1808 04001

ON THE CHOWLA AND TWIN PRIMES CONJECTURES
OVER Fq [T ]
arXiv:1808.04001v2 [math.NT] 7 Sep 2019
WILL SAWIN AND MARK SHUSTERMAN
Abstract. Using geometric methods, we improve on the function field

version of the Burgess bound, and show that, when restricted to certain
special subspaces, the Möbius function over Fq [T ] can be mimicked by
Dirichlet characters. Combining these, we obtain a level of distribution
close to 1 for the Möbius function in arithmetic progressions, and re-
solve Chowla’s k-point correlation conjecture with large uniformity in
the shifts. Using a function field variant of a result by Fouvry-Michel
on exponential sums involving the Möbius function, we obtain a level of
distribution beyond 1/2 for irreducible polynomials, and establish the
twin prime conjecture in a quantitative form. All these results hold for
finite fields satisfying a simple condition.
1. Introduction
Our main results are the resolutions of two open problems in number
theory, except with the ring of integers Z replaced by the ring of polynomials
Fq [T ], under suitable assumptions on q.
We first fix some notation. Define the norm of a nonzero f ∈ Fq [T ] to be
(1.1) |f | = q deg(f ) = |Fq [T ]/(f )|.
The degree of the zero polynomial is negative ∞, so we set its norm to be 0.
1.1. The main result - twin primes. Our main result covers the twin
prime conjecture in its quantitative form. The latter is the 2-point prime
tuple conjecture of Hardy-Littlewood, predicting for a nonzero integer h that
X
(1.2) #{X ≤ n ≤ 2X : n and n+h are prime} ∼ S(h) 2 , X → ∞,
log (X)
where
Y
(1.3) S(h) = (1 − p−1 )−2 (1 − p−1 − p−1 1p∤h ),
p
with 1p∤h equals 1 if h is not divisible by p, and 0 otherwise.

For the function field analogue, we set
Y −2
(1.4) Sq (h) = 1 − |P |−1 1 − |P |−1 − |P |−1 1P ∤h
P
W.S. served as a Clay Research Fellow while working on this paper.

1
2 WILL SAWIN AND MARK SHUSTERMAN
where q is a prime power, P ranges over all primes (monic irreducibles) of

Fq [T ], and h ∈ Fq [T ] is nonzero.
Theorem 1.1. For an odd prime number p, and a power q of p satisfying

q > 685090p2 , the following holds. For any nonzero h ∈ Fq [T ] we have
X
(1.5) #{f ∈ Fq [T ] : |f | = X, f and f + h are prime} ∼ Sq (h)
log 2q (X)
as X → ∞ through powers of q. Moreover, we have a power saving (depend-

ing on q) in the asymptotic above.
For example, the 2-point Hardy-Littlewood conjecture holds over
(1.6) F315 , F511 , F79 , F118 , F6850933 .
In case h ∈ Fq [T ] is a monomial, the fact that the count above tends

to ∞ as X → ∞ (under the weaker assumption q > 105) has been proved
in [CHLPT15, Theorem 1.3] using an idea of Entin. Their work builds on
the recent dramatic progress on this problem over the integers, particularly
[Ma15]. The strongest result known over the integers is [PM14, Theorem
16(i)], which says that for any ‘admissible tuple’ of 50 integers, there exists
at least one difference h between two elements in the tuple such that there
are infinitely many pairs of primes separated by h.
Remark 1.2. Our proof of Theorem 1.1 establishes also the analog of the
Goldbach problem over function fields, and can be modified to treat more
general linear forms in the primes.
The proof of Theorem 1.1 passes through some intermediate results which
may be of independent interest. We will discuss these results in the remain-
der of the introduction. Once Theorem 1.7 and Theorem 1.9 below are
established, Theorem 1.1 will follow from a quick argument involving a con-
volution identity relating the von Mangoldt function, which can be used to
count primes, to the Möbius function.
1.2. The key ingredient - Chowla’s conjecture. The main ingredient

in the proof of Theorem 1.1 is the removal of the ‘parity barrier’. More
precisely, we confirm Chowla’s k-point correlation conjecture over Fq [T ] for
some prime powers q. Over the integers, this conjecture predicts that for
any fixed distinct integers h1 , . . . , hk , one has
X
(1.7) µ(n + h1 )µ(n + h2 ) · · · µ(n + hk ) = o(X), X → ∞.
n≤X
The only completely resolved case is k = 1 which is essentially equivalent to

the prime number theorem.
ON THE CHOWLA AND TWIN PRIMES CONJECTURES OVER Fq [T ] 3
For the function field analogue, we recall that the Möbius function of a
monic polynomial f is given by

 2
0, #{P : P | f } > 0
(1.8) µ(f ) = 1, #{P : P | f } ≡ 0 mod 2


−1, #{P : P | f } ≡ 1 mod 2,
and denote by Fq [T ]+ the set of monic polynomials over Fq .
Theorem 1.3. For an odd prime number p, an integer k ≥ 1, and a power
q of p satisfying q > p2 k2 e2 , the following holds. For distinct h1 , . . . , hk ∈
Fq [T ] we have
X
(1.9) µ(f + h1 )µ(f + h2 ) · · · µ(f + hk ) = o(X), X → ∞.
f ∈Fq [T ]+
|f |≤X
For instance, the 2-point Chowla conjecture holds over

(1.10) F36 , F55 , F74 , F313 .
In fact, in Theorem 1.3 we obtain a power saving inversely proportional
to p, and the shifts h1 , . . . , hk can be as large as any fixed power of X
(the corresponding assumption on q becomes stronger as this power grows
larger). In Corollary 6.1 we also get cancellation in case the sum is restricted
to prime polynomials f .
Over the integers, the k = 2 case of the Chowla conjecture, with log-
arithmic averaging, was proven by Tao [Tao16, Theorem 3], building on
earlier breakthrough work of Matomäki and Radziwill [MR16]. The k odd
case, again with logarithmic averaging, was handled by Tao and Teräväinen
[TT17]. Generalizations of some of these arguments to the function field set-
ting are part of a work in progress by Klurman, Mangerel, and Teräväinen.
In contrast to these works, which deal with any sufficiently general (i.e.
non-pretentious) multiplicative function, our result relies on special prop-
erties of the Möbius function (in positive characteristic). Specifically, we
observe that for any fixed polynomial r, the function µ(r + sp ) essentially
equals χDr (s + cr ) where χDr is a quadratic Dirichlet character and cr is a
shift, both depending only on r (and not on s). This observation is very
closely related to the properties of Möbius described in [CCG08], specifi-
cally [CCG08, Theorem 4.8]. Conrad, Conrad, and Gross prove a certain
quasiperiodicity property in s for a general class of expressions of the form
r + sp , while we give a more precise description via Dirichlet characters in a
special case.
Our observation on µ(r + sp ) arises from the connection between the
parity of the number of prime factors of a squarefree f ∈ Fq [T ], and the sign
(inside the symmetric group) of the Frobenius automorphism acting on the
roots of f . In odd characteristic, this sign is determined by the value of the
quadratic character of F× q on the discriminant of f , i.e. the resultant of f
and its derivative f ′ . In characteristic p, the derivative of f = r + sp is equal

to the derivative of r, so the aforementioned sign of Frobenius is determined
by the quadratic character of the resultant of f with the fixed polynomial
r ′ . The latter is a quadratic Dirichlet character of f , and thus equals an
additively shifted Dirichlet character of s.
To use our observation, we restrict the sum in Theorem 1.3 to f of the
form r + sp for any fixed r, and obtain a short sum in s of a product of
additively shifted Dirichlet characters. As the conductors of these characters
are typically essentially coprime, we arrive at a short sum of a single Dirichlet
character.
Typically in analytic number theory, short character sums are handled by
the method of Burgess, who showed in [Bur63] that for a real number η >
1/4, a squarefree integer M , a real number X ≥ |M |η , and a nonprincipal
Dirichlet character χ mod M , one has

X

(1.11) sup χ(s + a) = o(X), |M | → ∞.
s∈Z
|a|≤X
Refining the method of Burgess is the focus of several works, but the expo-
nent 1/4 has not yet been improved (even conditionally). However, in the
function field setting, we can do better by a geometric method, as long as q
is sufficiently large.
Theorem 1.4. Fix η > 0. Then for a prime power q > e2 /η 2 the follow-
ing holds. For a squarefree M ∈ Fq [T ], a real number X ≥ |M |η , and a
nonprincipal Dirichlet character χ mod M , we have

X

(1.12) sup χ(s + a) = o(X), |M | → ∞.
s∈Fq [T ] |a|≤X
By further enlarging q, we get arbitrarily close to square root cancellation.

This is stated more precisely in Corollary 2.7.
To prove Theorem 1.4, we express the problem geometrically, viewing
the short interval {s + a : |a| ≤ X} as an affine space over Fq , and the
character χ as arising from a sheaf on that space. Following a strategy from
[Hoo91, appendix by Katz], we use vanishing cycles theory to compare the
cohomology of this sheaf for the s = 0 short interval and its cohomology for
a general short interval. Vanishing cycles can only occur when the vanishing
locus of χ is not (geometrically) a simple normal crossings divisor. Arguing
as in [Kat89], we split the modulus M of χ into a product of distinct linear
terms over Fq , which makes our vanishing locus a union of the hyperplanes
where the linear terms vanish, so we can check that this is a simple normal
crossings divisor away from some isolated points. This implies that the
cohomology groups vanish until almost the middle degree. Since we can
precisely calculate the vanishing cycles at the isolated points, we get a very
good control of the dimensions of cohomology groups as well.
Remark 1.5. The relation between Möbius and multiplicative characters

is less powerful the larger p is, as then fewer polynomials share a given
derivative. On the other hand, our geometric character sum bounds become
stronger as q grows. Thus, to make this method of proving Chowla work,
we need q to be sufficiently large with respect to p.
Remark 1.6. The study of the statistics of polynomial factorizations by ex-
amining Frobenius as an element of the symmetric group has been very fruit-
ful in the ‘large finite field limit’, where (in the notation of Theorem 1.3)
X is kept fixed and q is allowed to grow. We refer to [CaRu14], [Ca15],
[GS18] (and references therein) for the large finite field analogs of Theo-
rem 1.3 which save a fixed power of q. Our methods likely give an improved
savings in the large finite field limit when the characteristic is fixed, as long
as the degrees of the polynomials are sufficiently large with respect to the
characteristic, but we have not carefully calculated the resulting bounds in
this range.
1.3. Further ingredients - level of distribution estimates. Another
ingredient in the proof of Theorem 1.1 is an improvement of the level of
distribution of the Möbius and von Mangoldt functions in arithmetic pro-
gressions. Over the integers, assuming the Generalized Riemann Hypothesis
(GRH), this level of distribution is (at least) 1/2, which means that
X X
X X X
(1.13) µ(n) = o , Λ(n) = +o
|M | ϕ(M ) |M |
n≤X n≤X
n≡a mod M n≡a mod M
1
where M, a are coprime integers, and |M | ≤ X 2 −ǫ (for any fixed ǫ > 0 and
A ∈ R). For Möbius over Fq [T ], we obtain a level of distribution close to 1.
Theorem 1.7. Fix η > 0. For an odd prime number p, and a power q of p
2
with q > p2 e2 η2 − 1 , the following holds. For coprime M, a ∈ Fq [T ], and
a real number X with X 1−η ≥ |M | we have
X
X
(1.14) µ(f ) = o , |M | → ∞.
+
|M |
f ∈Fq [T ]
|f |≤X
f ≡a mod M
As in the previous theorems, we obtain a power savings estimate. Here

however (as opposed to Theorem 1.3), every f in our sum may have a dif-
ferent derivative, so a somewhat more elaborate implementation of our ob-
servation on the Möbius function is required. We put f = M g + a, and
wishfully write
a
(1.15) µ(M g + a) ≈ µ(M )µ g +
M
in order to create coincidences among the derivatives of the inputs to the
Möbius function. This is carried out more formally in Lemma 3.2, where we
show that for a power q of an odd prime p, and coprime a, M ∈ Fq [T ], the

function s 7→ µ(a + sp M ) is essentially proportional to an additive shift of
a (quadratic) Dirichlet characters in s, with the modulus of the character
depending on a and M in an explicit way. To visualize the power of this
claim, we view Fq [T ] as a rank p lattice over its subring Fq [T p ]. Restricting
the Möbius function to any line in this lattice gives a Dirichlet character
whose modulus varies with the line.
In order to deduce from Theorem 1.7 an improved level of distributions for

primes, we establish in the appendix a function field variant of [FM98] giving
quasi-orthogonality of the Möbius function and ‘inverse additive characters’.
While Fouvry-Michel work with characters to prime moduli, in order to
establish Theorem 1.1 we need arbitrary squarefree moduli.
Theorem 1.8. Let q be a prime power, and let ǫ > 0. Then for a squarefree
M ∈ Fq [T ], and an additive character ψ mod M , we have
X 3 25
(1.16) µ(f )ψ f ≪ |M | 16 +ǫ X 32 , X, |M | → ∞
f ∈Fq [T ]+
|f |≤X
(f,M )=1
where f denotes the inverse of f mod M , and the implied constant depends
only on q and ǫ.
For nonzero M ∈ Fq [T ] we define Euler’s totient function by

