LXXIX.4 (1997)
On the average of character sums for a group of characters
by
D. A. Burgess (Nottingham)
To J. W. S. Cassels
1. Introduction. Let r, k, h be positive integers. For any Dirichlet char- acter χ modulo k write
W
r(χ) = X
k x=1X
h m=1χ(x + m)
2r. In [2] it was shown that, if χ is a primitive character, then (1) W
2(χ) kh
2+ k
1/2+εh
4,
which implies that
(2) W
2(χ) k
1+εh
2for h ≤ k
1/4. In [6] it was shown that
(3) X
primitive χ (mod k)
W
2(χ) k
2+εh
2,
so that (2) holds for all h, on average for all primitive characters modulo k.
Thus it is reasonable to conjecture that (2) might hold for some h > k
1/4, on average for the primitive characters of a large subgroup of the characters modulo k. In [8] such a result was obtained, it being shown that, for any prime p,
(4) X
χ mod p3 χp2=χ0
W
2(χ) p
5h
2+ p
3h
4,
where χ
0is the principal character, and thus (2) holds for h ≤ k
1/3, on average for the characters modulo k = p
3in the group of order p
2.
1991 Mathematics Subject Classification: Primary 11L40.
[313]
In this paper the argument is strengthened to show in the following theorem that, for the non-principal characters of this group, (2) remains true for h ≤ k
1/2, on average.
Theorem 1. Let p be an odd prime number. Let
S = X
χ mod p3 χp2=χ0 χ6=χ0
p3
X
x=1
X
h m=1χ(x + m)
2r.
Then in the case r = 2 we have
S p
2h
4+ p
5h
2. From [1] it follows that if k is prime then (5) W
3(χ) kh
3+ k
1/2h
6for all positive h. By the methods of this paper it is shown that (5) can be improved for h > p
3/4, on average for the non-principal characters modulo k = p
3in the group of order p
2.
Theorem 2. Let p be an odd prime number. Let
S = X
χ mod p3 χp2=χ0
χ6=χ0
p3
X
x=1
X
h m=1χ(x + m)
2r.
Then in the case r = 3 we have
S p
2h
6+ min(h, p)
3p
5h + min(h, p)
2p
6.
In Section 8 we shall describe some corollaries of these theorems.
2. Preliminary transformation of the problem. For S as in the statement of both theorems, we have
(6) S = X
χ mod p3 χp2=χ0
p3
X
x=1
X
h m=1χ(x+m)
2r−
p3
X
x=1
X
hm=1
χ
0(x+m)
2r= S
1−S
2,
say.
Now
S
2= X
m p3
X
x=1
χ
0(f
1(x))χ
0(f
2(x)),
where m ∈ Z
2rsatisfies 0 < m
i≤ h for 1 ≤ i ≤ 2r, and
f
1(x) = (x + m
1) . . . (x + m
r), f
2(x) = (x + m
r+1) . . . (x + m
2r).
Thus
(7) S
2= X
m
p3
X
x6≡−mx=1i(mod p)
1 = X
m
p
2#{x : 1 ≤ x ≤ p, p - f
1(x)f
2(x)}.
We also have S
1= X
m p3
X
x=1
X
χp2=χ0
χ(f
1(x))χ(f
2(x)) = X
m
p3
X
x=1 p
-
f1(x)f2(x)X
χp2=χ0
χ
f
1(x) f
2(x)
= p
2X
m
#{1 ≤ x ≤ p
3, 1 ≤ z < p : p - f
1(x)f
2(x),
f
1(x) − z
p2f
2(x) ≡ 0 (mod p
3)}.
Thus, writing
(8) f
(z)(x) = f
1(x) − z
p2f
2(x), we have
(9) S
1= S
3+ S
4,
where
(10) S
3= p
2X
m
X
p−1 z=1#{1 ≤ x ≤ p
3: x 6≡ −m
i(mod p),
f
(z)(x) ≡ 0 (mod p
3), f
(z)0(x) 6≡ 0 (mod p)}
and
(11) S
4= p
2X
m
X
p−1 z=1#{1 ≤ x ≤ p
3: x 6≡ −m
i(mod p),
f
(z)(x) ≡ 0 (mod p
3), f
(z)0(x) ≡ 0 (mod p)}.
