This paper investigates the integral Hasse principle on twisted Markoff surfaces using Brauer-Manin obstruction, providing explicit examples with 4-torsion Brauer group elements and detailed local invariant computations.
Contribution
It introduces new examples of twisted Markoff surfaces with nontrivial Brauer groups and explicit representatives, advancing understanding of the Brauer-Manin obstruction in this context.
Findings
01
Identified examples with 4-torsion elements in the Brauer group
02
Constructed explicit representatives for these Brauer group elements
03
Computed local invariants at special places for these examples
Abstract
We study the integral Hasse principle for affine varieties of the form ax^2+y^2+z^2-xyz=m ,using Brauer-Manin obstruction, and we produce examples whose Brauer groups include 4-torsion elements .We will construct their explicit representatives and compute local invariants in special places.
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAlgebraic Geometry and Number Theory · Advanced Algebra and Geometry · Advanced Combinatorial Mathematics
We study the integral Hasse principle for affine varieties of the form
[TABLE]
using Brauer-Manin obstruction, and we produce examples whose Brauer groups include 4-torsion elements .
We use methods of [5] to describe them and in some cases we show that there is no
Brauer-Manin obstruction to the integral Hasse principle for them.
1. Introduction
In recent papers [4] and [10], Colliot-Thélène , Wei, Xu, D. Loughran and V. Mitankin, studied the integral Hasse principle
and strong approximation for Markoff surfaces, using the Brauer-Manin obstruction.
For Markoff surfaces , D. Loughran and V. Mitankin obtained the following beautiful result :
Assume that m∈Z is such that affine surface Um defined by
[TABLE]
has a Brauer-Manin obstruction to the integral Hasse principle. Then
[TABLE]
As they pointed out , this can be seen as an analogue of the finiteness of exceptional spinor classes in the study of the representation of an integer by a
ternary quadratic form (see [3 ,§7]).
Now, fix m,a∈Z,m=0,4a. Let U⊂AZ3
be the affine scheme over Z defined by the equation
[TABLE]
We study the Brauer-Manin obstruction to the integral Hasse principle for U.
In particular, we have similar results :
Theorem 1.1**.**
Assume that [Q(a,m,m−4a):Q]=8, let (a,m)=p1n1p2n2⋯psns, where (a,m) is the greatest common divisor of a and m, pi are prime for 1≤i≤s .
If there is a Brauer-Manin obstruction to the integral Hasse principle for U, we have
[TABLE]
Moreover, we will give examples whose Brauer groups include 4-torsion elements,
and with some assumptions, we can show that there is no Brauer-Manin obstruction to the integral Hasse principle for them.
As noted in [11], an often used strategy for proving that a class A∈Br(U) of order n gives no obstruction to the Hasse principle is to demonstrate the
existence of a finite place v of k such that the evaluation map X(kv)→(Brkv)[n], sending a
point P∈Y(kv) to the evaluation A(P)∈(Brkv)[n], is surjective .However,the local invariant of non-cyclic algebra is difficult to compute in general. Based on ideas of [5],
we will construct explicit representatives of non-cyclic Brauer classes on affine surfaces, and compute its local invariants
in special places.
Notation Let k be a field and k a separable closure of k.
If X is a k-variety,we write X=X×kk.
If X is an integral k-variety, we let k(X) denote the function field of X.
If X is a geometrically integral k-variety, we let k(X) denote
the function field of X.
We let Pic(X)=Heˊt1(X,Gm) denote the Picard group of a scheme X. We let
Br(X)=Heˊt2(X,Gm) denote the Brauer group of a scheme X.
If X is a regular integral k-variety, the natural map
[TABLE]
is injective. We let
[TABLE]
denote the algebraic Brauer group of a k-variety.
