Papers, books, translations, surveys
(The numbering is not quite consecutive because certain preprints
have been moved to the Unpublished section.)79 
Research paper
(preprint, 2020)

Equidistribution
from the Chinese Remainder Theorem, with
.
We prove that if one is given some subsets of residues classes modulo primes, and extend these to subsets modulo all squarefree moduli using the Chinese Remainder Theorem, then the fractional parts of these sets will become equidistributed modulo 1 if either (1) the sets modulo primes have at least two elements for a positive density of primes, and the moduli have sufficiently many prime factors; or (2) the sets modulo primes has growing size, and the moduli have at least two prime factors. This puts in a much wider context (and strenghtens) a previous result of Hooley concerning roots of polynomial congruences. We also prove higherdimensional analogues and consider various applications, e.g., to roots of pseudopolynomials (in the sense of Hall). 
78 
Survey paper
(Contemporary
Math., 2019)

Lectures
on applied ℓadic cohomology, with
, and .
This is a detailed survey of many of the results about, or proved with, trace functions over finite fields in the last years. It is an expanded version of the lectures given by Ph. Michel during the 2016 Arizona Winter School. 
77 
Research paper
(Forum
of Math, Sigma, 2020)

Periodic twists of GL_{3}modular forms, with
, and .
This paper extends the method recently devised by Holowinsky and Nelson (motivated by earlier ideas of Munshi) to prove subconvexity for critical values of Lfunctions of a fixed GL_{3}modular form twisted by Dirichlet characters modulo a prime, and proves a general estimate for sums of Heckeeigenvalues of such a modular form twisted by any qperiodic function with bounded discrete Fourier transform, where the sum has length close to q^{3/2}. This applies in particular to trace functions of Fourier sheaves, and in that respect it is similar to the paper 51 (with Fouvry and Michel); however, it turns out that the problem is somewhat easier, due to the longer length of the sums compared to the modulus (although it is the natural length for applications to subconvexity). 
76 
Book
(SMF, Cours spécialisés, 2019)

An introduction to expander graphs
This is the published version of my lecture notes on expander graphs. A preliminary version is available, as well as a list of corrections (updated Jan. 29, 2020). 
75 
Research monograph
(Memoirs of the AMS, to appear)

The second moment theory of families of Lfunctions, with , , and and .
Building on our previous works 66 and 68, one can prove asymptotic formulas with power saving for the first and second twisted moments of the family of Lfunctions of a fixed cusp form twisted by Dirichlet characters of large prime modulus. It is wellunderstood that this information, when available for any reasonable family of cusp forms, opens the possibility of many arithmetic applications. In this monograph, we present the general picture first informally, and then derive many of these applications in our particular case. We also consider some applications that are much more specific to this specific family: (1) the asymptotic computation of the variance of modular symbols of a fixed holomorphic cusp form of level 2 modulo a prime (answering partly a question of Mazur and Rubin), (2) positive proportion of nonvanishing of central values of twisted Lfunctions with a condition on the discrete Mellin transform of a suitable trace function (this last part relies on the recent work of Katz on Mellin transforms over finite fields). 
74 
Research paper
(Annali della Scuola Normale Superiore di Pisa, to appear)

Stratification and averaging for exponential sums:
Bilinear
forms with generalized Kloosterman sums,
with and .
This is a followup to our paper 68: we refine our methods to obtain nontrivial bounds for bilinear forms with Kloosterman sums with characters in a range of the variables that is identical with the bestknown results for much simpler (monomial) exponentials. Analytically, the idea is to use higher Hölder inequalities, but this is not sufficient by itself. The key new geometrical idea is to show that even when the resulting "sumproduct" sheaves fail to be irreducible, their decomposition reflects that of the "input" sheaves, except for parameters in a highcodimension subset. This fact is proved by a subtle interplay between étale cohomology in its algebraic and diophantine incarnations. We derive a first application to a first moment of a family of degree 3 Lfunctions. This preprint was briefly known as "Bilinear forms with Kloosterman sums, II". 
72 
Survey
(L'enseignement
mathématique, 2018)