(1.17) ϕ(M ) = (Fq [T ]/(M ))× ,
and for f ∈ Fq [T ]+ , we define the von Mangoldt function by
(
deg(P ), f = P n
(1.18) Λ(f ) =
0, otherwise.
1
Theorem 1.9. Fix δ < 126 .
For an odd prime p and a power q of p with
2
2 2 51 − 26δ
(1.19) q>p e ,
1 − 126δ
the following holds. For a squarefree M ∈ Fq [T ], a polynomial a ∈ Fq [T ]
1
coprime to M , and X a power of q with X 2 +δ ≥ |M | we have
X
X X
(1.20) Λ(f ) = +o , |M | → ∞.
+
ϕ(M ) |M |
f ∈Fq [T ]
|f |≤X
f ≡a mod M
1
We have not ventured too much into improving the constant 126 , as our
1
method cannot give anything above 6 , even if Theorem 1.8 would give square
root cancellation. Since our proof of Theorem 1.9 is based on the ‘convo-
lutional’ connection of von Mangoldt and Möbius, it is not surprising that
a level of distribution of 32 = 21 + 61 , which is a longstanding barrier for

the divisor function over Z (perhaps the most basic convolution), is a natu-
ral limit of our techniques. A large finite field variant of Theorem 1.7 and
Theorem 1.9 was earlier proved in [BBSR15, Theorem 2.5].
Remark 1.10. It would be interesting to see whether our results can be
extended to characteristic 2, perhaps in a manner similar to which [Ca15]
extends the results of [CaRu14].
1.4. Additional results in small characteristic. Throughout this work,
we have not made every possible effort to reduce the least values of the
prime powers q to which our theorems apply. Instead, we present in the last
section some results that hold for q as small as 3.
The first concerns sign changes of Möbius in short intervals. Improving
on many previous works, Matömaki and Radziwill have shown in [MR16]
that for any η > 1/2, and any large enough positive integer N , there exist
integers a, b with |a|, |b| ≤ N η such that µ(N + a) = 1, µ(N + b) = −1. In
characteristic 3, we show that the exponent 1/2 can be improved to 3/7.
Theorem 1.11. Let q be a power of 3, and fix 3/7 < η < 1. Then for any
f ∈ Fq [T ]+ of large enough norm, there exist g, h ∈ Fq [T ] with |g|, |h| ≤ |f |η
such that µ(f + g) = 1 and µ(f + h) = −1.
We follow the same proof strategy relating Möbius to characters, but since
we allow small values of q, we cannot apply Theorem 1.4 anymore. Now
however, once we are interested in sign change only (and not cancellation),
we can focus on just one of the derivatives appearing. It turns out that if
this derivative has a relatively large order of vanishing at 0, the conductor of
the associated character is relatively small, and we can apply a function field
version (see [Hsu99]) of the aforementioned result of Burgess. For p > 3,
the character sums arising are too short for the Burgess bound to apply.
For large enough q, Theorem 1.11 follows from Theorem 1.7 (since T 7→
1/T allows one to think of short intervals as arithmetic progressions), and
also from the k = 1 case of Theorem 1.3 (as we have sufficient uniformity in
the shift).
Our last result, in the spirit of [Gal72], shows that Möbius enjoys cancel-
lation in the arithmetic progression 1 mod a growing power of a fixed prime
P , no matter how slowly does the length of the progression increase.
Theorem 1.12. Let q be a power of 3. Fix an irreducible P ∈ Fq [T ]. Then
for a positive integer n we have
X
X X
(1.21) µ(f ) = o n
, → ∞.
+
|P | |P |n
f ∈Fq [T ]
|f |≤X
f ≡1 mod P n
As before, we can obtain a power saving. Since the progressions are so

short, this result does not follow from the previous ones, even if q is large.
In view of Maier’s phenomenon, we cannot expect to obtain the theorem for

all residue classes.
1.5. Further directions. In future work, we hope to use some of the meth-
ods introduced here to address the following problems:
• Obtaining cancellation in ‘polynomial Möbius sums’ such as
X
µ f2 + T
f ∈Fq [T ]+
|f |≤X
which may be relevant for counting primes of the form f 2 + T .

• Calculating the variance (and higher moments) of the Möbius func-
tion in short intervals (and arithmetic progressions) over Fq [T ].
1.6. Notation. From this point on, it will be more convenient to work with
degrees of polynomials instead of absolute values. For g ∈ Fq [T ] we denote
its degree by d(g). By convention, the latter is −∞ if g = 0. The letter q
denotes a prime power, and is often suppressed from notation such as

(1.22) Md = g ∈ Fq [T ]+ : d(g) = d .
2. Character sums
The main result of this section is the following.
Theorem 2.1. Let t ≤ m be natural numbers, let f ∈ Fq [T ], let g ∈ Mm
be squarefree, and let χ : (Fq [T ]/g)× → C× be a nontrivial character. Then

X m−1 t
1/2
(2.1) χ(f + h) ≤ (q + 1) q2.
t
h∈Fq [T ]
d(h)<t
gcd(f +h,g)=1
To prove this theorem, we use the following geometric setup. View At as

a space parameterizing polynomials h of degree less than t and let (c1 , c2 )
be coordinates on A2 . Let U ⊆ At × A2 be the open set consisting of points
(h, (c1 , c2 )) where c1 f + h + c2 T t is prime to g. Let j : U → Pt × A2 be the
open immersion, embedding At into Pt in the usual way. Let π : Pt ×A2 → A2
be the projection.
Let T be the torus parameterizing of polynomials of degree less than
m that are relatively prime to g. (The space T isa torus because, over
an algebraically closed field, we may factor g and write T as Gm m , with
the coordinates given by evaluating the polynomial on each of the roots
of g.) On T , we have a character sheaf Lχ whose trace function is χ,
constructed by the Lang isogeny. Let Lχ (c1 f + h + c2 T t ) be the pullback of
this sheaf to U along the natural map from U to T that sends (h, (c1 , c2 ))
to c1 f + h + c2 T t . We will prove (2.1) using geometric properties of the

complex Rπ∗ j! Lχ (c1 f + h + c2 T t ).
Lemma 2.2. The complex Rπ∗ j! Lχ (c1 f + h+ c2 T t ) is geometrically isomor-
×
phic to its pullback under the map (c1 , c2 ) → (λc1 , λc2 ) for any λ ∈ Fq .
Proof. Because U is invariant under multiplying c1 , c2 , and h by λ, as are the
maps j and π, its suffices to check that Lχ (c1 f + h + c2 T t ) is geometrically
isomorphic to its pullback under this multiplication map. It is sufficient that
Lχ on T is geometrically isomorphic to its pullback under any multiplicative
translation, which follows from its construction as a character sheaf.
Lemma 2.3. The stalk of Rπ∗ j! Lχ (c1 f + h+ c2 T t ) at the point (0, 1) is
supported in degree t, where it has rank m−1
t .
Proof. When c1 = 0, c2 = 1, the polynomial c1 f + h + c2 T t = T t + h is monic
and has degree t. By the proper base change theorem, our stalk is thus
equivalent to the compactly supported cohomology of the space of degree t
monic polynomials that are prime to g, with coefficients in Lχ (T t + h). We
t
may view this space as the quotient Spec Fq [x, g(x)−1 ] /St , and denote by
t t
(2.2) ρ : Spec Fq x, g(x)−1 → Spec Fq x, g(x)−1 /St
the quotient map. Then (ρ∗ ρ∗ Qℓ )St = Qℓ , so
S
(2.3) ρ∗ ρ∗ Lχ (T t + h) t = Lχ (T t + h)
and thus
t
H∗ Spec Fq x, g(x)−1 /St , Lχ (T t + h) =
t St
(2.4) H∗ Spec Fq x, g(x)−1 /St , ρ∗ ρ∗ Lχ (T t + h) =
t St
H ∗ Spec Fq x, g(x)−1 , ρ∗ Lχ (T t + h) .
Now ρ is the map defined by factorizing a polynomial into linear terms,
so ρ∗ Lχ (T t + h) = (Lχ (T − x))⊠t . By the Künneth formula, it follows that
the stalk of Rπ∗ j! Lχ (c1 f + h + c2 T t ) at (0, 1) is
⊗t St
(2.5) H ∗ Spec Fq x, g(x)−1 , Lχ (T − x) .
Because χ is nontrivial, Lχ (T − x) has nontrivial mondromy, so

(2.6) H ∗ Spec Fq x, g(x)−1 , Lχ (T − x)
vanishes in degrees other than 1. Because it arises from a character sheaf
on a torus, Lχ has tame local monodromy, so the rank of this cohomology
group in degree 1 is the Euler characteristic of Spec Fq [x, g(x)−1 ], which is
m − 1. Thus
⊗t
(2.7) H ∗ Spec Fq x, g(x)−1 , Lχ (T − x)
is supported in degree t, where it equals the t-th tensor power of an m − 1-

dimensional vector space, with St acting the usual way, twisted by the sign
character because of the Koszul sign in the tensor product of a derived
category. Thus taking St -invariants is equivalent to taking the t-th wedge

power of this m − 1-dimensional vector space, which has dimension m−1 t .

Lemma 2.4. The complement of U in Pt × A2 is a divisor with simple
normal crossings relative to A2 away from

(2.8) (h, (c1 , c2 )) ∈ At × A2 : d gcd(c1 f + h + c2 T t , g) > t .
Proof. Because this is a purely geometric question, we may assume that g
splits completely, and let α1 , . . . , αm be its roots. Then the complement of
U is the union of the hyperplane at ∞ in Pt with the hyperplanes
(2.9) c1 f (αi ) + h(αi ) + c2 αti = 0, 1 ≤ i ≤ m.
This union has simple normal crossings if the intersection of each subset of
our hyperplanes has the expected dimension.
We first consider the case of a subset not including the hyperplane at ∞.
For any S ⊆ {1, . . . , m}, the intersection of the hyperplanes corresponding
to the elements of S is
( )
Y
t 2 t
(2.10) (h, (c1 , c2 )) ∈ A × A : (T − αi ) c1 f + h + c2 T .
i∈S
Q
Since i∈S (T − αi ) is a polynomial of degree |S|, the above has codimension
|S| for any c1 , c2 as long as |S| ≤ t, because the set of h of degree less than t
which are multiples of this polynomial has the expected dimension. On the
other hand, when |S| > t we get that

(2.11) d gcd(c1 f + h + c2 T t , g) ≥ |S| > t,
so removing these points, we obtain simple normal crossings.
Now we consider the case where we have a set of hyperplanes correspond-
ing to S ⊆ {1, . . . , m} and also the divisor at ∞. We can take coordinates
on the divisor at ∞ to be (h, (c1 , c2 )), with h nonzero and well-defined up
to scaling. In these coordinates, the equation for the intersection of any
hyperplane with the divisor at ∞ is its original equation with all terms hav-
ing degree zero in h removed. Thus the equation for the i-th hyperplane,
restricted to the divisor at ∞, is simply h(αi ) = 0, so the intersection of the
divisor at ∞ Q with the hyperplanes in S consists of (h, (c1 , c2 )) where h is
a multiple of i∈S (T − αi ). This can only happen if |S| < t, as |S| is the
degree of this polynomial and d(h) < t, but then our intersection always has
the expected dimension, so we have simple normal crossings.
Lemma 2.5. Away from a finite union of lines through the origin, the
complex Rπ∗ j! Lχ is supported in degree t with rank m−1
t .
Proof. Let l1 : A1 → A2 send c to (c, 1). We consider the vanishing cycles

at zero RΦc l1∗ j! Lχ (f + h + λT t ) of the pullback of j! Lχ (f + h + λT t ) under
l1 . Let us first check that the complement of U is a simple normal crossings
divisor everywhere in the fiber over zero. To do this, we apply Lemma 2.4
and observe that we cannot have d(gcd(c1 f + h + c2 T t , g)) > t in this fiber
as c1 = 0, c2 = 1 and