We consider the non-singular roots of f
(z). Since the numbers of non- singular roots modulo p
3and p are the same we have from (10) that
S
3≤ p
2X
m p−1
X
z=1
#{1 ≤ x ≤ p : p - f
1(x)f
2(x), f
(z)(x) ≡ 0 (mod p)}
= p
2X
m
X
p x=1 p-
f1(x)f2(x)#{1 ≤ z < p : f
1(x) − zf
2(x) ≡ 0 (mod p)}
= p
2X
m
#{1 ≤ x ≤ p : p - f
1(x)f
2(x)} = S
2from (7). Now from (6) and (9) we have
(12) S ≤ S
4.
3. Estimates for solution sets of polynomials. In the proof of our theorems we shall require some lemmas concerning the number of solutions of congruences to a prime power modulus. We shall use the following gen- eralisation of the well known estimate for the number of non-singular roots of a congruence.
Lemma 1. Let F be a polynomial of degree n having integer coefficients.
Let p be a prime, d a positive integer , and α, β and γ be non-negative integers satisfying γ = dα/de. Then
#{1 ≤ x ≤ p
γ: p
α+β| F (x), p
βk F
(d)(x)} ≤ n.
P r o o f. This is Proposition 1 of [7].
We shall require an estimate for the number of solutions of a congruence in many variables to a prime modulus. The following will suffice.
Lemma 2. Let G be a polynomial in x
1, . . . , x
twhich is not identically zero modulo the prime p. Let 0 < h
i≤ p for all i. Then the number of x, satisfying 0 < x
i≤ h
ifor all i, for which G(x) ≡ 0 (mod p) is O(( Q
h
i)/ min h
i).
P r o o f. This is an easy modification of Lemma 5 of [4].
We shall also use the following estimate which, under favourable condi- tions, can provide an optimal estimate for the average number of singular roots of a set of polynomials.
Lemma 3. Let F
i(y) (0 < i ≤ n) be polynomials in ν variables y
k(0 <
k ≤ ν). Let p be a prime and 2β ≥ α
1≥ . . . ≥ α
m> β ≥ α
m+1≥ . . . ≥ α
nbe positive integers. Let H ∈ N
ν. Write
N = #{y : ∀k ≤ ν 0 < y
k≤ H
k, ∀i F
i(y) ≡ 0 (mod p
αi)}.
Put λ
k= dH
k/p
βe for all k ≤ ν. Then
N X
∀k≤ν |BBk|<λk
#
y : ∀k ≤ ν 0 < y
k≤ p
β, ∀i ≤ m F
i(y) ≡ 0 (mod p
β),
∀i > m F
i(y) ≡ 0 (mod p
αi), ∀i ≤ m X
ν k=1B
k∂F
i∂y
k(y) ≡ 0 (mod p
αi−β)
. P r o o f. Clearly
N ≤ #{y : ∀k ≤ ν 0 < y
k≤ λ
kp
β, ∀i F
i(y) ≡ 0 (mod p
αi)}.
For 1 ≤ k ≤ ν write y
k= a
k+ p
βb
k, where 0 < a
k≤ p
β, 0 ≤ b
k< λ
k. Then, for m < i ≤ n, F
i(y) ≡ 0 (mod p
αi) becomes
F
i(a) ≡ 0 (mod p
αi), while for i ≤ m, F
i(y) ≡ 0 (mod p
αi) becomes
F
i(a) + X
ν k=1∂F
i∂y
k(a)b
kp
β≡ 0 (mod p
αi).