Good
cyclic codes and the uncertainty principle, with and .
This is a survey of variants of the uncertainty principle over various fields (especially finite fields), and how even a rather weak form of such a result would ensure the existence of sequences of "good" cyclic codes, a longstanding open problem in coding theory. Arithmetically, there are many fascinating connections with the properties of the order of elements of the multiplicative group of a finite field, e.g., Artin's conjecture on primitive roots. 
71 
Research paper
(Proc. Steklov
Inst. Math., 2017)

Some applications of smooth bilinear forms with
Kloosterman sums, with , , and .
This is a complement to our previous paper 66, based on a recent bound of I. Shparlinski and T.P. Zhang for smooth bilinear forms with coefficients Kl_{2}(mn;p). 
70 
Research paper
(in "Exploring the Riemann zeta function", Springer, 2017)

Bagchi's Theorem for families of automorphic forms.
This is a quick discussion of a GL_{2} case of Bagchi's Theorem on the functional limit behavior of families of Lfunctions on the right of the critical line. The argument is adapted from my lecture notes on probabilistic number theory (where the original case of the Riemann zeta function is discussed); see also the summary on my blog. 
68 
Research paper
(Annals
of Math., 2017)

Bilinear forms with Kloosterman sums and applications, with and .
We prove nontrivial estimates for general bilinear forms with coefficients of the type K(mn;p), where K(*;p) are (hyper)Kloosterman sums. This has applications for instance to the average of twisted Lfunctions of cusp forms, as in paper 66. Indeed, we obtain a powersaving asymptotic formula in that problem. The heart of the paper is a detailed and subtle study of auxiliary sheaves that we call "sumproduct sheaves", which we need to prove are (generically) geometrically irreducible. This requires many deep results from algebraic geometry, going beyond the basic formalism of the Riemann Hypothesis (e.g., to determine precisely the local monodromy of sumproduct sheaves at suitable singularities). 
67 
Research paper
(Annales de l'Institut Fourier, 2017)

On short sums of trace functions, with , , , and .
We consider sums of oscillating functions on intervals in cyclic groups of size close to the square root of the size of the group. We first prove nontrivial estimates for intervals of length slightly larger than this square root (bridging the "PólyaVinogradov gap" in some sense) for bounded functions with bounded Fourier transforms. We then prove that the existence of nontrivial estimates for ranges slightly below the squareroot bound is stable under the discrete Fourier transform, and we give a number applications related to trace functions over finite fields. (Remark. This paper contains a considerable strengthening of the basic "sliding sum" bound of reference 58, but not the results on generalized arithmetic progressions there.) 
66 
Research paper
(American J. of
Math., 2017)

On moments of twisted Lfunctions, with , , and .
We study the average over Dirichlet characters of prime modulus of the product of two central values of twisted Lfunctions of modular forms. When the two forms are a special Eisenstein series, this is the fourth moment of the central values of the Dirichlet Lfunctions. Using new bounds for shifted convolution sums (avoiding a dependency on the RamanujanPetersson conjecture) and new bounds for bilinear forms in Kloosterman sums, we obtain an asymptotic formula with power saving when one of the two forms is the Eisenstein series; in the second case, this improves (and streamlines) significantly a wellknown result of M. Young. We also show how a conjecture on complete twovariable sums over finite fields leads to an asymptotic formula with power saving for two cusp forms. (Remark. The conjecture mentioned was proved by Ph. Michel, W. Sawin and myself in the paper 68, hence the asymptotic formula for two cusp forms is now unconditional.) 
65 
Research paper
(Compositio Math., 2016)

Kloosterman paths and the shape of exponential sums, with .
We study the statistics of the shape of the graphs joining successive partial sums of Kloosterman sums (see the Kloostermania page), as well as of other types of exponential sums. We prove convergence in the space of continuous functions for a suitable family, with limit given by a specific remarkable random Fourier series. This result turns out to depend on Katz's determination of the monodromy groups of Kloosterman sheaves (and the Riemann Hypothesis over finite fields), and on estimates for averages of short sums at the edge of the PólyaVinogradov range. Using some probabilistic results of Talagrand and MontgomerySmith in particular, we deduce bounds on the tails of the maximum of the normalized partial sums. 
64  Book 
An introduction to the representation theory of groups
This book is based on my introductory lecture notes on representation theory, and presents the fundamental concepts of representation theory for groups, as well as some applications. I have started collecting some corrections and additions. 
63 
Research survey
(Phil. Trans. Royal
Soc. A, 2015)