2. Algebraic Brauer group of cubic surface
Follow J.-L.Colliot-Thélène,Dasheng Wei,and Fei Xu, we have
Lemma 2.1**.**
Let X⊂Pk3 defined by equation
[TABLE]
over a field k of characteristic zero with a∈k×,a∈/k2, then X is smooth if and only if m=0,4a.In this case,the 27 lines in X are defined over k(a,m,m−4a) by the following equations:
[TABLE]
and
[TABLE]
*with ε=±1, and δ=±1.Moreover the intersection number
[TABLE]
Proof.
The results follow from straight forward computation.
∎
.
Proposition 2.2**.**
Let X⊂Pk3 defined by equation
[TABLE]
over a field k of characteristic zero with a∈k×,a∈/k2,and m=0,4a.Then Br(X)/Br0(X)=0 or Br(X)/Br0(X)≅Z/2 with generator ((tx)2−4,m−4a).
Proof.
The proof is completely similar to [4,Proposition3.2] , for later application,we only give computations of Br(X)/Br0(X) in somes cases.
Since X is geometrically rational,we have Br(X)=Br1(X). SinceX(k)=∅, we have the following isomorphism
[TABLE]
by the Hochschild-Serre spectral sequence. By [7,Chapter V,Proposition4.8],there is l∈Pic(X) satisfying the following intersecton property
(l,l)=1(l,li(1,1))=0 for 1≤i≤6.
such that {Hi(1,1):1≤i≤6}∪{l} forms a basis of Pic(X).
Since
[TABLE]
where 1≤j≤3,1≤i≤6. One concludes that
[TABLE]
in Pic(X) for 1≤j≤3 by [7,Chapter V,Proposition 4.8(e)].
For simplicity, we write li for li(1,1) with 1≤i≤6 . If [k(a,m,m−4a):k]=2, there exists σ∈Gal(k(a)/k) such that σ(a)=−a .
1.m∈k,m−4a∈k, we have
[TABLE]
Since Ker(1+σ)=(l−l2−L3−l4), Im(σ−1)=(l−l2−L3−l4),
we have H1(k,Pic(X))=0 .
Arguing in the same way as in the proof of [4,Theorem 3.4], one obtains the generators (x−2,m−4a) and (x+2,m−4a).
Indeed since
[TABLE]
is a closed subset of codimension ≥2 on U, one obtains that (x±2,m−4a)∈Br1(U).
This implies that
[TABLE]
Now we show that B is not constant.
[TABLE]
The generic fibre Uηπηη induces
[TABLE]
by [6,Theorem 5.4.1].
Since [k(a,m,m−4a):k]=8, the residue of (x2−4,m−4a) at (m−ax2) is different from that of (x2−4,m−ax2).
This implies that πη∗(x2−4,m−4a) is not constant by the Faddeev exact sequence. Since πη∗(x2−4,m−4a) is the pull-back of B by the projection map Uη→U, one concludes that B is not constant.
∎
3. Examples of Brauer-Manin obstruction
We now give examples of Brauer-Manin obstruction to the integral Hasse principle. Here the results are inspired by the results in [10,§5.3,§5.4]
Lemma 3.1**.**
If p is an odd prime with (p,m−4a)=1, then the following elements
[TABLE]
vanish over U(Zp). If m−4a>0,these elements vanish over U(R).
In particular, if a<0, (x2−4,m−4a)∞=(z2−4a,m−4a)∞=(y2−4a,m−4a)∞=0.
Proof.
Arguing in the same way as in the proof of [4,Lemma 5.1 ], one can easily verify this.
∎
Lemma 3.2**.**
Let p∣(m−4a) be odd , if p∤a, any singular point T(x,y,z)∈U(Fp) satisfies
[TABLE]
Proof.
Since T∈U(Fp) is singular, we have
[TABLE]
We obtain x2=4,y2=4a.
∎
Lemma 3.3**.**
*Let p≥3 such that p∣(m−4a) and p∤a, with ordp(m−4a) even but m−4a∈/Qp×2.
Let B1=(x2−4,m−4a),B2=(x+2,m−4a). For all T∈U(Zp), we have
If (pa)=−1,*
[TABLE]
If (pa)=1,
[TABLE]
Proof.