(2.12) d gcd(h + T t , g) ≤ d(h + T t ) ≤ t.
By [SGA7-II, XIII Lemma 2.1.11] it follows that the vanishing cycles
(2.13) RΦc l1∗ j! Lχ (f + h + λT t )
vanish everywhere. Hence the cohomology of the nearby and general fibers
is isomorphic. Using Lemma 2.3 to compute the cohomology of the special
fiber, it follows that the cohomology of the general fiber is supported in
degree t with rank m−1 t . By constructibility, the stalk cohomology must
have the same description at every point in some nonempty open set. By
Lemma 2.2, the same description holds for the stalks in the Gm -orbit of this
open set, which is the complement of finitely many lines.
Let l2 : A1 → A2 send c to (1, c).
Lemma 2.6. Taking vanishing cycles at zero, the complex
(2.14) RΦc l2∗ j! Lχ (f + h + λT t )
has the following properties:

• it is supported on C = (h, 0) ∈ At × A1 : d (gcd(f + h, g)) > t ;
• it is supported in degree t;
• the rank of its stalk at (h, 0) ∈ C is d(gcd(f +h,g))−1
t .
Proof. The first property follows immediately from [SGA7-II, XIII Lemma
2.1.11] and Lemma 2.4.
We note that l2∗ j! Lχ (f + h + λT t )[t + 1] is the extension by zero of a
lisse sheaf in degree −(t + 1) on a variety of dimension t + 1 and is thus
semiperverse. The dual complex is the pushforward of a lisse sheaf in degree
−(t + 1) on a variety of dimension t + 1 along an affine open immersion and
thus is semiperverse by Artin’s affine theorem. We conclude that
(2.15) l2∗ j! Lχ (f + h + λT t )[t + 1]
is a perverse sheaf. By [Ill94, Corollary 4.6], the vanishing cycles of a per-
verse sheaf are perverse up to a shift by one, so
(2.16) RΦc l2∗ j! Lχ (f + h + λT t )[t]
is perverse.
The support of the perverse sheaf above is the closed set C. Because C
does not intersect the divisor at ∞, C is finite. A perverse sheaf supported
at a finite set is necessarily a sum of skyscraper sheaves supported in degree
zero, so we obtain the second property in the statement of this lemma.
It remains to calculate the rank of the stalk at a particular point (h0 , 0) ∈

C. We do that by working locally in an étale neighborhood.
We set
(2.17) g′ = gcd(f + h0 , g), g∗ = g/g ′ ,
and factor T as the product of the torus T ′ of residue classes mod g′ and
the torus T ∗ of residue classes mod g∗ . This lets us factor Lχ as the tensor
product of Lχ′ (the pullback of a character sheaf from T ′ ) and Lχ∗ (the
pullback of a character sheaf from T ∗ ). Because f + h0 is relatively prime to
g∗ , the map U → T → T ∗ extends to a well-defined map in a neighborhood
of the point (h0 , 0), so Lχ∗ (c1 f + h + c2 T t ) extends to a lisse sheaf in a
neighborhood of (h0 , 0). Because tensoring with a lisse rank one sheaf does
not affect vanishing cycles, it suffices to calculate
(2.18) RΦc l2∗ j!′ Lχ′ (c1 f + h + c2 T t ),
where j ′ is the inclusion of the open set where gcd(c1 f + h + c2 T t , g′ ) = 1.
By changing variables, we may replace (f, h) with f ′ , h′ where f ′ = f + h0
and h′ = h − h0 - we are then tasked with calculating the vanishing cycles
RΦc l2∗ j!′ Lχ′ (c1 f ′ + h′ + c2 T t ) at zero. Having done this, we observe that f ′
is a multiple of g′ , so translation by f ′ does not affect Lχ′ (c1 f ′ + h′ + c2 T t ),
thus these vanishing cycles are the same as RΦc l2∗ j!′ Lχ′ (h′ + c2 T t ).
As d(h′ + c2 T t ) ≤ t, we can only have

(2.19) d gcd(h′ + c2 T t , g′ ) > t
if h′ + c2 T t = 0. Hence, by Lemma 2.4, the complement of the image of j ′
is a simple normal crossings divisor way from the point (0, 0). Therefore,
by [SGA7-II, XIII Lemma 2.1.11], the vanishing cycles are supported at this
point. By the vanishing cycles long exact sequence, the Euler characteristic
of the vanishing cycles complex is the difference between the Euler charac-
teristic of the generic fiber and the Euler characteristic of the special fiber.
Because the vanishing cycles sheaf is supported at a single point in a single
degree, its Euler characteristic is (−1)t times its rank at that point. We will
calculate these Euler characteristics and thereby calculate the rank.
′
By Lemma 2.5, the Euler characteristic of the generic fiber is (−1)t d(g t)−1 .
So it remains to check the Euler characteristic at the special point is zero.
Because Lχ is lisse of rank one and tame, the Euler characteristic of the
special fiber is the Euler characteristic of the space of polynomials of degree
less than t and prime to g ′ . Because this admits a free action of Gm by
scaling, its Euler characteristic is zero.
We can now prove Theorem 2.1.
Proof. We have a vanishing cycles long exact sequence
(Rπ∗ j! Lχ (c1 f + h + c2 T t ))(1,0) → (Rπ∗ j! Lχ (c1 f + h + c2 T t ))(1,η)
→ H ∗ (Pt , RΦc l2∗ j! Lχ (f + h + λT t )).
By Lemma 2.5, (Rπ∗ j! Lχ (c1 f + h + c2 T t ))(1,η) is supported in degree t

with rank m−1t . By Lemma 2.6, the complex RΦc l2∗ j! Lχ (f + h + λT t ) is
supported in degree t and at finitely many points, so the third term above
is also supported in degree t and is simply the sum of the stalks at those
points, and thus has rank
X
d (gcd(f + h, g)) − 1
(2.20) r(f, g, t) = ,
t
h∈Fq [T ]
d(h)<t
d(gcd(f +h,g))>t
again using Lemma 2.6.

We conclude that, upon suppressing c1 f + h + c2 T t for brevity, the van-
ishing cycles long exact sequence becomes
0 → (Rt π∗ j! Lχ )(1,0) → (Rt π∗ j! Lχ )(1,η)
(2.21)
→ H t (Pt , RΦc l2∗ j! Lχ ) → (Rt+1 π∗ j! Lχ )(1,0) → 0.
Thus (Rπ∗ j! Lχ (c1 f + h + c2 T t ))(1,0) is supported in degrees t and t + 1, with

rank at most m−1 t in degree t and rank bounded by r(f, g, t) in degree t+1.
By Deligne’s theorem, the absolute values of the eigenvalues of Frobenius
on the i-th cohomology group are at most q i/2 , so the absolute value of the
trace of Frobenius on cohomology is at most

m−1 t t+1
(2.22) q 2 + r(f, g, t)q 2 .
t
By the Lefschetz fixed point formula, the trace of Frobenius on cohomol-
ogy equals the sum of the trace of Frobenius on the stalks, which is
X
(2.23) χ(f + h).
h∈Fq [T ]
d(h)<t
gcd(f +h,g)=1

Finally, we check that r(f, g, t) ≤ m−1 t . To do this fix a root α of g, and
d(gcd(f +h,g))−1

note that t does not exceed the number of degree t divisors of
g, prime to T − α, that divide f + h. Each such divisor of g prime to T − α
contributes at most once to the sum in Eq. (2.20), so this sum is bounded
by the number of such divisors, which is m−1 t .
Corollary 2.7. Fix η > 0 and 0 < β < 1/2. Then for a prime power
2
q ≥ (eη −1 ) 1−2β the following holds. For a nonprincipal character χ to a
squarefree modulus g ∈ Fq [T ], f ∈ Fq [T ], and t ≥ η · d(g), we have
X
(2.24) χ(f + h) ≪ q (1−β)t
d(h)<t
as t → ∞, with the implied constant depending only on q.

Furthermore, if we have t ≤ η · d(g) ≤ t′ , then we still have
X ′
(2.25) χ(f + h) ≪ q (1−β)t .
d(h)<t
Proof. If η ≥ 1, the left hand side vanishes and the bound is trivial. Other-
wise, we apply Theorem 2.1 to the left side, taking m = d(g), to obtain
X
1/2+1 m − 1 t/2
χ(f + h) ≤ (q ) q
t
d(h)<t
(2.26) −t
m t/2 t m − t t−m t/2
≪ q ≤ q
t m m
where the last inequality follows from
m k
t m−t m X m t m − t m−k
1= + =
m m k m m
(2.27) k=0
t
m t m − t m−t
≥ .
t m m
From the Taylor series we can see that − log(1 − x) ≤ x/(1 − x) if x > 0
so (1 − x)−(1−x)/x ≤ e. Applying this to x = t/m, we get

m − t t−m
(2.28) ≤ et
m
so we obtain
X −t
t
(2.29) χ(f + h) ≪ et q t/2 .
m
d(h)<t
2
Because q ≥ (eη −1 ) 1−2β and t/m ≥ η, we have
−1
t 1
(2.30) e ≤ eη −1 ≤ q 2 −β
m
hence
−t
t
(2.31) et q t/2 ≤ q (1−β)t ,
m
as desired.
To handle the case where t ≤ η · d(g) ≤ t′ , first note that we may assume
′
t ≤ m. We observe that the left hand side of Eq. (2.31) is an increasing
function of t because its logarithm
(2.32) t log m − t log t + t + t(log q)/2
has derivative
(2.33) log m − log t + (log q)/2
which is positive in the range t ≤ m. Thus we can get a bound for the
shorter sum which is at least as good as our bound for the longer sum.
3. The Möbius Function

From now on, we will assume that the characteristic p of Fq is odd. Be-
cause of this, F×
q admits a unique quadratic character, which we denote ψ.
We use freely the basic properties of resultants (see [Jan07]) and the Jacobi
symbol (see [Ros13, Chapter 3]).
The following lemma recalls the relation between the (real valued) Jacobi
symbol and the quadratic character of a resultant.
Lemma 3.1. Let f ∈ Fq [T ], g = an T t + · · · + a0 of degree n ≥ 1. Then

f
(3.1) = ψ(an )max{d(f ),0} ψ (Res(g, f )) .
g
Proof. Fix f 6= 0, and note that both sides above are completely multiplica-
tive in g, so we assume that g is irreducible. For a root θ of g we have
d(g)−1 d(g)−1
Y i Y i q d(g) −1
(3.2) Res(g, f ) = and(f ) f θq = and(f ) f (θ)q = and(f ) f (θ) q−1
i=0 i=0
so we get the mod p congruence

q d(g) −1
d(f )
ψ (Res(g, f )) = ψ(an ) ψ f (θ) q−1
(3.3)
d(f ) q d(g) −1
d(f ) f
≡ ψ(an ) f (θ) 2 ≡ ψ(an )
g
which suffices for the lemma.
Given a D ∈ Fq [T ] we write rad(D) for the product of primes that appear
in the factorization of D, and rad1 (D) for the product of primes that appear
with odd multiplicity in the factorization of D. The derivative of D (with
respect to T ) is denoted by D ′ .
The next lemma interprets the Möbius function (on an arithmetic pro-
gression) as a Dirichlet character that ‘depends only on the derivative’.
Lemma 3.2. Let m, k ≥ 0, d ≥ 1 be integers with k 6= d+m. For M ∈ Mm ,
g ∈ Md , and a ∈ Mk coprime to M , define the polynomials
a ′ rad(D) rad1 (D)
(3.4) D = M 2 g + , E= , E1 = .
M gcd (M, rad(D)) gcd (M, rad1 (D))
Then
(3.5) µ(a + gM ) = S · χ(w + g)
where w = wa,M,g′ ∈ Fq [T ], χ = χa,M,g′ is a (real) multiplicative character
mod E with conductor E1 , and
(3.6) S = Sd,a,M,g′ ∈ {0, 1, −1}
with S = 0 if and only if D = 0.
Proof. Pellet’s formula (see [Con05, Lemma 4.1]) gives
(3.7) µ(a + gM ) = (−1)d(a+gM ) ψ (Disc(a + gM ))

and since d(a + gM ) = max{k, d + m}, we see that (−1)d(a+gM ) can be
absorbed into S. Our assumption that k 6= d + m implies a + gM is monic,
so ψ (Disc(a + gM )) equals, up to a sign that S absorbs,

(3.8) ψ Res(a + gM, a′ + g ′ M + gM ′ ) .
By Lemma 3.1, and the fact that a + gM is monic, the above equals
′
a + g′ M + gM ′
(3.9)
a + gM
and since gcd(a, M ) = 1 by multiplicativity, this equals
′ −1
a M + g ′ M 2 + gM M ′ M
(3.10) .
a + gM a + gM