A study in sums of products, with and .
This is to a large extent a survey of the applications of the ideas of Katz surrounding the GoursatKolchinRibet criterion for monodromy groups of direct sums of sheaves to the estimation of sums over finite fields where the summands are products of trace functions of suitable sheaves. Such sums arise frequently in analytic number theory, and we provide a convenient reference for general statements as well as for important special cases. We illustrate the method with a number of applications, old and new. A shorter version appears in the proceedings of the Royal Society Meeting Number fields and function fields: coalescences, contrasts and emerging applications. 
62 
Survey
(Bourbaki seminar, 2014)

Gaps
between prime numbers and prime numbers in arithmetic
progressions, after Y. Zhang and J. Maynard
This is the written English version of my lecture at the Bourbaki seminar in March 2014 concerning the work of Y. Zhang and J. Maynard on gaps between primes. A French translation is in preparation. 
61 
Survey
(Colloquium De
Giorgi, 2015)

Trace functions over finite fields and their applications, with and .
This is a survey of (most) of our papers on trace functions, with an emphasis on 51, 55 and 57. It is a written and slightly expanded version of a lecture I gave in early 2013 at the Colloquio de Giorgi of the Scuola Normale Superiore di Pisa, and appears in the series of proceedings volumes. 
60 
Research paper
(GAFA, 2014)

Fourier coefficients of GL(N) automorphic forms in arithmetic progressions, with .
This paper generalizes the results of my previous paper with É. Fouvry, S. Ganguly and Ph. Michel (number 56 below) to all cusp forms on GL(n) for all N≥ 3, and to the Nary divisor function d_{N}, obtaining a Gaussian distribution result for Fourier coefficients in arithmetic progressions in suitable ranges. This is therefore a new case of an analytic property that is now known for automorphic forms on all linear groups GL(N)! From this point of view, one can note that, besides the RankinSelberg convolution and bounds towards the RamanujanPetersson conjecture, which occur in the proofs of other "universal" results, we require deep equidistribution statements for products of hyperKloosterman sums. 
59 
Research paper
(preprint, 2014; updated 2019)

On the conductor of cohomological transforms, with and .
We show that the conductor of ℓadic sheaves on the affine line defined by "cohomological transforms" with a kernel given by an ArtinSchreier sheaf in two variables is bounded in terms of the conductor of the input sheaf and the degree of the numerator and denominator of the rational function defining the kernel. In particular, this applies to the Fourier transform in one variable, and recovers in that case a result we used in our previous works and which we had deduced from the deep local Fourier transform theory of Laumon. W. Sawin has developed a much more general theory in work in progress, so this paper is mostly of historical interest, and can be useful as an illustration of elementary cohomological techniques, just a bit beyond the formalism of the Riemann Hypothesis. 
58 
Research paper
(preprint, 2013)

The sliding sum method for short exponential sums, with and .
We develop a very simple idea to obtain nontrivial estimates for sums of suitable functions over intervals in a finite abelian group (or more generally, over proper generalized arithmetic progressions) when the size of the summation set is just within the gap between the squareroot of the size of the group and the range where the PólyaVinogradov completing method becomes relevant. The condition on the function is related to the size of its additive correlation sums, and we show that trace functions satisfy this condition, except when it is obviously false. We then derive some applications to equidistribution in the same ranges. (Remark. The basic inequality has been considerably strengthened in reference 67, jointly with CS. Raju, J. Rivat and K. Soundararajan; some results on generalized arithmetic progressions are however not considered in that later paper.) 
57 
Research paper
(Mathematika, 2014)

On the exponent of distribution of the ternary divisor function, with and .
This paper is an illustration of the power and flexibility of the results in our first two papers (number 51 and 55 below) concerning trace functions over finite fields and their uses in analytic number theory. We give a much more streamlined proof of the result of Friedlander and Iwaniec (later improved by HeathBrown) that the exponent of distribution of the ternary divisor function in arithmetic progressions (to prime moduli at least) is strictly larger than 1/2, and in fact obtain a value for this exponent which also improves the one of HeathBrown. Combining this with "Kloostermania", we also improve the exponent of distribution on average over (prime) moduli for a fixed residue class. 
56 
Research paper
(Commentarii
Math. Helv., 2014)