The latter congruences imply, for i ≤ m, F
i(a) ≡ 0 (mod p
β) so that, say,
∀i ≤ m F
i(a) = c
i(a)p
β. Thus we have
N ≤ X
∀i≤m Fi(a)≡0 (mod pa β)
∀i>m Fi(a)≡0 (mod pαi)
#
b : ∀k ≤ ν 0 ≤ b
k< λ
k,
∀i ≤ m X
ν k=1∂F
i∂y
k(a)b
k≡ −c
k(a) (mod p
αi−β)
. Now the number of solutions of the inhomogeneous congruences
∀i ≤ m X
ν k=1∂F
i∂y
k(a)b
k≡ −c
k(a) (mod p
αi−β)
in the variables b satisfying 0 ≤ b
k< λ
kfor all k ≤ ν is at most the number of solutions of the homogeneous congruences
∀i ≤ m X
ν k=1∂F
i∂y
k(a)B
k≡ 0 (mod p
αi−β)
in the variables B satisfying |B
k| < λ
kfor all k ≤ ν. Thus we have
N ≤ X
∀i≤m Fi(a)≡0 (mod pa β)
∀i>m Fi(a)≡0 (mod pαi)
#
B : ∀k ≤ ν |B
k| < λ
k,
∀i ≤ m X
ν k=1∂F
i∂y
k(a)B
k≡ 0 (mod p
αi−β)
≤ X
∀k≤ν |BBk|<λk
#
a : ∀k ≤ ν 0 < a
k≤ p
β, ∀i ≤ m F
i(a) ≡ 0 (mod p
β),
∀i > m F
i(a) ≡ 0 (mod p
αi), ∀i ≤ m X
ν k=1B
k∂F
i∂y
k(a) ≡ 0 (mod p
αi−β)
.
4. Proof of Theorem 1. Clearly in proving the theorem we may sup- pose that p > 2. We consider here the case r = 2.
It remains to consider the singular roots. Noting (11) we write
(13) S
4= S
5+ S
6,
where S
5= p
2X
m
#{1 ≤ x ≤ p
3, 1 ≤ z < p : x 6≡ −m
i(mod p),
f
(z)(x) ≡ 0 (mod p
3), f
(z)0(x) ≡ 0 (mod p), f
(z)00(x) ≡ 0 (mod p)}, and
S
6= p
2X
m p−1
X
z=2
#{1 ≤ x ≤ p
3: x 6≡ −m
i(mod p), f
(z)(x) ≡ 0 (mod p
3), f
(z)0(x) ≡ 0 (mod p), f
(z)00(x) 6≡ 0 (mod p)}.
Clearly we have S
5≤ p
2X
m
#{1 ≤ x ≤ p
3:
(m
1+ m
2− m
3− m
4)x + (m
1m
2− m
3m
4) ≡ 0 (mod p
3), (m
1+ m
2− m
3− m
4) ≡ 0 (mod p)}.
Write
(14) p
δ= highest common factor(p
3, m
1+ m
2− m
3− m
4), where 1 ≤ δ ≤ 3. For solubility of the congruence
(m
1+ m
2− m
3− m
4)x + (m
1m
2− m
3m
4) ≡ 0 (mod p
3) we require also
(15) p
δ| (m
1m
2− m
3m
4).
The congruence then has at most p
δsolutions satisfying 1 ≤ x ≤ p
3. (14) and (15) imply
(m
1− m
3)(m
2− m
3) ≡ 0 (mod p
δ).
Suppose that p
ε| (m
1− m
3) and p
δ−ε| (m
2− m
3). Then the number of such m is
O
h
1 + h
p
ε1 + h p
δ−ε1 + h p
δ= O
h
4p
2δ+ h
2. Thus
(16) S
5p
2X
δ,ε
p
δh
4p
2δ+ h
2ph
4+ p
5h
2.
On the other hand, we have S
6p
2X
m
X
p−1 z=2#{1 ≤ x ≤ p
3: f
(z)(x) ≡ 0 (mod p
3),
f
(z)0(x) ≡ 0 (mod p), f
(z)00(x) 6≡ 0 (mod p)}.