Note that ax2+y2+z2−xyz=m is equivalent to
[TABLE]
As (m−4a)∈/Qp×2 and ordp(m−4a) is even, it follows that ordp((x2−4)(y2−4a)) is even.
If ordp(x2−4) is even ,one obtains ordp(x+2) is even, hence {invpB1(T),invpB2(T)}={{0,0}}.
Assume that ordp(x2−4) is odd, then ordp(y2−4a) is odd, thus y2≡4amodp. If (pa)=−1, this is a contradiction.
Now let (pa)=1, if ordp(x+2) is odd, we obtain
[TABLE]
If ordp(x−2) is odd, we obtain
[TABLE]
We now show that these possibilities can be realised. We first consider (pa)=−1,
we have smooth point (2,0,0)∈U(Fp) by lemma 3.2, hence U(Zp)=∅.
Now we consider (pa)=1, we have smooth point (0,2a,0)∈U(Fp) by lemma 3.2, hence there exists
T∈U(Zp) such that
[TABLE]
Suppose ordp(m−4a)=2, let s∈Fp× such that (pp2m−4a−s)=1. Let s′∈Zp such that
s′≡smodp . There is y0∈Zp such that y02−4a=ps′ by Hensel lemma. Let x0=p−2, we consider the following
equation
[TABLE]
over Zp.
That is
[TABLE]
By Hensel lemma, one can see that the equation has solutions. Let t0 denote one of the solutions and let z0∈Zp such that
2z0−x0y0=pt0, then T0=(x0,y0,z0)∈U(Zp), we have
[TABLE]
If we let x1=p+2, one can see that there exists T1=(x1,y1,z1)∈U(Zp), we have
[TABLE]
For ordp(m−4a)>2, the proof is similar .
∎
Proposition 3.4**.**
Suppose conditions of Proposition 2.3 are satisfied, if there exists p≥5 such that p∤a and ordp(m−4a) is odd,
there is no Brauer–Manin obstruction to integral Hasse principle.
Proof.
We can assume that U(AZ)=∅, where AZ=R×∏pZp ,otherwise there is nothing to prove.
Let
[TABLE]
to prove the proposition ,
it suffices to show for all (ε1,ε2)∈(Z/2Z)2, there exists
ζ∈U(Zp) such that
[TABLE]
We first consider the case (pa)=1.
Since p≥5,there exist t,s∈Fp, such that the Legendre symbol (pt2−4a)=1, (ps2−4a)=−1, let
[TABLE]
One can see that υ1 and υ2 are smooth points of U(Fp) by lemma 3.2, hence there exist μ1,μ2∈U(Zp)
such that μi≡υimodp for 1≤i≤2, we obtain
[TABLE]
Since p≥5 ,we can choose e,f∈Fp such that the Legendre symbol (pe)=−1,(pe2−4e)=1,
(pf)=−1, (pf2−4f)=−1. Let
[TABLE]
One can see that υ3 and υ4 are smooth points of U(Fp) by lemma 3.2, hence there exist μ3,μ4∈U(Zp)
such that μi≡υimodp for 3≤i≤4, we obtain
[TABLE]
Now assuing (pa)=−1.
Let η1=μ1,η2=μ2, we have
[TABLE]
Let
[TABLE]
where α2=ae,β2=af,one can see that ξ3 and ξ4 are smooth points of U(Fp) by lemma 3.2, hence there exist η3,η4∈U(Zp)
such that ηi≡ξimodp for 3≤i≤4, we obtain
[TABLE]
The proposition is established.
∎
As corollary, we obtain Theorem 1.1 .
Lemma 3.5**.**
Given a,m∈Z, the equation ax2+y2+z2−xyz=m
has solutions in (Zp)3 for all primes p except for the following two cases:
(i)
a≡1mod4,m≡3mod4.
2. (ii)
a≡1mod3,m≡±3mod9.
Proof.