M
Using quadratic reciprocity, we can absorb a+gM into S. Subtracting
M ′ (a + gM ) from the numerator of the first Jacobi symbol above, we get
(in the notation of equation (3.4))

D
(3.11)
a + gM
and set w = 0, S = 0 in case D = 0. Otherwise (if D 6= 0) we apply
quadratic reciprocity once again to obtain

a + gM
(3.12)
D
up to a sign that goes into S.
We write ξ(a + gM ) for the Jacobi symbol above, so that ξ is a multi-
plicative character mod rad(D) with conductor rad1 (D). Since rad(D) is
squarefree, we see that M is coprime to E (from equation (3.4)). There-
fore, by the Chinese remainder theorem, there exists a unique decomposition
ξ = ξE ξN to characters mod E and N = gcd (M, rad(D)) respectively. In
this notation, our Jacobi symbol equals ξE (a + gM )ξN (a + gM ) and the
second factor is simply ξN (a), so we immerse it in S. Taking M ∈ Fq [T ]
with M M ≡ 1 (E) we can write
(3.13) ξE (a + gM ) = ξE (aM + g)ξE (M )
and conclude by setting w = aM , χ = ξE , and dumping ξE (M ) into S.
Remark 3.3. With notation as in Lemma 3.2, suppose that χ is principal.
We can then write D = AB 2 for some A, B ∈ Fq [T ] with A | M , simply by
taking A = rad1 (D) since principality gives E1 = 1.
4. Linear forms in Möbius

Proposition 4.1. Let p be a prime, let q be a power of p, let m, d be non-
negative integers, let M ∈ Mm be squarefree, and let a ∈ Fq [T ]. Then
# {h : h = g′ for some g ∈ Md , h ≡ a mod M }
n j ko
− min m, d−1
≤q p
.
# {g′ : g ∈ Md }
Proof. Let 0 ≤ j ≤ p − 1 be the unique integer congruent to d mod p. Any
g ∈ Md can then be uniquely expressed as
p−1
X
d−i
(4.1) g= T i gip , gj ∈ M d−j , d(gi ) ≤ for i 6= j.
p p
i=0
For the derivative we then have

p−1
X
(4.2) g′ = iT i−1 gip
i=1
so we set ai = iT i−1 for 1 ≤ i ≤ p − 1, and consider the congruence

p−1
X
(4.3) ai gip ≡ a mod M.
i=1
Since M is squarefree, we can uniquely pick b, b1 , . . . , bp−1 ∈ Fq [T ] with

(4.4) bp ≡ a mod M, bpi ≡ ai mod M, 1 ≤ i ≤ p − 1,
so the congruence above becomes
p−1
X
(4.5) g1 ≡ b − bi gi mod M
i=2
and from this restriction on g1 the proposition follows.
The following technical proposition follows from the arguments of [BGP92,

Page 371] or [CG07, Section 9] (see also [BZ02]), which obtain stronger state-
ments over Z in place of Fq [T ].
Proposition 4.2. Fix α, ǫ > 0, and a prime power q. Then for integers
d, m, k ≥ 0 with d ≥ ǫ(m + k), and M ∈ Mm , A ∈ Mk , a ∈ Fq [T ], we have
1
# g ∈ Fq [T ] : d(g) < d, a + gM = λAB 2 , λ ∈ Fq , B ∈ Fq [T ] ≪ q ( 2 +α)d
as d → ∞, with the implied constant depending only on ǫ, α and q.
We can now prove Theorem 1.3. We first prove the “generic” special case
where the derivatives of certain parameters are distinct, and then prove the
general case.
Proposition 4.3. Fix ǫ, δ > 0, 0 < β < 1/2, and a positive integer n. Let
q be a power of an odd prime p such that
  2
1−2β
 pne 
(4.6) q> n o .
ǫ ǫδ
min ǫ+2 , ǫ+δ
Then for nonnegative integers d, m1 , . . . , mn , k1 , . . . , kn with
(4.7) d ≥ max{ǫm1 , . . . , ǫmn , δk1 , . . . , δkn }, ki 6= d + mi , 1 ≤ i ≤ n,
and pairs (ai , Mi ) ∈ Mki × Mmi for 1 ≤ i ≤ n such that the derivatives
′
ai
Mi are all distinct, we have
n
X Y 1− βp
(4.8) µ (ai + gMi ) ≪ |Md |
g∈Md i=1
as d → ∞, with the implied constant depending only on β, ǫ, δ, n and q.

Remark 4.4. Note that the statement of this proposition remains meaningful
even if ǫ and δ are very large, though it is at its strongest when ǫ and δ are
small.
Proof. Let us first assume that for every 1 ≤ i ≤ n we have gcd(ai , Mi ) = 1.
We say that g1 , g2 ∈ Md are equivalent if g1′ = g2′ , and let R be a complete
set of representatives of equivalence classes. So for each g ∈ Md there exists
a unique r ∈ R such that (g − r)′ = 0, and therefore also a unique s ∈ Fq [T ]
such that g − r = sp . We can thus write our sum as
X X Y n
d
(4.9) µ (ai + (r + sp )Mi ) , t = .
p
r∈R d(s)<t i=1
By Lemma 3.2 (the notation of which is used throughout), our sum equals
XY n X Y n
(i)
(4.10) Sr χ(i)
r wr(i) + sp
r∈R i=1 d(s)<t i=1
(i) (i)
with χr a character to a squarefree modulus Er (defined in Lemma 3.2
(i) (i)
using Dr ). Hence, there exist fr ∈ Fq [T ] with
p
(4.11) fr(i) ≡ wr(i) mod Er(i) , 1 ≤ i ≤ n,
(i)
so using the fact that χr is real, we see that our inner sum equals
X Y n
(4.12) χ(i)
r f (i)
r + s .
d(s)<t i=1
For 1 ≤ i < j ≤ n we set

(4.13) Gr(i,j) = gcd Er(i) , Er(j) , Ur = lcmi,j Gr(i,j) , ℓr = d(Ur ),
and claim that, for every integer ℓ ≥ 0 and γ > 0, we have

d 1− p1 +γ −min d−1 ,ℓ
(4.14) # {r ∈ R : ℓr ≥ ℓ} ≪ q p
, d → ∞.
(i,j)
To do this, first note that, for any 1 ≤ i < j ≤ n , since Gr divides
(i) (j) (i) (j)
Er , and divides Er , it also divides Dr , and divides Dr so therefore it
divides the polynomial
′
2 (i) 2 (j) 2 2 ai 2 2 aj ′
(4.15) Mj Dr − Mi Dr = Mi Mj − Mi Mj
Mi Mj
which is nonzero by our initial assumption. The degree of the above poly-
nomial is at most d/δ + 3d/ǫ, so by the divisor bound (see [IK04, Equation
(i,j)
1.81]), the polynomial
Gr attains ≪ q 2γd/(n(n−1)) values, for any γ > 0.
(i,j)
Hence, the tuple Gr attains ≪ q γd values. For each possible
1≤i<j≤n
(i,j) (i,j)
tuple G∗ , we can recover the residue class of r ′ mod G∗ from the con-
gruence
(4.16) Mj2 r ′ ≡ Mj′ aj − a′j Mj mod Gr(i,j) .
(i,j) (j)
and the fact that Gr is prime to Mj (because Dr is prime to Mj , by
definition). Combining these for all i, j we can also recover the residue class
(i,j)
of r ′ mod the least common multiple U∗ of G∗ .
Proposition 4.1 tells us that for any α ∈ Fq [T ] we have
n j ko
′ − min d(U∗ ), d−1 d 1− p1 −min d−1 ,ℓ
#{r ∈ R : r ≡ α mod U∗ } ≤ q p
#R ≪ q p
since d(U∗ ) ≥ ℓ, so our claim is established.

Next, we observe from Eq. (4.14)
that the contribution of those r ∈ R
d−1 d 1− p1 +γ
with ℓr ≥ p to Eq. (4.9) is ≪ q and thus can be ignored as we
can choose γ small enough that 1 − + γ ≤ 1 − βp . So we may assume that
1
p
ℓr < d−1
p .
We further set
(i)
(4.17) E b(i) = Er , 1 ≤ i ≤ n,
e(i) = gcd E (i) , Ur , E
r r r
er(i)
E
er(i) , E
and note that gcd(E br(i) ) = 1 as Er(i) is squarefree. The Chinese remain-
der theorem then gives a unique decomposition
(4.18) χ(i)
r =χe(i)
r χ b(i)
r , 1 ≤ i ≤ n,
er(i) and E
to characters mod E br(i) respectively. In this notation, our sum reads
X Y n Y n
(i) (i)
(4.19) er fr + s
χ b(i)
χ r fr(i) + s
d(s)<t i=1 i=1
so splitting according to the residue class u of s mod Ur we get

X Y n X Y n
(4.20) e(i)
χ r f (i)
r + u χb(i)
r f (i)
r + u + hU r .
d(u)<ℓr i=1 d(h)<t−ℓr i=1
Since gcd(Eer(i) , E
br(i) ) = 1, we get that gcd(Ur , E br(i) ) = 1 for 1 ≤ i ≤ n.
(i) (i) br(i) . Summing trivially
Hence, there exist Vr ∈ Fq [T ] with Ur Vr ≡ 1 mod E
over u, we may thus consider
Yn X Y n
(i)
(4.21) χbr (Ur ) br(i) fr(i) Vr(i) + uVr(i) + h .
χ
i=1 d(h)<t−ℓr i=1
er , E (i) (i)
br ) = 1 we moreover conclude that {E
br }n are pairwise (i)
From gcd(E i=1
coprime. The Chinese remainder theorem then gives an fr,u ∈ Fq [T ] with
(4.22) b(i) ,
fr,u ≡ fr(i) Vr(i) + uVr(i) mod E 1 ≤ i ≤ n,
r
(1) (n) br , the sum
br · · · E (1) (n)
so defining the character χr = χ
br · · · χ
br , mod Er = E
above becomes
X
(4.23) χr (fr,u + h).
d(h)<t−ℓr
We have
n o
ǫ ǫδ
t t min ǫ+2 , ǫ+δ
(4.24) ≥ = =·· η
d(Er ) n max{d + 2d/ǫ, d/δ + d/ǫ} pn
′
so if χr is nonprincipal, Corollary 2.7 bounds the sum above by ≪ q (1−β )t ,
for some β ′ > β + pγ, with γ > 0 arbitrarily small. Because we have
  2
1−2β
pne
(4.25) q> n o
ǫ ǫδ
min ǫ+2 , ǫ+δ
by assumption, using our definition of η this can be writeen as
2
(4.26) q > eη −1 1−2β
and therefore
2
(4.27) q > eη −1 1−2β ′
for any sufficiently small choice of γ.