Thus, by Lemma 1,
S
6p
3#{m, 1 < z < p : ∃x f
(z)(x) ≡ 0 (mod p
2), f
(z)0(x) ≡ 0 (mod p)}.
Thus
(17) S
6p
3#{m, z, x : ∀i 0 < m
i≤ h, 0 < x ≤ p, 0 < z < p, f
(z)(x) ≡ 0 (mod p
2), f
(z)0(x) ≡ 0 (mod p)}.
Put
(18) λ =
h p
, µ =
n p if λ > 1, h if λ = 1.
Now we apply Lemma 3, treating x, z as constants and the m
ias our vari- ables, to obtain
(19) S
6p
3X
∀k≤4 |BBk|<λ
#
a, x, z : ∀k ≤ 4 0 < a
k≤ µ, 0 < x ≤ p,
0 < z < p, f
(z)(x) ≡ 0 (mod p), f
(z)0(x) ≡ 0 (mod p),
X
4 k=1B
k∂f
(z)∂m
k(x) ≡ 0 (mod p)
if λ > 1, while if λ = 1 this follows immediately from (17).
Given B, x, z, a
3, a
4we have, from (19),
f
(z)0(x) ≡ 2(1 − z)x + (a
1+ a
2− za
3− za
4) ≡ 0 (mod p),
from which a
1+ a
2is determined modulo p. Then also from (19) we have f
(z)(x) ≡ (1 − z)x
2+ (a
1+ a
2− za
3− za
4)x + (a
1a
2− za
3a
4) ≡ 0 (mod p), and so a
1a
2is also determined modulo p. Thus there are at most two choices for a
1, a
2. Use these congruences to eliminate a
1, a
2. We have on writing
π
1= a
1+ a
2, π
2= a
1a
2, %
1= a
3+ a
4, %
2= a
3a
4, the identity
(π
12− 4π
2)(B
1− B
2)
2− (2B
2a
1+ 2B
1a
2− π
1(B
1+ B
2))
2= 0.
Now from (19) we have X
4k=1
B
k∂f
(z)(x)
∂m
k≡ (B
1+ B
2− zB
3− zB
4)x
+ (a
1B
2+ a
2B
1− za
3B
4− za
4B
3)
≡ 0 (mod p).
Thus eliminating a
1, a
2we have
(20) (z
2%
21− 4z%
1(1 − z)x − 4x
2z(1 − z) − 4z%
2)(B
1− B
2)
2−z
2(2(B
3a
4+ B
4a
3) + 2x(B
3+ B
4) − (%
1+ 2x)(B
1+ B
2))
2≡ 0 (mod p).
Thus from (19), (21) S
6p
3X
B
#{x, z, a
3, a
4:
(z
2%
21− 4z%
1(1 − z)x − 4x
2z(1 − z) − 4z%
2)(B
1− B
2)
2−z
2(2(B
3a
4+ B
4a
3) + 2x(B
3+ B
4) − (%
1+ 2x)(B
1+ B
2))
2≡ 0 (mod p)}.
By Lemma 2, for a given choice of B, (20) has at most O(p
2µ) solutions in x, z, a
3, a
4, unless this polynomial is identically zero modulo p. If the coefficient of za
3a
4is zero we have
−4(B
1− B
2)
2≡ 0 (mod p),
and thus B
1≡ B
2(mod p). Under this condition if the coefficient of z
2a
23is zero we have
−(2B
4− B
1− B
2)
2≡ 0 (mod p), and if the coefficient of z
2a
24is zero we have
−(2B
3− B
1− B
2)
2≡ 0 (mod p).
Thus if the polynomial is identically zero modulo p we have B
1≡ B
2≡ B
3≡ B
4(mod p).
Hence the number of such cases is O(λ(1 + λ/p)
3).
