (β, γ) ∈ (Bm × Bm ) ∩ Ur,s
r s
m , (r, s even and positive).
where ∼ means equality up to permutation between factors.
In order to make these conditions more explicit, we introduce the additive semi-
group Mm of non-negative solutions (x1 , . . . , xm−1 ; y) with y > 0, to the following
system of linear equations:
m−1
X
< ti > xi = my for all t ∈ Z∗m
i=1
Also, define Mm (y) as those solutions where y is fixed. Note that by Gordan’s
lemma, Mm is finitely generated.
Definition 2.4. An element a ∈ Mm is called decomposable if a = c + d for
some c, d ∈ Mm , otherwise it’s called indecomposable. An element is called
quasi-decomposable if a + b = c + d for some a ∈ Mm (1) and c, d ∈ Mm with
c, d 6= a.
With this notation we can identify elements of Bnm with elements of Mm using
the map:
n n
{} : α = (a0 , . . . , an+1 ) ∈ Bnm 7→ {α} = (x1 (α), . . . , xm−1 (α), + 1) ∈ Mm ( + 1)
2 2
where xk (α) is the number os i’s such that < ai >= k.
Note that α satisfies (P 1) above if and only if {α} is decomposable. If α satisfies
(P 2) then {α} is quasi-decomposable. Conversely, if the latter is true then α
satisfies (P 1) or (P 2). So it makes sense to introduce the following conditions:
(Pmn
) Every indecomposable elemets of Mm (y) with 3 ≤ y ≥ n2 + 1, if any, is
quasi-decomposable.
(Pm ) Every indecomposable elemets of Mm (y) with y ≥ 3 is quasi-decomposable.
By the results above we conclude:
Theorem 2.5. [5] If condition (Pm ) is satisfied, then the Hodge conjecture is true
n n n
for Xm for any n. If (Pm ) is satisfied then the Hodge conjecture is true for Xm .
For m prime or m = 4, Mm is generated by Mm (1) which gives:
n
Theorem 2.6. [4, 5] If m is prime or m = 4, the Hodge conjecture is true for Xm
for all n.
Shioda manually verified condition (Pm ) for m ≤ 20 and concluded:
n
Theorem 2.7. [5] If m ≤ 20, the Hodge conjecture is true for Xm for all n.
4 GENIVAL DA SILVA JR.
Based on the values above and the ones already computed, we believe the following
is true:
pk−1 +1
Conjecture 2. For p > 2 prime, we have φ(pk ) = 2 , and φ(2l ) = 2l−2 + 1
for l > 2.
Computing φ(m) for m < 48 gives the following:
17
m+3
13 y= 3
φ(m)
9
7
5
3
1
5 10 15 20 25 30 35 40 45
m
Question 2. It seems that φ(m) ≤ f (m) for some linear function f . Can f be
described explicitly?
For m ≥ 48, computations become more and more time consuming, even for the
computer, and specially if m has a lot of prime powers in its prime decomposition.
But the results obtained here give us a glimpse about the structure of Mm and
consequently, the Hodge conjecture.
References
1. N. Aoki, Some new algebraic cycles on Fermat varieties. J. Math. Soc. Japan 39 (1987), no.
3, 385–396.
2. G. da Silva Jr, https://sites.google.com/view/genival-da-silva-jr/in%C3%ADcio (Ac-
cessed December 25, 2020).
3. J. Lewis, A survey of the Hodge Conjecture, CRM Monograph, Ser. 10, AMS, Providence, RI,
1999.
4. Z. Ran, Cycles on Fermat hypersurfaces. Compositio Mathematica, Tome 42 (1980) no. 1, pp.
121-142.
5. T. Shioda,The Hodge conjecture for Fermat varieties. Math. Ann. 245, 175–184 (1979).