Hence, those r ∈ R for which ℓr = ℓ and χr is nonprincipal, contribute
′
≪ q (1−β )t q ℓ individually, so the total contribution to our initial sum is ≪

1
(1−β ′ )t+ℓ d 1− p +γ −ℓ
1
(1−β)t d 1− p γd−(β ′ −β)t d 1− βp d(γ−(β ′ −β)/p)
q q =q q q =q q .
The increase from summing over the possible values of ℓ is linear in d and
′
thus can be bounded by the exponential q d((β −β)/p−γ)
, so the contribution
d 1− βp
of all the terms where χ is nonprincipal is ≪ q .
Let r ∈ R for which χr is principal. Pairwise coprimality implies that

(1) (1) (1)
χ
br is principal as well, so the conductor of χr divides that of χ er , and
the latter divides Ur . Thus, for some λ ∈ Fq and monic B ∈ Fq [T ] we have

M12 r ′ + a′1 M1 − a1 M1′ = Dr(1) = rad1 Dr(1) λB 2 = λAAB 2 , A | M1 , A | Ur
(1)
where A is the greatest common divisor of M1 and rad1 (Dr ), and A is the
(1)
conductor of χr . Since Ur divides the nonzero polynomial Q defined to be

Y
2 2 ai ′ aj ′
(4.28) Mi Mj − ,
Mi Mj
1≤i<j≤n
it follows that A | Q as well, so Proposition 4.2 ensures that the number of

r for which the equation above holds is
(4.29) ≪ d2 (M1 Q)q (1/2+ζ)d .
The divisor bound then allows us to neglect those r for which
χr is principal,

1 1 β 1 1 1
as long as 2 + ζ + p < 1 − p , which is alright as β < 2 ≤ p 2 − p so we
can choose ζ small enough that this inequality holds.
Let us now handle the case when ai and Mi are not coprime. Set
(4.30) Hi = gcd(ai , Mi ), 1 ≤ i ≤ n.
If some Hi is not squarefree, the sum vanishes and the bound is trivial.
Otherwise, we have the identity

µ ai + g Mi µ(Hi ) if gcd ai + g Mi , Hi = 1
Hi Hi Hi Hi
(4.31) µ(ai + gMi ) =
0 ai Mi
if gcd Hi + g Hi , Hi 6= 1.
Q
Thus we can write the sum as the constant factor ni=1 µ(Hi ) times a
similar sum, except
that the degrees
of ai and Mi are reduced and the
ai Mi
terms where gcd Hi + g Hi , Hi 6= 1 are removed. The degree reduction
preserves
our inequalities
and so is no trouble. Removing the terms with
ai Mi
gcd Hi + g Hi , Hi 6= 1 amounts to removing those g which lie in a par-
ticular
P residue class modulo each prime factor of Hi , for a total of at most
Γ = ni=1 ω(Hi ) residue classes.
We perform the same argument to this restricted sum. The only change
that occurs is when we write g = r + sp , we must assume that s avoids
a corresponding set of residue classes modulo these primes. By inclusion-
exclusion, we can write a Dirichlet character sum avoiding Γ residue classes
as an alternating sum of Dirichlet character sums in at most 2Γ residue
classes, and hence as an alternating sum of at most 2Γ shorter Dirichlet
character sums. Because the sums over each residue class are shorter, we
can get the same bound for them by Corollary 2.7. Thus our final bound
for this case is worse by a factor of
Pn
(4.32) 2Γ = q i=1 o(d(Hi ))
= q o(d) .
We can absorb this into our bound by slightly increasing β so that it still
satisfies the strict inequality (4.6).
Theorem 4.5. Fix ǫ, δ > 0, 0 < β < 1/2, and a positive integer n. Let q
be a power of an odd prime p such that
  2
1−2β
 pne 
(4.33) q> n o .
ǫ ǫδ
min ǫ+2 , ǫ+δ

and pairs (ai , Mi ) ∈ Mki × Mmi for 1 ≤ i ≤ n with ai /Mi distinct, we have
n
X Y 1− βp
(4.35) µ (ai + gMi ) ≪ |Md |
g∈Md i=1
as d → ∞, with the implied constant depending only on ǫ, δ, β, n and q.

Proof. Our initial assumption is that there are no coincidences among (ai /Mi ),
for 1 ≤ i ≤ n, so we can find a prime P not dividing
n
Y Y
(4.36) Mi ai Mj − aj Mi ,
i=1 1≤i<j≤n
with d(P ) = o(d). Let t′ = d(P ). Splitting our initial sum according to the
residue class z of g mod P we get
X X n
Y
(4.37) µ (ai + zMi + f Mi P ) .
d(z)<t′ f ∈Md−t′ i=1
We can bound the sums over residue classes by applying Proposition 4.3.
Indeed, if for some 1 ≤ i < j ≤ n we have

ai + zMi ′ aj + zMj ′
(4.38) =
Mi P Mj P
then we get that ai Mj ≡ aj Mi mod P , a contradiction. Because the
′
length of the sum in this case is q d−t , we obtain a savings in each term
′
of q (d−t )β/2p from Proposition 4.3. To obtain our desired savings of q dβ/2p ,
we must choose β + o(1) instead of β in the statement of Proposition 4.3 .
Similarly to ensure that d − t′ ≥ max{ǫm1 , . . . , ǫmn , δk1 , . . . , δkn } we must
choose ǫ − o(1) and δ − o(1). However because the inequality from Eq. (4.33)
is strict, we may increase β by o(1) and reduce ǫ and δ by o(1) in such a
way that this inequality is still satisfied.
We prove two corollaries that give weaker results under conditions that
are simpler to state.
Corollary 4.6. Fix ǫ, δ > 0 and a positive integer n. Let q be a power of

an odd prime p such that

2 2 2 2 1 1 2
(4.39) q > p n e max 1 + , + .
ǫ ǫ δ
and distinct coprime pairs (ai , Mi ) ∈ Mki × Mmi for 1 ≤ i ≤ n, we have
n
X Y
(4.41) µ (ai + gMi ) = o (|Md |)
g∈Md i=1
as d → ∞ for fixed ǫ, δ, n, q.
Proof. The assumed lower bound on q is equivalent to
 2
pne
(4.42) q> n o .
ǫ ǫδ
min ǫ+2 , ǫ+δ
If q satisfies this inequality, we can take β small enough that the inequality
  2
1−2β
 pne 
(4.43) q> n o
ǫ ǫδ
min ǫ+2 , ǫ+δ
holds, and then apply Theorem 4.5.

Corollary 4.7. Let q be a power of an odd prime p such that
(4.44) q > p2 n2 e2 ,
and let (ai , Mi ) ∈ Mki × Mmi be distinct coprime pairs for 1 ≤ i ≤ n. Then
n
X Y
(4.45) µ (ai + gMi ) = o (|Md |)
g∈Md i=1
as d → ∞ for fixed q, n, ai , Mi .
Proof. We can take ǫ, δ large enough that

2 1 1 2
(4.46) q > p2 n2 e2 max 1 + , + .
ǫ ǫ δ
For this ǫ and δ, the inequalities
will be satisfied for all d sufficiently large. We can then apply Corollary 4.6
to deduce the claim.
5. Level of distribution
The following will be obtained using the techniques of [FM98] and [FKM14]
in the appendix.
Theorem 5.1. Fix an odd prime power q. Then for any θ > 0, for non-
negative integers d, m with d ≤ m, squarefree M ∈ Mm , and an additive
character ψ mod M , we have
X 3 25
(5.1) µ(g)ψ(g) ≪ q ( 16 +θ)m+ 32 d
g∈Md
(g,M )=1
as d → ∞, with the implied constant depending only on q and θ.

The following proposition allows us to identify the main term in sums of
von Mangoldt in arithmetic progressions.
Proposition 5.2. Fix a prime power q. For nonnegative integers d, m, and
a squarefree M ∈ Mm we have
d
X X qm
(5.2) kq −k µ(A) = − + q o(m+d)−d .
ϕ(M )
k=1 A∈Mk
(A,M )=1
Proof. The left hand side above is the sum of the first d coefficients of the
power series
d Y
∞
d X −k k X
u q u µ(A) = u 1 − ud(P ) |P |−1
du du
k=1 A∈Mk P ∤M
(A,M )=1
 
d Y −1
= u (1 − u) 1 − ud(P ) |P |−1 .
du
P |M
Summing all the coefficients of a power series, evaluates it at u = 1.

Hence, the main term comes from the equality
Y −1
d
u ((1 − u)F (u)) (1) = −F (1), F (u) = 1 − ud(P ) |P |−1 .
du
P |M
The coefficients of degree greater than d contribute to the error term. To

bound the sum of these coefficients, we can write the degree k coefficient as
I
(1 − u)F (u)
(5.3) k
uk+1
|u|=r
for r < q, getting a bound of

(5.4) (1 + r)r −k−1 max F (u).
|u|=r
As long as the above maximum is subexponential in m for all r < q, the

expression above will be q o(m) (1 + r)r −k−1 , so the sum of the coefficients of
degree greater than d is
X
(5.5) q o(m) k(1 + r)r −k−1 = q o(m) (r − o(1))−d .
k>d
Taking r arbitrarily close to q, the above is q o(m) (q − o(1))d = q o(m+d)−d .

The value of F (u) is indeed subexponential in m, because M has o(m)
prime divisors and each contributes at most
d(P ) !−1
r r −1
(5.6) 1− ≤ 1− .
q q
In the following we deduce, from our results on the Möbius function, a

level of distribution beyond 1/2 for primes in arithmetic progressions to
squarefree moduli. We shall use the convolution identity Λ = −1 ∗ (µ · deg)
which for f ∈ Fq [T ]+ of degree d ≥ 0 says that
d
X X X
(5.7) Λ(f ) = − k µ(A).
k=1 A∈Mk B∈Md−k
AB=f
Corollary 5.3. For any 0 < ω < 1/32, for any odd prime p and power q
2
50
of p such that q > p2 e2 1 + 1−32ω , the following holds. For nonnegative
integers d, m with d ≥ (1 − ω)m,, squarefree M ∈ Mm , and a ∈ Fq [T ] with
d(a) < d + m and coprime to M , we have
X q d+m
(5.8) Λ(a + gM ) = + Eq
ϕ(M )
g∈Md
as d → ∞, with a power saving error term Eq .
Proof. We can assume ℓ = m, and use Eq. (5.7) to write our sum as
X d+m
X X X
(5.9) Λ(f ) = − k µ(A) 1
f ∈Mm+d k=1 A∈Mk B∈Mm+d−k
f ≡a mod M (A,M )=1 AB≡a mod M
so by Proposition 5.2 the range k ≤ d contributes the main term.

The (absolute value of the) contribution of any k > d is

X X

q d−k µ(A) ψ aA ψ(T m+d−k ) ≤

A∈Mk ψ : Fq [T ]/M →C×
(A,M )=1 ψ(f )=0 if d(f )<m+d−k
(5.10)

X X

q d−k µ(A)ψ aA

ψ : Fq [T ]/M →C×
A∈Mk
(A,M )=1

ψ(f )=0 if d(f )<m+d−k
so by Theorem 5.1 we get

3 25 3 25
(5.11) ≪ q d−k q k−d q ( 16 +θ)m+ 32 k = q ( 16 +θ)m+ 32 k
which gives a power saving as long as

3 25
(5.12) + 2θ m + k < d.
16 32
The (absolute value of the) contribution of any other k is at most

X
X X X

(5.13) µ(A) = µ(b + gM )

B∈Mm+d−k A∈Mk B∈Mm+d−k g∈Mk−m
(B,M )=1 AB≡a mod M (B,M )=1
where b is the unique monic polynomial of degree m congruent to aB −1

modulo m. We apply Theorem 4.5 with some fixed β > 0 and with

32 1
(5.14) ǫ=δ= − ω − 2θ .
25 32
Then we because (5.12) does not hold, we have

32 31 32 31
k−m≥ d− + 2θ m ≥ (1 − ω)m − + 2θ m = ǫm
25 32 25 32
so the conditions of Theorem 4.5 are satisfied as long as
!! 2
1−2β
25 1
(5.15) q > pe 1 + 1 .
16 32 − ω − 2θ
Because we may take β and θ arbitrarily small, it suffices to have
!2 2
2 2 25 1 2 2 50
(5.16) q >p e 1+ 1 = p e 1 +
16 32 −ω 1 − 32ω
Summation over k gives only an extra logarithmic factor, so it preserves
our power savings.
6. Twins
6.1. Chowla sums over primes. We establish cancellation in Möbius au-
tocorrelation over primes.
Corollary 6.1. Fix ǫ̃, δ̃ > 0, 0 < α < 1, 0 < β < 1/2, and a positive integer
n. Let q be a power of an odd prime p such that
!! 2
2 + 2α + 4ǫ̃−1 1 + α + 2ǫ̃−1 + 2δ̃−1 1−2β
q > p(n + 1)e max 1 + , .

1−α 1−α
Take nonnegative integers d, m1 , . . . , mn , k1 , . . . , kn with
(6.1) d ≥ max{ǫ̃m1 , . . . , ǫ̃mn , δ̃k1 , . . . , δ̃kn }, ki 6= d + mi , 1 ≤ i ≤ n,
and distinct coprime pairs (ai , Mi ) ∈ Mki × Mmi for 1 ≤ i ≤ n.
Furthermore let m, k be nonnegative integers with m ≤ αd, k < m + d,
let M ∈ Mm , and let a ∈ Mk with (a, M ) distinct from (ai , Mi ) for every
1 ≤ i ≤ n. Then
X Yn
1− β(1−α)
(6.2) Λ(a + gM ) µ (ai + gMi ) ≪ |Md | 2p
g∈Md i=1
as d → ∞, with the implied constant depending only on ǫ, δ, α, n and q.