Consequently, from (21) we have S
6p
3λ
4p
2µ +
λ + λ
4p
3p
2µ
2p
2h
4+ p
5h
2. The theorem follows from (12), (13) and (16).
5. Introduction to proof of Theorem 2. We may suppose that p > 2.
We consider here the case r = 3. Noting (11) we write
(22) S
7= X
m
#{x, z : 0 < x ≤ p
3, 0 < z < p, f
1(x)f
2(x) 6≡ 0 (mod p), f
(z)(x) ≡ 0 (mod p
3), f
(z)0(x) ≡ 0 (mod p), either f
(z)06≡ 0 (mod p
2) or f
(z)00(x) 6≡ 0 (mod p)}
and
(23) S
8= X
m
#{x, z : 0 < x ≤ p
3, 0 < z < p, f
1(x)f
2(x) 6≡ 0 (mod p), f
(z)(x) ≡ 0 (mod p
3), f
(z)0(x) ≡ 0 (mod p
2), f
(z)00(x) ≡ 0 (mod p)}
so that
(24) S
4= p
2S
7+ p
2S
8. We estimate S
7and S
8in Section 7.
We shall use also the polynomials g
i(x) given by
∀i ≤ 3 g
i(x) = f
1(x)/(x + m
i), ∀i ≥ 4 g
i(x) = f
2(x)/(x + m
i).
Thus we have, from (8),
f
(z)(x) = g
1(x)(x + m
1) − z
p2g
4(x)(x + m
4) and
f
(z)0(x) = g
1(x) + g
2(x) + g
3(x) − z
p2g
4(x) − z
p2g
5(x) − z
p2g
6(x).
Write
C
1(m) = g
1(x)(x + m
1) − zg
4(x)(x + m
4),
C
2(m) = (2x + m
2+ m
3)(x + m
1) − z(2x + m
5+ m
6)(x + m
4) + (g
1(x) − zg
4(x)),
C
3(m) = 2(x + m
1) − 2z(x + m
4)
+ ((4x + 2m
2+ 2m
3) − z(4x + 2m
5+ 2m
6)),
C
4(m) = b
1g
1(x) + b
2g
2(x) + b
3g
3(x) − b
4zg
4(x) − b
5zg
5(x) − b
6zg
6(x)
= (b
2(x + m
3) + b
3(x + m
2))(x + m
1)
− z(b
5(x + m
6) + b
6(x + m
5))(x + m
4) + (b
1g
1(x) − b
4zg
4(x)) and
C
5(m) = b
1(2x + m
2+ m
3) + b
2(2x + m
1+ m
3) + b
3(2x + m
1+ m
2)
− b
4z(2x + m
5+ m
6) − b
5z(2x + m
4+ m
6)
− b
6z(2x + m
4+ m
5)
= (b
2+ b
3)(x + m
1) − z(b
5+ b
6)(x + m
4)
+ (b
1(2x + m
2+ m
3) + b
2(x + m
3) + b
3(x + m
2)
− zb
4(2x + m
5+ m
6) − zb
5(x + m
6) − zb
6(x + m
5)).
We define λ and µ by (18).
6. Minor lemmas Lemma 4. Write
(25) D
1=
g
1(x) −zg
4(x) 2x + m
2+ m
3−z(2x + m
5+ m
6)
. Then X
∀i |bbi|<λ
#{m, x, z : 0 < m
i≤ µ, 0 < x ≤ p, 0 < z < p,
f
1(x)f
2(x) 6≡ 0 (mod p), C
1(m) ≡ C
2(m) ≡ 0 (mod p), D
1≡ 0 (mod p)}
pµ
4λ
6. P r o o f. From C
1(m) ≡ C
2(m) ≡ D
1≡ 0 (mod p) it follows that
g
1(x) ≡ zg
4(x) (mod p),
from which z is uniquely determined. Then from C
1(m) ≡ 0 (mod p) it follows that m
1= m
4. Finally,
D
1= z((m
5+ m
6− m
2− m
3)x
2+ 2(m
5m
6− m
2m
3)x + m
5m
6(m
2+ m
3) − m
2m
3(m
5+ m
6)),
so that, by Lemma 2, D
1/z ≡ 0 (mod p) has O(pµ
3) solutions as a function of m
2, m
3, m
5, m
6, x. Thus the required estimate follows trivially.