We break up the proof into several cases. Let
[TABLE]
For p>3,we have
(i)
p∤a,p∣m. If p≡1mod4, there exists (0,t,s)∈(Fp)3, such that f(0,t,s)≡0modp,fy(0,t,s)≡0modp, hence f has a zero in (Zp)3.
If p≡3mod4,we first consider(pa)=−1,there exists (1,0,t)∈(Fp)3,such that f(1,0,t)≡0modp,fz(1,0,t)≡0modp, hence f has a zero in (Zp)3. Now suppose (pa)=1,
there exists (3,t,2t)∈(Fp)3,such that f(3,t,2t)≡0modp,fy(3,t,2t)≡0modp,
hence f has a zero in (Zp)3.
2. (ii)
p∤a,p∤m. If (pm)=−1, since p>3, there exists (0,t,s)∈(Fp)3,such that f(0,t,s)≡0modp,fy(0,t,s)≡0modp, hence f has a zero in (Zp)3.
If (pm)=1,there exists (0,t,0)∈(Fp)3,such that f(0,t,0)≡0modp,fy(0,t,0)≡0modp,
hence f has a zero in (Zp)3.
3. (iii)
p∣a,p∣m. There exists (2,1,1)∈(Fp)3, such that f(2,1,1)≡0modp,fx(2,1,1)≡0modp, hence f has a zero in (Zp)3.
4. (iv)
p∣a,p∤m. An argument similar to (ii), one can prove f has a zero in (Zp)3.
For p=3, we have
(i)
3∤a,3∣m. If a≡1mod3, f has only zero (0,0,0)∈(F3)3. So f has no zeros in (Z3)3
when m≡±3mod9.
Now assume 9∣m,one can easily check the equation ax2+y2+z2−3xyz=9m has a solution (x0,y0,z0)∈(Z3)3 using Hensel lemma.
Hence f(3x0,3y0,3z0)=0.
If a≡2mod3, since f(1,1,0)≡0mod3,fy(1,1,0)≡0mod3, f has a zero in (Z3)3.
2. (ii)
3∤a,3∤m. If m≡1mod3, since f(0,1,0)≡0mod3,fy(0,1,0)≡0mod3,
f has a zero in (Z3)3.
If m≡2mod3, since f(0,1,1)≡0mod3,fy(0,1,1)≡0mod3,f has a zero in (Z3)3.
3. (iii)
3∣a,3∣m. Since f(2,1,1)≡0mod3,fx(2,1,1)≡0mod3,f has a zero in (Z3)3.
4. (iv)
3∣a,3∤m. If m≡1mod3, since f(0,1,0)≡0mod3,fy(0,1,0)≡0mod3,
f has a zero in (Z3)3.
If m≡2mod3, since f(0,1,1)≡0mod3,fy(0,1,1)≡0mod3, f has a zero in (Z3)3.
For p=2, we have
(i)
2∤a,2∣m. Since f(1,1,1)≡0mod2,fy(1,1,1)≡0mod2, f has a zero in (Z2)3.
2. (ii)
2∤a,2∤m. If a≡mmod8, since f(1,0,0)≡0mod8,ord2(fx(1,0,0))=1, f has a zero in (Z2)3.
If a≡m−4mod8, since f(1,2,0)≡0mod8,ord2(fx(1,2,0))=1, f has a zero in (Z2)3.
If a≡3mod4,m≡1mod4, we first consider m≡1mod8, since f(0,1,0)≡0mod8,ord2(fy(0,1,0))=1,
f has a zero in (Z2)3. Now suppose m≡5mod8, since f(0,1,2)≡0mod8,ord2(fy(0,1,2))=1,
f has a zero in (Z2)3.
If a≡1mod4,m≡3mod4, note that all solutions of f≡0mod2 are (1,0,0),(0,1,0) and (0,0,1).
If we take for (1,0,0), this implies a≡mmod4,a contradiction. If we take for (0,1,0) or (0,0,1), this implies m≡1mod4, a contradiction. So f has no zero in (Z2)3 in this case.