Proof. We can assume (a, M ) = 1, set z = d+m
2 , and use Eq. (5.7) to write
X z X d+m−z−1
X X
a + gM
Λ(a+gM ) = − b µ(B)− (d+m−c) µ .
C
b=1 B∈Mb c=0 C∈Mc
B|a+gM C|a+gM
For every B above, taking a monic N = N (B) ∈ Fq [T ] with

(6.3) M N ≡ −a mod B, d(N ) 6= ki − mi , d(N ) ≤ b + n,
and writing g = N + hB with h ∈ Md−b , we see that the sum over b
contributes
⌊ d+m
X2 ⌋ X X Y n
(6.4) b µ(B) µ (ai + N Mi + hBMi )
b=1 B∈Mb h∈Md−b i=1
so we can apply Theorem 4.5 to the innermost sum, taking

1−α d−m d − d+m d−b
(6.5) ǫ= ≤ = d+m 2 ≤
1 + α + 2ǫ̃−1 d + m + 2mi 2 + mi
b + mi
and

1−α 1−α d−b d−b
δ = min , ≤ min ,
2δ̃−1 1 + α + 2ǫ̃−1 ki b + mi
(6.6)
d−b d−b
≤ min , + o(1) .
deg ai deg N Mi
To obtain the condition on q, note that, when calculating max(1+ 2ǫ , 1ǫ + 1δ ),

we can treat δ as 21−α
δ̃−1
, because if the other term is smaller, then 1ǫ + 1δ is
dominated by 1 + 2ǫ anyways.
Taking L ∈ Fq [T ]+ with
(6.7) M L ≡ −a mod C, d(L) 6= ki − mi , d(L) 6= d, d(L) ≤ c + n + 1,
and writing g = L + hC with h ∈ Md−c , we see that the sum over c
contributes
d+m−z−1
X X X Yn
a + LM
(d+m−c) µ + hM µ (ai + LMi + hCMi )
C
c=0 C∈Mc h∈Md−c i=1
and we again apply Theorem 4.5 to the innermost sum, with the same values
of ǫ and δ. Everything is the same as before, with c replaced by b, except
for two things.
(1) We can no longer use the inequality b ≤ m+d2 , but rather the slightly
m+d+1
weaker inequality c ≤ .
2
(2) The term µ a+LM
C + hM appears, which means we must check that

a + LM
(6.8) (d − c) ≥ ǫm, (d − c) ≥ ǫd .
C
However (1) is no difficulty as we may assume d sufficiently large and
perturb the parameters ǫ and δ slightly to insure the inequalities of Theo-
rem 4.5 still hold. For this reason we will assume c ≤ m+d
2 while handling
(2) as well.
To check that (d−c) ≥ ǫm we observe that, because c ≤ m+d
2 and d < αm,
we have d − c ≥ d−m 2 , and thus d−c
m ≥ 1−α
2α ≥ 1−α
1+α ≥ ǫ.
To check that

a + LM
(6.9) d ≤ max(m + d − c, m + n) ≤ δ−1 (d − c)
C
we observe
1−α 2(d − m) 1 1 d−c
δ≤ = = m ≤ m = .
1+α d+m (d−m)/2 +1 d−c+1 m+d−c

6.2. Singular series. For nonzero a ∈ Fq [T ] define

Y −1 Y
(6.10) Sq (a) = 1 − |P |−1 1 − (|P | − 1)−2 .
P |a P ∤a
The following propositon allows us to identify the main term in our twin
prime number theorem. An analogous result over the integers is proved in
[GY03, Lemma 2.1].
Proposition 6.2. Fix a prime power q. Then for an integer n ≥ 1 and a

nonzero a ∈ Fq [T ] we have
n
X X µ(M )
(6.11) k = −Sq (a) + (q − 1)o(n+d(a))−n .
ϕ(M )
k=1 M ∈Mk
(M,a)=1
Proof. We are interested in the sum of the first n coefficients of

d Y
∞
d X k X µ(M )
Z(u) = u u =u 1 − ud(P ) (|P | − 1)−1
du ϕ(M ) du
k=1 M ∈Mk P ∤a
(6.12) (M,a)=1
d
=u ((1 − u)G(u))
du
where G(u) is
Y −1 Y −1
1 − ud(P ) |P |−1 1 − ud(P ) |P |−1 1 − ud(P ) (|P | − 1)−1 .
P |a P ∤a
The sum of all coefficients equals

(6.13) Z(1) = −G(1) = −Sq (a)
because
−1
(6.14) 1 − |P |−1 1 − (|P | − 1)−1 = 1 − (|P | − 1)−2 .
As in Proposition 5.2, to prove the bound for the error term it suffices
to prove that G(u) is bounded subexponentially in a for u on each circle of
radius < q − 1.
Note that
−1
1 − ud(P ) |P |−1 1 − ud(P ) (|P | − 1)−1 =
(6.15)
ud(P )
1−
|P | (|P | − 1) 1 − ud(P ) |P |−1
so G(u) can be rewritten as
!
Y ud(P ) Y −1
1− 1 − ud(P ) (|P | − 1)−1 .
P
|P | (|P | − 1) 1 − ud(P ) |P |−1
P |a
The first product above is independent of a and converges on the disc where
|u| < q. On a circle of radius r, the value of each term in the second product
is at most
!−1 −1
r d(P ) r
(6.16) 1 − d(P ) ≤ 1−
q −1 q−1
and thus is bounded, and the number of terms is o(d(a)), so the product is
subexponential in a.

6.3. Hardy-Littlewood.
Theorem 6.3. For every odd prime number p, and power q of p with
(6.17) q > 685090p2 ,
there exists λ > 0 such that the following holds. For nonnegative integers
d > ℓ, and a ∈ Mℓ we have
X
(6.18) Λ(f )Λ(f + a) = Sq (a)q d + O q (1−λ)d
f ∈Md
as d → ∞, with the implied constant depending only on q.

Proof. Using Eq. (5.7) our sum becomes
d
X X X
(6.19) − k µ(M ) Λ(a + N M ).
k=0 M ∈Mk N ∈Md−k
The appearance of the factor µ(M ) allows us to consider only squarefree M ,

so by Corollary 5.3, the contribution of the range 0 ≤ k ≤ d/(2 − ω) is
d/(2−ω)
X X µ(M ) d
(6.20) − qd k + O dq 2−ω Eq
ϕ(M )
k=0 M ∈Mk
(a,M )=1
which equals by Proposition 6.2 to

d d
d ′ d
(6.21) q Sq (a) + O dq 2−ω Eq + q Eq .
2−ω
The error term is of power savings size.
In the other range, we need to prove a power savings bound for
d
X X X
(6.22) k Λ(a + M N )µ(M )
k>d/(2−ω) N ∈Md−k M ∈Mk
and this is done by applying Corollary 6.1 to the innermost sum with
d−k
(6.23) n = 1, ǫ̃ = ∞, δ̃ = ∞, α = 1 − ω >
k
and β > 0 but very small. This requires
2 2
2 + 2 − 2ω 4
(6.24) q > 2pe 1 + = 2pe −1
ω ω
and Corollary 5.3 requires
2
50
(6.25) q > p2 e2 1 +
1 − 32ω
so the optimal bound is obtained by solving

4 50
(6.26) 2 −1 =1+
ω 1 − 32ω
whose solution is
q
30803
103 − 3
(6.27) ω= = .0261 . . .
64
which satisfies
2
4
(6.28) 2e −1 < 685090.
ω

7. Results for Small q

We prove Theorem 1.11.
Proof. Set d(f ) = d. We need to show that by suitably changing the coeffi-
cients of f in degree at most ηd, one can arrive at a polynomial with a given
(nonzero) Möbius value.
Let c < ηd be the largest even integer not divisible by 3. Note that
(7.1) c ≥ ηd − 4.
We take the coefficient of Tc
in f to be 1, and the coefficient of T k to be 0
for every k < c that is not divisible by 3. Hence, it is enough to show that

(7.2) 1, −1 ∈ µ(f + b3 ) : b ∈ M⌊c/3⌋ .
By Lemma 3.2 with M = 1, a = f , and g = b3 , our set equals

(7.3) S · χ(w + b3 ) : b ∈ M⌊c/3⌋
and since the highest power of T that divides (f + b3 )′ = f ′ is T c−1 , we
conclude that S = ±1 and from Remark 3.3 that χ is a nonprincipal charac-
ter (to a squarefree modulus E). Arguing as in Eq. (4.11) to ’extract third
roots’, we are thus led to consider

(7.4) χ(w
e + b) : b ∈ M⌊c/3⌋ .
From Lemma 3.2 we further conclude that
(7.5) d(E) ≤ d − c + 1 ≤ (1 − η)d + 5,
and on the other hand
c ηd − 4 η
(7.6) ⌊c/3⌋ ≥ −1≥ − 1 ≥ d − 3,
3 3 3
so combining the two we get
η
⌊c/3⌋ 3d − 3
(7.7) ≥ .
d(E) (1 − η)d + 5
Since we have assumed that 3/7 < η < 1, the right hand side of the above
tends to a quantity greater than 1/4 as d → ∞. Consequently, we can use
the (function field version of the) Burgess bound (as stated for instance in
[Bur63, Theorem 2]) to show that 1 and −1 belong to the set above. Such
a version is obtained in [Hsu99].
Now we prove Theorem 1.12.

Proof. Set k = d(P ). For positive integers d, n, we seek cancellation in
X
(7.8) µ(a + gP n )
g∈Md
where a ∈ Fq [T ]+ satisfies d(a) < nk and a ≡ 1 mod P n−2 . We assume first

that 3 | n, and follow the proof of Proposition 4.3 up to Eq. (4.23) getting
X
(7.9) χr (f + h)
d(h)<t
with χr a character mod Er . If χr is principal, then by Remark 3.3 we have

(7.10) P 2n r ′ + a′ P n = Dr(1) = AB 2 , A, B ∈ Fq [T ], A | P n , P ∤ B
and since P n−3 | a′ , we conclude that
a′ eB
e 2 , A,
eB e ∈ Fq [T ], A
e | P.
(7.11) P 3r′ + =A
P n−3
There are ≪ q d/2 choices of r ∈ R satisfying the above, so those can be
neglected.
For r ∈ R with χr nonprincipal, we note that

2n ′ ′ n
a′
(7.12) d(Er ) ≤ d rad P r + a P ≤ d P 3 r′ + +k
P n−3
so for large enough d we have t/d(Er ) > 1/4, hence cancellation in Eq. (7.9)
is guaranteed by Burgess.
Suppose now that 3 | n + β for some β ∈ {1, 2}, and write
X X X
(7.13) µ(1 + gP n ) = µ(1 + g0 P n + g1 P n+β )
g∈Md d(g0 )<βk g1 ∈Md−βk
for d ≥ β. We have thus reduced to the previous case with a = 1+g0 P n .

Remark 7.1. We see from the proof that 1 is not the only residue class for
which the argument works. Also, the modulus does not have to be a power
of a fixed prime, but it has to be ‘multiplicatively close’ to a cube.
Appendix A. Orthogonality of Möbius and

inverse additive characters
We explain how a variant of the results of [FM98] carries over to function
fields and gives Theorem 5.1.
A standard strategy in the treatment of sums such as those from Eq. (5.1)
is to use a combinatorial identity for the Möbius function. Following [FM98],
we use Vaughan’s identity, which for f ∈ Fq [T ]+ gives
X X X X
(A.1) µ(f ) = − µ(g)µ(h) + µ(g)µ(h)
d(g)≤α d(h)≤β d(g)>α d(h)>β
gh|f gh|f
where summation is over monic polynomials, and α, β are nonnegative in-

tegers with max{α, β} < d(f ). For a proof (that is also valid for function
fields) see [IK04, Proposition 13.5].
By applying Vaughan’s identity as in [FM98, Section 6], we reduce our
task to bounding sums of type I:
(I)
X X
(A.2) Σk,r = γf ψ(f g)
f ∈Mk g∈Mr
(f g,M )=1
where k ≤ 18 d(M ) + 7
16 d, k + r ≤ d, |γf | ≤ 1 and sums of type II:
(II)
X X
(A.3) Σk,r = γf δg ψ(f g)
f ∈Mk g∈Mr
(f g,M )=1
7
with k ≥ 81 d(M ) + 16 d, r ≥ 7
16 d − 38 d(M ), k + r ≤ d, |δg | ≤ 1. For every
ǫ > 0, we need the bounds
3 25 1 k
Σk,r ≪ q ( 16 +ǫ)d(M )+ 32 d , Σk,r ≪ q d+ǫd(M )− 2 + q d+( 4 +ǫ)d(M )− 2
(I) (II) r
(A.4)
that are analogous to [FM98, Equation 6.4]. The bounds (A.4) then imply
(5.1) by the argument of [FM98, Section 6].
A.1. Sums of type I. Following first the bilinear shifting trick argument
of [FM98, §4], we obtain the inequality