Lemma 5. Write D
2=
m
2m
3−zm
5m
60
m
2+ m
3−z(m
5+ m
6) m
2m
3− zm
5m
6b
2m
3+ b
3m
2−zb
5m
6− zb
6m
5b
1m
2m
3− b
4zm
5m
6.
Then X
∀i |bbi|<λ D2identically 0 (mod p)
#{m
2, m
3, m
5, m
6, x, z :
0 < m
i≤ µ, 0 < x ≤ p, 0 < z < p}
p
2µ
4λ
λ p + 1
5. P r o o f. We have
D
2= (b
6− b
1)zm
22m
23m
5+ (b
5− b
1)zm
22m
23m
6+ (b
1− b
3)zm
22m
3m
5m
6+ (b
1− b
2)zm
2m
23m
5m
6+ (b
4− b
6)z
2m
2m
3m
25m
6+ (b
4− b
5)z
2m
2m
3m
5m
26+ (b
3− b
4)z
2m
2m
25m
26+ (b
2− b
4)z
2m
3m
25m
26.
This is identically 0 (mod p) only if
b
1≡ b
2≡ b
3≡ b
4≡ b
5≡ b
6(mod p).
The required estimate follows trivially.
Lemma 6. We have X
∀i |bbi|<λ
#{m, x, z : 0 < m
i≤ µ, 0 < x ≤ p,
0 < z < p, f
1(x)f
2(x) 6≡ 0 (mod p), C
1(m) ≡ C
2(m) ≡ C
3(m) ≡ 0 (mod p), D
1≡ 0 (mod p)}
pµ
3λ
6, where D
1is defined by (25).
P r o o f. From C
1(m) ≡ C
2(m) ≡ D
1≡ 0 (mod p) it follows that g
1(x) ≡ zg
4(x) (mod p),
from which z is uniquely determined. Then from C
1(m) ≡ 0 (mod p) it follows that m
1= m
4and, since f
1(x)f
2(x) 6≡ 0 (mod p), from C
2(m) ≡ 0 (mod p) that
(2x + m
2+ m
3) − z(2x + m
5+ m
6) ≡ 0 (mod p).
Now substituting into C
3(m) ≡ 0 (mod p) we obtain z = 1 and so also m
2+ m
3≡ m
5+ m
6(mod p). But also we have g
1(x) ≡ g
4(x) (mod p) and so m
2m
3≡ m
5m
6(mod p). Thus m
2, m
3is a permutation of m
5, m
6. The required estimate follows trivially.
Lemma 7. Write D
3=
g
1(x) g
4(x) 0
2x + m
2+ m
32x + m
5+ m
6g
1(x)
1 1 2x + m
2+ m
3and
D
4=
g
1(x) g
4(x) 0
2x + m
2+ m
32x + m
5+ m
6g
4(x)
1 1 2x + m
5+ m
6. Then X
∀i |bbi|<λ
#{m
2, m
3, m
5, m
6, x :
0 < m
i≤ µ, 0 < x ≤ p, D
3≡ D
4≡ 0 (mod p)}
pµ
2λ
6. P r o o f. The conditions D
3≡ D
4≡ 0 (mod p) expand to give
(g
1(x) − g
4(x))((2x + m
5+ m
6)(2x + m
2+ m
3) − g
1(x))
−g
4(x)(m
2+ m
3− m
5− m
6)(2x + m
2+ m
3)
≡ (g
1(x) − g
4(x))((2x + m
5+ m
6)
2− g
4(x))
− g
4(x)(m
2+ m
3− m
5− m
6)(2x + m
5+ m
6) ≡ 0 (mod p).