3. (iii)
2∣a,2∣m. Since f(0,1,1)≡0mod2,fx(1,1,1)≡0mod2, f has a zero in (Z2)3.
4. (iv)
2∣a,2∤m. Since f(1,1,1)≡0mod2,fx(1,1,1)≡0mod2, f has a zero in (Z2)3.
∎
Proposition 3.6**.**
Let U be the scheme over Z given by
[TABLE]
where a,d are odd integers such that (a,d)=1, 3∤(a−1), a∈/Q,
p≡±1mod8 or (pa)=−1 for p∣d.
Then there is a Brauer-Manin obstruction to the integral Hasse principle for U.
Proof.
By lemma 3.5, we have U(AZ)=∅ .
Let
[TABLE]
we will show that for each point T∈U(Zp), we have
[TABLE]
If p∤2d2 the claim follows from Lemma 3.1 . If p∣d and p≡±1mod8 ,we have 2∈Qp×2.
Thus 2d2∈Qp×2.If p∣d and (pa)=−1, the claim follows from lemma 3.3. Finally, since m−4a>0,
the claim is trivial for p=∞ . It remains to examine p=2.
Assme now p=2. Let T∈U(Z2), one easily see that there is at least one coordinate of T belonging to Z2×.
A simple Hilbert symbol calculation implies the claim for p=2.
∎
Proposition 3.7**.**
Let U be the scheme over Z given by
[TABLE]
where a is an even integer such that a≡1mod3 and a∈/Q,
p≡±1mod12 or (pa)=−1 for p∣d.
When 4a+3d2∈/Q, there is a Brauer-Manin obstruction to the integral Hasse principle for U.
Proof.
One can see U(AZ)=∅ by lemma 3.5.
Let B=(x2−4,3)=(z2−4a,3)=(y2−4a,3), we will show that for each point T∈U(Zp), we have
[TABLE]
so that B gives an obstruction to the Hasse principle.
If p∤6d2 the claim follows from Lemma 3.1 . If p∣d and p≡±1mod12 ,we have 3∈Qp×2.
Thus 3d2∈Qp×2.If p∣d and (pa)=−1, the claim follows from lemma 3.3. Finally, since m−4a>0,
the claim is trivial for p=∞ . It remains to examine p=2,3.
Assme now p=2. Let T∈U(Z2), one easily see that there is at least one coordinate of T belonging to Z2×.
A simple Hilbert symbol calculation implies the claim for p=2.
For p=3, note that ax2+y2+z2−xyz=4a+3d2 is equivalent to the following equations
[TABLE]
Then for any P∈U(Z3), there are two coordinates of P belonging to 3Z3. We can assume x,y∈3Z3,
since (x2−4,3)3=(y2−4a,3)3=21, one concludes that inv3B(P)=21.
The proposition is established.
∎
Proposition 3.8**.**
Let U be the scheme over Z given by
[TABLE]
where 4∣a,a≡1mod3 and a∈/Q,
d∈Z whose prime divisors are congruent to ±1mod12 and ±1mod8 or ±5mod12 and ±3mod8.
Then there is a Brauer-Manin obstruction to the integral Hasse principle for U.
Proof.
Note that if U(Z2)=∅, since 4∣a,for any local solution T(x,y,z)∈U(Z2), y or z is in Z2×.
We assume z is in Z2×,hence z2−4a≡1mod8. Thus z2−4a∈Q2×2. Let B=(x2−4,6)=(z2−4a,6)=(y2−4a,6),
we obtain inv2B(T)=0.
By lemma 3.5, we have U(AZ)=∅. A similar argument in the proof of Proposition 3.6, we obtain
[TABLE]
so that B gives an obstruction to the Hasse principle.
∎
Proposition 3.9**.**
Let U be the scheme over Z given by
[TABLE]
where a,d are odd integers such that (a,d)=1, ord5(a)≥2 ,and a∈/Q,
the prime divisors of d are congruent to ±1mod8 and ±1mod5 or ±3mod8 and ±2mod5.