1 X X X X
(I) −1
(A.5) Σk,r ≪ ψ af (a g + b)
V
a∈MdA f ∈Mk g∈Mr b∈MdB
(a,M )=1 (a−1 g+b,M )=1

where, in analogy with the variables A, B from [FM98, (4.8)],
3r − k k+r
(A.6) dA = , dB = −1
4 4
and
(A.7) V = # {ab : a ∈ MdA , b ∈ MdB , (a, M ) = 1} ≫ q dA +dB −ǫd(M ) .
The parameter t and the term e(−bt) from [FM98] do not appear here,
as they arise from the failure of an archimedean interval to be perfectly
invariant under a shift.
Now following the Hölder’s inequality argument from [FM98, §4.a], we

get as in [FM98, Equation 4.6], the bound
6

X X X X

ψ af (a−1 g + b) ≪

(a,M )=1 a−1 g+b∈A×

(A.8)

X
5(k+r+dA )+ǫd(m) X X
q ψ (∆b (h, s))

b1 ,...,b′3 ∈MdB h∈A d(s)≤k+dA
h+bi ∈A (s,M )=1
×
h+b′ ∈A×
i
where A = Fq [T ]/(M ),
(A.9) b = (b1 , b2 , b3 , b′1 , b′2 , b′3 ) ∈ Fq [T ]6 ,
and the function ∆b (h, s) is defined by
X3
1 1
(A.10) ∆b (h, s) = − .
(h + bi )s (h + b′i )s
i=1
Because of our conditions h + bi ∈ A× , h + b′i ∈ A× , and (s, M ) = 1, the

function ∆b is never evaluated where it is undefined. The Hölder’s inequality
argument is slightly more involved because the invertibility assumptions are
more complex in case M is not prime, but the idea is essentially the same.
As in [FM98], we complete the right hand side of Eq. (A.8) and get
X X
ψ (∆b (h, s)) =
h∈A d(s)≤k+dA
h+bi ∈A× (s,M )=1
h+b′i ∈A×
(A.11) X X X
q k+dA −d(M ) ψ (∆b (h, s) + zs) .
z∈Fq [T ] h∈A s∈A×
d(z)<d(M )−k−dA h+bi ∈A×
h+b′i ∈A×
In order to treat the innermost sum on the right hand side, we note that
3
1X 1 1
(A.12) ∆b (h, s) + zs = − + zs
s (h + bi ) (h + bi′ )
i=1
so the aforementioned innermost sum over s is a Kloosterman sum. We put

(A.13) Ab = x ∈ A : (x + b1 ) · · · (x + b′3 ) ∈ A× ,
define a function Rb : Ab → A by
3
X
1 1
(A.14) Rb (x) = − ,
x + bi x + b′i
i=1
and a Kloosterman sum

X
(A.15) S(x, z) = ψ xy −1 + zy , x ∈ A, z ∈ Fq [T ].
y∈A×
In this notation, for any z ∈ Fq [T ] we have

X X X
(A.16) ψ (∆b (h, s) + zs) = S(Rb (x), z)
h∈Ab s∈A× x∈Ab
and the following claim.

Proposition A.1. For σ ∈ S3 and

(A.17) Mσb = gcd M, b1 − b′σ(1) , b2 − b′σ(2) , b3 − b′σ(3)
we have if p > 3 the bound

X
(A.18) S(R (x), z) ≪ |M |d2 (M )4 lcmσ∈S gcd(M b , z)
b 3 σ
x∈Ab
with the implied constant depending only on q.
If p = 3, with

(A.19) b
M△ = gcd M, b1 − b2 , b2 − b3 , b′1 − b′2 , b′2 − b′3 ,
we have the bound

X 4
(A.20) S(R (x), z) ≪ |M |d2 (M ) lcm gcd(M b
, z)
b σ∈S3 ∪{△} σ
x∈Ab
with the implied constant depending only on q.
Proof. Since both the bound and the sum are multiplicative in M , it suffices
to handle the case when M is prime, where we show that

X

(A.21) S(Rb (x), m) ≤ 16|A|

x∈A\{b1 ,...,b′ }
3
unless z = 0 and either

• for some σ ∈ S3 we have bi = b′σ(i) for all 1 ≤ i ≤ 3;
• or p = 3, b1 = b2 = b3 , and b′1 = b′2 = b′3 ;
in which case we have the trivial bound

X
(A.22) S(Rb (x), 0) ≤ |A|2 .

x∈A\{b1 ,...,b′ }
3
The relevance of these conditions is that the residue of the pole of Rb at

a point x equals
(A.23) #{1 ≤ i ≤ 3 : bi = −x} − #{1 ≤ i ≤ 3 : b′i = −x}
so it is nonzero whenever these two numbers are not equal, except when
p = 3, one of these numbers is 3, and the other is zero. Hence, Rb has a
pole unless each bi is equal to some bσ(i)′ , or p = 3, all the bi are equal, and
the b′i are also all equal.
Excluding these ‘trivial’ values of b for z = 0, we get that the rational
function Rb is nonconstant and at most 6 to 1. Hence, if Rb (x) 6= 0 we get
S(Rb (x), 0) = −1, while for the values of x with Rb (x) = 0, at most 6 in
number, we have
(A.24) S(0, 0) = |A| − 1.
In total, we get

X
(A.25) S(Rb (x), 0) ≤ |A| + 6|A| = 7|A|.

x∈A\{b1 ,...,b′ }
3
Suppose from now on that z 6= 0. Note that if the rational function Rb

is constant then it necessarily vanishes identically, so we get

X

(A.26) S(0, z) ≤ |A|.

x∈A\{b1 ,...,b′ }
3
We can thus assume that Rb is nonconstant, and let Kℓ2 be the Kloosterman
sheaf defined by Katz. With this notation, our sum can be written as
X
− tr Frob|A| , (Kℓ2 )Rb (x)z =
x∈A\{b1 ,...,b′3 }
X
− tr Frob|A| , ([zRb ]∗ Kℓ2 )x =
(A.27)
x∈A\{b1 ,...,b′3 }
2
X
− (−1)i tr Frob|A| , Hci A1Fq \ {−b1 , . . . , −b′3 }, [zRb ]∗ Kℓ2
i=0
by the Grothendieck-Lefschetz fixed point formula. Because the geometric
monodromy group of Kℓ2 is SL2 , which is connected, the geometric mon-
odromy group of its pullback by any finite covering map is SL2 , which has
no monodromy coinvariants, so the cohomology groups in degree 0 and 2
vanish. As Kℓ2 is pure of weight 1, its pullback by a finite covering map
is mixed of weight at most 1, so by Deligne’s theorem, the eigenvalues of
Frob|A| on Hc1 (A1F \ {−b1 , . . . , −b′3 }, [zRb ]∗ Kℓ2 ) have absolute value at most
q
|A|. Hence, in order to bound our sum by 16|A|, it suffices to prove that the
dimension of the above cohomology group is at most 16.
By the aforementioned vanishing of cohomology in degrees 0 and 2, the
dimension of our cohomology group equals minus the Euler characteristic.
Since [zRb ]∗ Kℓ2 is lisse of rank 2 on AF1 \ {−b1 , . . . , −b′3 }, its Euler char-
q
acteristic is twice the Euler characteristic of A1F \ {−b1 , . . . , −b′3 }, which
q
is

(A.28) 2 1 − #{−b1 , . . . , −b′3 } ,
minus the sum of the Swan conductors at each singular point. Because
the rational function mRb has a zero at ∞ and a pole of order at most 1
at each bi or b′i , the Swan conductor of [mRb ]∗ Kℓ2 at ∞ vanishes and the
Swan conductor of [zRb ]∗ Kℓ2 at bi or b′i is at most 1, so the total Euler
characteristic is at least
(A.29) 2 − 3#{−b1 , . . . , −b′3 } ≥ 2 − 3 · 6 = −16.

Corollary A.2. Keeping the same notation, we have if p > 3
X X
3(r+k)

ψ (∆b (h, s)) ≪ d2 (M )5 |M | + q 4 lcmσ∈S3 Mσb .
h∈Ab s∈A×
d(s)≤k+dA
b also included in the lcm.
and, if p = 3, the same bound but with M△
Proof. Using Eq. (A.11), Eq. (A.16), and Proposition A.1 we get a bound
of
X
k+dA 4
(A.30) ≪q d2 (M ) lcmσ∈S3 gcd Mσb , z
z∈Fq [T ]
d(z)<d(M )−k−dA
for the left hand side above. Summing over the possible values of the least
common multiple, we get
3(r+k) X X
≪ q 4 d2 (M )4 |L| 1
L|lcmσ∈S3 Mσb z∈Fq [T ]
3(r+k)
d(z)<d(M )− 4
(A.31) L|z
3(r+k) X n 3(r+k)
o
≪q 4 d2 (M )4 |L| max q d(M )− 4 −d(L) , 1 .
L|lcmσ∈S3 Mσb
3(r+k)
The contribution of q d(M )− 4 −d(L) (respectively, of 1) is the first (respec-
tively, the second) summand of the right hand side in our corollary.
Corollary A.3. Notation unchanged, we have

X X X 6
(A.32) ψ (∆b (h, s)) ≪ |M |d2 (M )12 q 4 (k+r)

b1 ,...,b′3 ∈MdB h∈Ab s∈A×

d(s)≤k+d A
Proof. By Corollary A.2 we have a bound of

X
3(r+k)

(A.33) ≪ d2 (M )5 |M | + q 4 lcmσ∈S3 Mσb .
b1 ,...,b′3 ∈MdB
Summing first over tuples Mσ of divisors of M we get

X X X 3(r+k)

(A.34) d2 (M )5 |M | + q 4 |lcmσ∈S3 Mσ | .
σ∈S3 Mσ |M b1 ,...,b′3 ∈Md
B
∀σ Mσ =Mσb
For each such tuple, the conditions

(A.35) Mσ = Mσb , σ ∈ S3
imply the congruences
(A.36) bi ≡ b′σ(i) mod Mσ , 1 ≤ i ≤ 3, σ ∈ S3
which determine b′1 , b′2 , b′3 mod lcmσ∈S3 Mσ once b1 , b2 , b3 are chosen. Hence,
for each tuple of divisors of M , the number of possible values of b is at most
n o
(A.37) max q 6dB |lcmσ∈S3 Mσ |−3 , q 3dB .
If p = 3, we need a slightly more complicated argument. There are at most
M△ 2 ways to choose the congruence classes of b modulo M , and then choos-
△
ing b1 , b2 , b3 arbitrarily now determines b′1 , b′2 , b′3 mod lcmσ∈S3 ∪{△} Mσ . Be-
cause the number of ways to choose b modulo Mσ and then choose b1 , b2 , b3
is
(
M△ 2 (q 3dB /M 3 ) ≤ q 3dB if M△ ≤ q dB
△
(A.38)
q 2dB ≤ q 3dB if M△ > q dB
in either case the number of possible values of b is at most
n −3 o
(A.39) max q 6dB lcmσ∈S3 ∪{△} Mσ , q 3dB .
Setting τ = d(lcmσ∈S3 Mσ ) (or adding △ if p = 3), and taking the maximal
possible contribution for every tuple of divisors of M , we get the bound
3(r+k)

(A.40) max d2 (M )|S3 |+1 q 6dB −3τ + q 3dB d2 (M )5 |M | + q 4 +τ .
0≤τ ≤d(M )
Expanding the brackets above, we see that each exponent is a linear function
of τ , hence maximized either at τ = 0 or at τ = d(M ). Using Eq. (A.6), one
3(r+k)
observes that the maximal terms q 6dB +d(M ) and q 3db + 4
+d(M )
agree and
arrives at the right hand side of Eq. (A.32).
It then follows from Eq. (A.8), Eq. (A.32), and the divisor bound that

X X X X
41 21 1
ψ af (a−1 g + b) ≪ q 24 r+ 24 k+( 6 +ǫ)d(M )

(a,M )=1
and thus (matching the ℓ = 3 case of [FM98, (1.2.3)] we get
(I) 17 7 1
(A.41) Σk,r ≪ q 24 r+ 8 k+( 6 +ǫ)d(M )
using Eq. (A.5). Using the fact that k ≤ 81 d(M ) + 7

16 d and k + r ≤ d we
arrive at the first bound in Eq. (A.4).
A.2. Sums of type II. Keeping the same notation, we follow the proof of
[FKM14, Theorem 1.17]. Applying Cauchy’s inequality and Polya-Vinogradov
completion as in [FKM14, Section 3], we get
X X
(II) 2
Σk,r ≤ q k δg1 δg2 ψ(f g1 − f g2 )
g1 ,g2 ∈Mr f ∈Mk
(f g1 ,M )=(f g2 ,M )=1
(A.42) X X
≤ qk q k−d(M ) C(g1 − g2 , h).
g1 ,g2 ∈Mr h∈Fq [T ]
d(h)<d(M )−k
where
X
(A.43) C(g, h) = ψ(gz)ep TrA
Fp (hz) , g, h ∈ Fq [T ].
z∈A×
We have the following analog of [FKM14, Proposition 3.1].