These will have only O(1) solutions x unless both polynomials are identically 0 (mod p). But in the first of these the coefficient of x
3is m
2+m
3−m
5−m
6, and if this is 0 (mod p) then the coefficient of x
2is 3(m
2m
3− m
5m
6). Thus if both polynomials in x are identically 0 (mod p) then the pair m
2, m
3is a permutation of m
5, m
6. This contributes
(26) λ
6pµ
2to our estimate.
Now consider the other case in which at least one of D
3and D
4is not identically 0 (mod p). Then there are only O(1) values for x. The two polynomial congruences D
3≡ D
4≡ 0 (mod p) are cubics. The difference between these polynomials is
g
1(x) g
4(x) 0
2x + m
2+ m
32x + m
5+ m
6g
1(x) − g
4(x)
1 1 m
2+ m
3− m
5− m
6. By row and column operations this simplifies to
m
2m
3m
5m
60
m
2+ m
3m
5+ m
6m
2m
3− m
5m
61 1 m
2+ m
3− m
5− m
6,
which is a polynomial in m
2, m
3, m
5, m
6. This polynomial is −1 when m
2= m
3= 1, m
5= m
6= 0. Thus it is not identically 0 (mod p) and so has O(µ
3) solutions in m
2, m
3, m
5, m
6. Thus this contributes
(27) λ
6µ
3to our estimate. The lemma follows from (26) and (27).
Lemma 8. Write D
5=
m
2m
3−m
5m
60
m
2+ m
3−(m
5+ m
6) m
2m
3b
2m
3+ b
3m
2−b
5m
6− b
6m
5b
1m
2m
3×
m
2m
3− m
5m
6m
5m
60 m
2+ m
3− m
5− m
6m
5+ m
6m
5m
60 1 m
5+ m
6−
m
2m
3−m
5m
60
m
2+ m
3−(m
5+ m
6) m
5m
6b
2m
3+ b
3m
2−b
5m
6− b
6m
5b
4m
5m
6×
m
2m
3− m
5m
6m
5m
60 m
2+ m
3− m
5− m
6m
5+ m
6m
2m
30 1 m
2+ m
3.
Then
X
∀i |bbi|<λ D5identically 0 (mod p)
#{m
2, m
3, m
5, m
6: 0 < m
i≤ µ} µ
4λ
λ p + 1
5.
P r o o f. Substitute m
6= 0 in D
5to obtain (b
6− b
1)m
32m
33m
35. Thus if D
5is identically 0 (mod p) then b
6≡ b
1(mod p). Similar arguments give
b
1≡ b
5≡ b
6, b
2≡ b
3≡ b
4(mod p).
Substituting this in D
5, and putting m
2= m
3= m
5= 1 we obtain
−(b
4− b
1)(1 − m
6)
2m
6also. The required estimate follows trivially.
7. Proof of Theorem 2 Lemma 9. We have
S
7h
6+ p
3hµ
3, where µ is defined by (18).
P r o o f. From Lemma 1 applied to (22) it follows that S
7p X
m
#{z : 0 < z < p, ∃x f
1(x)f
2(x) 6≡ 0 (mod p),
f
(z)(x) ≡ 0 (mod p
2), f
(z)0(x) ≡ 0 (mod p)}.
We can rewrite this as (28) S
7p
X
p x=1p−1
X
z=1
#{m : f
1(x)f
2(x) 6≡ 0 (mod p),
f
(z)(x) ≡ 0 (mod p
2), f
(z)0(x) ≡ 0 (mod p)}, say. Write
(29) N = #{m : f
1(x)f
2(x) 6≡ 0 (mod p), f
(z)(x) ≡ 0 (mod p
2), f
(z)0(x) ≡ 0 (mod p)}.
Thus by Lemma 3,
(30) N X
∀i |bbi|<λ
#{m : ∀i 0 < m
i≤ µ, f
1(x)f
2(x) 6≡ 0 (mod p), 0 ≡ C
1(m) ≡ C
2(m) ≡ C
4(m) (mod p)}
if λ > 1, and follows immediately from (29) if λ = 1. Thus from (28) we have
S
7p X
∀i |bbi|<λ