Then there is a Brauer-Manin obstruction to the integral Hasse principle for U.
Proof.
By lemma 3.5, one can prove U(AZ)=∅. Let B=(x2−4,10)=(z2−4a,10)=(y2−4a,10),
we will show that for each point T∈U(Zp), we have
[TABLE]
so that B gives an obstruction to the Hasse principle.
If p∤10d2 the claim follows from Lemma 3.1 . If p∣d, then 10∈Qp×2.
Thus 10d2∈Qp×2. Finally, since m−4a>0, the claim is trivial for p=∞ . It remains to examine p=2,5.
Assme now p=5. Since 25∣a,for any local solution T(x,y,z)∈U(Z5), y or z is in Z5×.
We assume z is in Z5×,hence z2−4a∈Q5×2.
we obtain inv5B(T)=0.
For p=2,for any local solution T(x,y,z)∈U(Z2), there is at least one coordinate of T belonging to Z2×.
We assume z is in Z2×,hence z2−4a≡5mod8.
we obtain inv2B(T)=1/2.
so that B gives a obstruction to the Hasse principle.
∎
Remark 3.10**.**
We can take m=4a+2qd2, where q is an odd prime , one easily obtains similar conclusions.
Proposition 3.11**.**
Let U be the scheme over Z given by
[TABLE]
where q is an odd prime, t is an odd integer such that 3∤(t−1) ,t∈/Q,
(t,d)=1 and the prime divisors of d are congruent to ±1mod8.
Then there is a Brauer-Manin obstruction to the integral Hasse principle for U.
Proof.
By lemma 3.5, one can prove U(AZ)=∅.Let B=(x2−4,2)=(z2−4a,2)=(y2−4a,2),
we will show that for each point T∈U(Zp), we have
[TABLE]
so that B gives an obstruction to the Hasse principle.
If p∤2q2d2 the claim follows from lemma 3.1 . If p∣d, then 2∈Qp×2.
Thus 2q2d2∈Qp×2. Finally, since m−4a>0, the claim is trivial for p=∞ .
Note that for any local solution T(x,y,z)∈U(Z2), there is at least one coordinate of T belonging to Z2×.
we obtain inv2B(T)=1/2. It remains to examine p=q.
If y or z is in Zq×, we have y2−4a∈Qq×2 or z2−4a∈Qq×2. If not,
for any point M(x,y,z)∈U(Zq),let y=qy′,z=qz′, then (x,y′,z′) is the solution of the following equation
[TABLE]
We obtain invqB(M)=0 by lemma 3.1 .
∎
Proposition 3.12**.**
Let U be the scheme over Z given by
[TABLE]
where q is an odd prime such that q≡±3mod8 ,
then there is a Brauer-Manin obstruction to the integral Hasse principle for U.
Proof.
One can easily check U(AZ)=∅ by lemma 3.5. Let B=(x2−4,2q)=(z2+4q,2q)=(y2+4q,2q),
we will show that for each point T∈U(Zp), we have
[TABLE]
We only need to consider p=2,q,∞.
Note that for any local solution T(x,y,z)∈U(Z2), there is at least one coordinate of T belonging to Z2×.
we obtain inv2B(T)=1/2. Since y2+4q>0, the claim is trivial for p=∞ . It remains to examine p=q.
If y or z is in Zq×, we have y2+4q∈Qq×2 or z2+4q∈Qq×2.If not,
for any point M(x,y,z)∈U(Zq),let y=qy′,z=qz′, then (x,y′,z′) is the solution of the following equation
[TABLE]
Thus x2≡2modq, a contradiction. We obtain invqB(M)=0.
∎
4. Review of bicyclic group cohomolgy
Let G=Z/n⊕Z/m, with generators t and s.Put Nt:=1+t+⋯+tn−1 and Δt:=1−t in Z[G],
similar put Ns:=1+s+⋯+sm−1 and Δs:=1−s in Z[G].