Proposition A.4. For g, h ∈ Fq [T ] and Mg,h = gcd(M, g, h) we have
p q
(A.44) |C(g, h)| ≤ d2 (M ) |M | |Mg,h |.
Proof. Since both C(g, h) and our putative bound are multiplicative in M ,
it suffices to show that for a prime P we have
p
(A.45) |C(g, h)| ≤ 2 |P |
unless g ≡ h ≡ 0 mod P . To demonstrate that, take f ∈ Fq [T ]/(P ) with

F [T ]/(P )
(A.46) ψ(z) = ep TrFpq (f z)
and note that

X
F [T ]/(P )
(A.47) C(g, h) = ep TrFpq (gf z −1 + hz) .
z∈(Fq [T ]/(P ))×
p
We have a Weil bound of 2 |P | for this exponential sum unless the rational
function gf z −1 + hz is an Artin-Schreier polynomial, which can only happen
if it is constant, as all its poles have order at most 1. The latter happens
only if g = h = 0, as desired.
By Eq. (A.42) we have

X X X
(II) 2
(A.48) k,r ≤ q 2k−d(M )
Σ C(g1 − g2 , h)
L|M h∈Fq [T ] g1 ,g2 ∈Mr
d(h)<d(M )−k Mg1 −g 2 ,h =L
so from Proposition A.4 and the divisor bound, this is at most

d(M ) X d(L) X X
(A.49) q 2k− 2 +ǫd(M ) q 2 1.
L|M h∈Fq [T ] g1 ,g2 ∈Mr
d(h)<d(M )−k Mg1 −g2 ,h =L
Since Mg1 −g2 ,h = L implies that g2 ≡ g1 , h ≡ 0 mod L, the above is at most

d(M ) X d(L)
(A.50) q 2k− 2 +ǫd(M ) q 2 +r+max(r−d(L),0)+max(d(M )−k−d(L),0)
L|M
so setting ξ = d(L) and applying the divisor bound once again, we arrive at
ξ d(M )
(A.51) max q 2k+ 2 +r+max{r−ξ,0}+max{d(M )−k−ξ,0}− 2
+ǫd(M )
.
0≤ξ≤d(M )
As a function of ξ, the exponent above is convex, so its values do not

exceed those at ξ = 0 and ξ = d(M ), which are
d(M )
+ǫd(M )+2r+k
(A.52) q 2 , q r+2k+ǫd(M )
in view of our assumption that r ≤ d ≤ d(M ). Taking a square root and
using the fact that k + r ≤ d we get the second bound in (A.4).
Remark A.5. There are (at least) two other potential approaches for bound-
ing our type II sums. The first is to follow the proof of [FM98, Proposi-
tion 1.3] that uses Bombieri’s bound on complete exponential sums [FM98,
Lemma 4.3]. One then argues as in [FM98, Section 5]. The second is to
follow the proof of [FM98, Theorem 1.4] given in [FM98, Section 7].
References
[BBSR15] E. Bank, L. Bary-Soroker, L. Rosenzweig, Prime polynomials in short intervals
and in arithmetic progressions, Duke Math. J. 164.2 (2015): 277-295.
[BGP92] E. Bombieri, A. Granville, J. Pintz, Squares in arithmetic progressions, Duke
Math. J. 66.3 (1992): 369-385.
[BZ02] E. Bombieri, U. Zannier, A Note on squares in arithmetic progressions, II, Atti
della Accademia Nazionale dei Lincei. Classe di Scienze Fisiche, Matematiche e Nat-
urali. Rendiconti Lincei. Matematica e Applicazioni 13.2 (2002): 69-75.
[Bur63] D. A. Burgess (1963), On Character Sums and L-Series II, Proc. Lond. Math.
Soc. 3(1), 524-536.
[Ca15] D. Carmon, The autocorrelation of the Möbius function and Chowla’s conjecture
for the rational function field in characteristic 2, Phil. Trans. R. Soc. A 2015.
[CaRu14] D. Carmon, Z. Rudnick, The autocorrelation of the Möbius function and
Chowla’s conjecture for the rational function field, Q. J. Math. (2014) 65 (1):53–61.
[CHLPT15] A. Castillo, C. Hall, R. Lemke Oliver, P. Pollack, L. Thompson, (2015)
Bounded gaps between primes in number fields and function fields, Proc. Amer. Math.
Soc.143(7), 2841-2856.
[CG07] J. Cilleruelo, A. Granville, Lattice points on circles, squares in arithmetic progres-
sions and sumsets of squares, Additive combinatorics, 43, 241-262, Amer. Math. Soc.
Providence, RI, 2007.
[CCG08] B. Conrad, K. Conrad, and R. Gross (2008) Prime specialization in genus 0,
Transactions of the AMS, 360(6), 2867-2908.
[Con05] K. Conrad (2005), Irreducible values of polynomials: a non-analogy, Number

Fields and Function Fields: Two Parallel Worlds, 71-85, Progress in Mathematics,
239, Birkhauser, Basel.
[FM98] E. Fouvry, P. Michel, Sur certaines sommes d’exponentielles sur les nombres pre-
miers, Ann. scient. Ec. Norm. Sup., 4e serie, t. 31, 1998, 93-130.
[FKM14] E. Fouvry, E. Kowalski, P. Michel (2014), Algebraic trace functions over the
primes, Duke Math. J. 163(9), 1683-1736.
[Gal72] P. X. Gallagher (1972), Primes in progressions to prime-power modulus, Invent.
math. 16(3), 191-201.
[GY03] D. Goldston, C. Yıldırım, Higher correlations of divisor sums related to primes I:
Triple correlations, Integers 3 (2003) A5.
[GS18] O. Gorodetsky, W. Sawin, Correlation of Arithmetic Functions over Fq [T ], arxiv
preprint, 2018.
[Hoo91] C. Hooley, On the number of points on a complete intersection over a finite field,
J. of Number Theory, 38.3 (1991), 338-358.
[Hsu99] C. N. Hsu (1999), Estimates for Coefficients of L-Functions for Function Fields,
Fin. Fields App. 5(1), 76-88.
[Ill94] L. Illusie (1994) Autour du théorême de monodromie locale, Astérisque 223 (1994)
9-57..
[IK04] H. Iwaniec, E. Kowalski, Analytic number theory, Vol. 53. Amer. Math. Soc. 2004.
[Jan07] S. Janson (2007), Resultant and discriminant of polynomials, Notes.
[Kat89] N. Katz (1989), An Estimate for Character Sums, JAMS, 2, 2, 197-200.
[Ma15] J. Maynard (2015), Small gaps between primes, Ann. Math. 181, 1-31.
[MR16] K. Matomäki, M. Radziwill (2016), Multiplicative functions in short intervals,
Ann. Math 1015-1056.
[PM14] Polymath, D. H. J. (2014) Variants of the Selberg sieve, and bounded intervals
containing many primes, Res. Math. sci. 1(1), 12.
[Ros13] M. Rosen (2013), Number theory in function fields (Vol. 210). Springer Science &
Business Media.
[Sa18] W. Sawin, Square-root cancellation for sums of factorization functions over short
intervals in function fields, arxiv preprint, 2018.
[SGA7-II] P. Deligne, N. Katz, eds. Séminaire de Géométrie Algébrique du Bois Marie -
1967-69 - Groupes de monodromie en géométrie algébrique - (SGA 7) - vol. 2, Lecture
Notes in Mathematics (in French), Vol. 340. Springer-Verlag.
[Tao16] T. Tao, The logarithmically averaged Chowla and Elliott conjectures for two-point
correlations, Forum of Mathematics, Pi, Vol 4, 2016, e8, Cambridge University Press.
[TT17] T. Tao, J. Teräväinen, The structure of logarithmically averaged correlations of
multiplicative functions, with applications to the Chowla and Elliott conjectures, arXiv
preprint, 2017.
Department of Mathematics, Columbia University, New York, NY 10027,

USA
E-mail address: sawin@math.columbia.edu
Department of Mathematics, University of Wisconsin-Madison, 480 Lincoln

Drive, Madison, WI 53706, USA
E-mail address: mshusterman@wisc.edu

1808 04001

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

1808 04001

Încărcat de

Drepturi de autor:

Formate disponibile

ON THE CHOWLA AND TWIN PRIMES CONJECTURES

WILL SAWIN AND MARK SHUSTERMAN

Abstract. Using geometric methods, we improve on the function ﬁeld

with 1p∤h equals 1 if h is not divisible by p, and 0 otherwise.

W.S. served as a Clay Research Fellow while working on this paper.

where q is a prime power, P ranges over all primes (monic irreducibles) of

Theorem 1.1. For an odd prime number p, and a power q of p satisfying

as X → ∞ through powers of q. Moreover, we have a power saving (depend-

For example, the 2-point Hardy-Littlewood conjecture holds over

(1.6) F315 , F511 , F79 , F118 , F6850933 .

In case h ∈ Fq [T ] is a monomial, the fact that the count above tends

1.2. The key ingredient - Chowla’s conjecture. The main ingredient

The only completely resolved case is k = 1 which is essentially equivalent to

For instance, the 2-point Chowla conjecture holds over

and its derivative f ′ . In characteristic p, the derivative of f = r + sp is equal

By further enlarging q, we get arbitrarily close to square root cancellation.

Remark 1.5. The relation between Möbius and multiplicative characters

As in the previous theorems, we obtain a power savings estimate. Here

show that for a power q of an odd prime p, and coprime a, M ∈ Fq [T ], the

In order to deduce from Theorem 1.7 an improved level of distributions for

a level of distribution of 32 = 21 + 61 , which is a longstanding barrier for

As before, we can obtain a power saving. Since the progressions are so

In view of Maier’s phenomenon, we cannot expect to obtain the theorem for

which may be relevant for counting primes of the form f 2 + T .

To prove this theorem, we use the following geometric setup. View At as

to c1 f + h + c2 T t . We will prove (2.1) using geometric properties of the

Because χ is nontrivial, Lχ (T − x) has nontrivial mondromy, so

is supported in degree t, where it equals the t-th tensor power of an m − 1-

Proof. Let l1 : A1 → A2 send c to (c, 1). We consider the vanishing cycles

It remains to calculate the rank of the stalk at a particular point (h0 , 0) ∈

By Lemma 2.5, (Rπ∗ j! Lχ (c1 f + h + c2 T t ))(1,η) is supported in degree t

again using Lemma 2.6.

as t → ∞, with the implied constant depending only on q.

3. The Möbius Function

so we get the mod p congruence

Proof. Pellet’s formula (see [Con05, Lemma 4.1]) gives

(3.7) µ(a + gM ) = (−1)d(a+gM ) ψ (Disc(a + gM ))

4. Linear forms in Möbius

For the derivative we then have

so we set ai = iT i−1 for 1 ≤ i ≤ p − 1, and consider the congruence

Since M is squarefree, we can uniquely pick b, b1 , . . . , bp−1 ∈ Fq [T ] with

and from this restriction on g1 the proposition follows.

The following technical proposition follows from the arguments of [BGP92,

as d → ∞, with the implied constant depending only on β, ǫ, δ, n and q.

For 1 ≤ i < j ≤ n we set

and claim that, for every integer ℓ ≥ 0 and γ > 0, we have

since d(U∗ ) ≥ ℓ, so our claim is established.

so splitting according to the residue class u of s mod Ur we get

for any suﬃciently small choice of γ.

Let r ∈ R for which χr is principal. Pairwise coprimality implies that

it follows that A | Q as well, so Proposition 4.2 ensures that the number of

Then for nonnegative integers d, m1 , . . . , mn , k1 , . . . , kn with

as d → ∞, with the implied constant depending only on ǫ, δ, β, n and q.

Corollary 4.6. Fix ǫ, δ > 0 and a positive integer n. Let q be a power of

holds, and then apply Theorem 4.5.

as d → ∞, with the implied constant depending only on q and θ.

Summing all the coeﬃcients of a power series, evaluates it at u = 1.

The coeﬃcients of degree greater than d contribute to the error term. To

for r < q, getting a bound of

As long as the above maximum is subexponential in m for all r < q, the

Taking r arbitrarily close to q, the above is q o(m) (q − o(1))d = q o(m+d)−d .

In the following we deduce, from our results on the Möbius function, a

as d → ∞, with a power saving error term Eq .