For trivial G-module Z ,we have the following resolution
[TABLE]
where
[TABLE]
If we are given a G-module M,then applying HomG(−,M) to the above complex,the
groups Hi(G,M) are homology groups of the following complex:
[TABLE]
where
[TABLE]
We introduce the notations: Z1(G,M):=ker(d1),and Z2(G,M):=ker(d2),then we have
[TABLE]
For subgroup ⟨t⟩, we have the following resolution
[TABLE]
The injection from Z[t] to the first factor Z[G] of Z[G]i+1 induces
the restriction
Hi(G,M)→Hi(⟨t⟩,M)
(a0,...ai)→a0
Similar for subgroup ⟨s⟩, the injection from Z[s] to the last factor Z[G] of Z[G]i+1 induces
the restriction
Hi(G,M)→Hi(⟨s⟩,M)
(a0,...ai)→ai
5. Special examples
Example 1**.**
Let U be an affine variety over a field of characteristic zero defined by the equation
[TABLE]
where a∈k×,a∈/k2,m=0,4a.
By[4,Proposition 2.2],Pic(U) is given by the following quotient group
[TABLE]
Here we give explicit condition which H1(k,Pic(U))≅Z/2⊕Z/4, and use methods of Colliot-Thélène, D. Kanevsky, J.-J. Sansuc [5]
to describe the 4 torsion elements .
Lemma 5.1**.**
When [k(a,m,m−4a):k]=4 and mam−4a∈k, H1(k,Pic(U))≅Z/2⊕Z/4
Proof.
Let G=Gal(k(a,m)/k),there exist σ,τ∈G,such that
[TABLE]
By computation of proposition 2.3, we have
(i)
\left\{\begin{array}[]{l}\sigma(\overline{l}_{1})=-\overline{l_{1}}\\
\sigma(\overline{l}_{2})=\overline{l}_{3}-\overline{l}_{1}\\
\sigma(\overline{l}_{3})=\overline{l}_{2}-\overline{l}_{1}\\
\sigma(\overline{l}_{4})=\overline{l}_{2}+\overline{l}_{3}-\overline{l}_{1}-\overline{l}_{4}\\
\end{array}\right.
2. (ii)
we have H1(⟨τ⟩,Pic(U))≅Z/2.
By computation,we have Pic(U)⟨τ⟩=(l1,l2−l3), and since
[TABLE]
one concludes that
[TABLE]
Hence,we have the following sequence
[TABLE]
by [6.proposition 3.3.14]. To show H1(G,Pic(U)) has 4-torsion elements, we use bicyclic group cohomology .
Now we identify classes in H1(G,Pic(U)) with pairs (a,b)∈Z1(G,Pic(U)) modulo those of the form (Δσ(v),Δτ(v)),
where
[TABLE]
Then any element of H1(G,Pic(U)) is the class of
[TABLE]
where x1,x2,y1 and y2∈Z.
If we let y2 be odd, it’s easy to prove it’s 4-torsion element.
∎
Remark 5.2**.**
Using methods of Colliot-Thélène,Dasheng Wei,and Fei Xu,we can obtain all 2-torsion elements:
(x+2,m−4a),(x−2,m−4a), (x2−4,m−4a)
We take x1=2,x2=1,y1=0,y2=1,we obtain this class (l1+l4−l2,l4),
since l1+l4=l2+l5 in Pic(U), (l5,l4) is a 4-torsion element in H1(G,Pic(U)).Let K=k(a,m), one has the following commutative diagram of exact sequences
[TABLE]
where the morphism ∂ is the connecting homomorphism of the following exact sequence
[TABLE]
d1(l5,l4)=(l5(1,1)+l5(1,−1),l5(1,1)+l4(−1,1)−l5(−1,1)−l4(1,1),l4(1,1)+l4(1,−1))∈Z2(G,Div(UK)),
let
[TABLE]
By [1,proposition 7.1(b) and proposition 8.4],we have
[TABLE]
This implies that div(fg)=l5(1,1)+l4(−1,1)−l5(−1,1)−l4(1,1)
Since div(y−m)=l5(1,1)+l5(1,−1), div(x−am)=l4(1,1)+l4(1,−1)), we obtain
[TABLE]
Now we claim (my−m,f(2a+m−m−4a)g(2a−m+m−4a),x−am)∈Z2(G,K(U)×),
it suffices to show
[TABLE]
in K(U)×, one can directly check this by using rational point (0,0,m) of UK.The cocycle determines a non-cyclic Azumaya algebras
A on U.
.
Proposition 5.3**.**
Let U be the affine scheme defined by
[TABLE]
*where a,m∈Z. Let K=Q(m,a), when
(i)
[K:Q]=4*, mam−4a∈Q *
2. (ii)
for any prime q,its decomposition group in Gal(K/Q) is cyclic
3. (iii)
*There exists a prime p≥5,such that p splits in Q(m) and has ramification index 2 in *Q(a)
then there is no Brauer-Manin obstruction to the integral Hasse principle for U.
Proof.
We can assume that U(AZ)=∅, where AZ=R×∏pZp , otherwise there is nothing to prove. Note that
since p splits in Q(m), its decomposition group is ⟨σ⟩. Hence for any T∈U(Zp),
A(T)=(my−m,a)∈Br(Qp).By (iii), we can assume ordp(a)=1. Note that
[TABLE]
Let B1=(x+2,a), B2=(y−m,a),to prove the proposition ,
it suffices to show for all (ε1,ε2)∈(Z/2Z)2, there exists
ζ∈U(Zp) such that
[TABLE]
Since ordp(a) is odd, we have p ∣m by (i) ,in fact ordp(m) is even by (iii).
Let s,t∈Fp×, such that the Legendre symbol (ps)=1, (pt)=−1, let
[TABLE]
One can see that υ1 and υ2 are smooth points of U(Fp), hence there exist μ1,μ2∈U(Zp)
such that μi≡υimodp for 1≤i≤2, we obtain
[TABLE]
By Dirichlet theorem,there exist a prime l,such that Legendre symbol (pl)=−1 and p∤(l+1). Let
[TABLE]
One can check that υ3 and υ4 are smooth points of U(Fp), hence there exist μ3,μ4∈U(Zp)
such that μi≡υimodp for 3≤i≤4, we obtain
[TABLE]
The proposition is established.
∎
Acknowledgements
I thank my advisor Fei Xu for countless helpful conversations and suggestions, and also for his patience. I also thank Yang Zhang for many useful discussions.
Bibliography16
The reference list from the paper itself. Each links out to its DOI / PubMed record.
1[1] W.Fulton , Intersection Theory. Springer Berlin , 1998.
2[2] J.-L. Colliot-Thélène et O. Wittenberg, Groupe de Brauer et points entiers de deux familles de surfacescubiques affines, American Journal of Mathematics 134 (2012), no. 5, 1303–1327
3[3] J.-L. Colliot-Thélène and F. Xu, Brauer-Manin obstruction for integral points of homogeneous spaces and representation by integral quadratic forms, Compos. Math. 145 (2009), no. 2, 309–363, with an appendix by D. Wei and F. Xu.
5[5] J.-L. Colliot-Théléne, D. Kanevsky, J.-J. Sansuc, Arithmétique des surfaces cubiques diagonales, in: Diophantine Approximation and Transcendence Theory, Lecture Notes in Math., vol. 1290, Springer, Berlin, 1987.
6[6] P. Gille and T. Szamuely, Central simple algebras and Galois cohomology, Cambridge Studies in Advanced Mathematics 101, Cambridge University Press, 2006.
7[7] R. Harshorne:, Algebraic Geometry, GTM 52, Springer.
8[8] A. Ghosh and P. Sarnak, Integral points in Markoff type cubic surfaces, ar Xiv: 1706.06712 v 2