Gaussian distribution for the divisor function and Hecke eigenvalues in arithmetic progressions, with , and .
We prove that the error term of the divisor function in arithmetic progressions modulo a prime number p has a Gaussian limiting distribution as a function of the residue class, in a suitable range. We also show a similar result for Hecke eigenvalues of classical primitive holomorphic cusp forms of level 1, and determine when the error term for the residue classes a and γ(a) are asymptotically independent, for a projective linear transformation γ modulo p. The proof uses the moment method and involves the Voronoi summation formula and a crucial use of the Riemann Hypothesis over finite fields, in the form of a result of independence of twists of Kloosterman sheaves. 
55 
Research paper
(Duke Math. Journal, 2014)

Algebraic trace functions over the primes, with and .
We prove that sums over primes of trace functions of middleextension sheaves on A^{1}/F_{p} are small, when the length is close enough to p in logarithmic scale, and a similar statement for correlation of trace functions with the Möbius function. The bound is valid for any trace function which is not proportional to the product of an additive and a multiplicative character. We then prove a number of applications (estimates for sums of multiplicative characters of a rather general polynomial over the primes, existence of a positive density of "large" Kloosterman sums with moduli having two prime factors, among others.) 
54 
Research paper
(Math. Proc. Cambridge
Phil. Soc., 2013)

An inverse theorem for Gowers norms of trace functions over F_{p}, with and .
We prove that the Gowers uniformity norms of functions over a prime field which arise as trace functions of a middleextension sheaf on the affine line are asymptotically "as small as possible", except if the sheaf contains ArtinSchreier sheaves of polynomials. This is a strong form of inverse theorem for these functions with algebraic structure. 
53 
Research paper
(Math. Res. Letters, 2013)

Counting sheaves using spherical codes, with and .
Using a result on the size of spherical codes in a suitable range, we obtain here a polynomialtype bound (in terms of the size of the finite field) for the number of lisse sheaves on an algebraic curve over a finite field which have bounded conductor (i.e., bounded rank and ramification). As an application, we show that a function defined on a finite field can not usually be expressed as a short linear combination of trace functions of sheaves with small complexity. 
52 
Survey
(MSRI
publications, 2014)

Sieve in discrete groups, especially sparse
This is the written version of the minicourse I gave during the "Hot Topics" workshop at MSRI on "Thin groups and superstrongapproximation" that I coorganized in February 2012. It has a certain intersection with my Bourbaki seminar talk on sieve and expanders (see paper number 47 below) but I have tried to make this survey complementary of the previous one. In particular there is more discussion of the recent versions and applications of the very general large sieve inequality by Lubotzky, Meiri and Rosenzweig. There is also a discussion of an ErdösKac Theorem in the context of the affine sieve. 
51 
Research paper
(GAFA, 2015)

Algebraic twists of modular forms and Hecke orbits, with and .
We consider correlations of Fourier coefficients of classical modular forms with functions "of algebraic origin", defined as trace functions of suitable sheaves on A^{1}/F_{p}. We are able to show a strong form of asymptotic orthogonality. Analytically, the main tools are the amplification method and the Kuznetsov formula, and geometrically we depend crucially on the full force of the Riemann Hypothesis over finite fields, and on the properties of the Fourier transform for sheaves, as defined by Deligne and studied by Katz, Laumon and others. This leads to applications concerning the distribution of "twisted" Hecke orbits, or discrete horocycles, on modular curves. 
50 
Research paper
(IMRN, 2012)

Explicit growth and expansion for SL_{2}
This contains a streamlined selfcontained account of fully explicit versions of the growth theorem of Helfgott and of the BourgainGamburd expansion theorem for the case of SL_{2}(F_{p}). Warning! The current version (updated on July 2, 2012) contains many changes and corrections, compared with the first draft, some of them quite important, and including all (!) numerical constants; many thanks to the remarkable work of the referee on this paper. 
49 
Survey
(Pub. Math. Besançon, 2013)

Families of cusp forms
This survey describes some ideas towards a definition (or some understanding...) of the notion of a family of cusp forms. This is influenced by similar ideas of P. Sarnak, and related to earlier work of CogdellMichel, as well as to the "recipe" of Conrey, Farmer, Keating, Rubinstein and Snaith for the prediction of moments of Lfunctions on the critical line. 
48 
Research paper
(IMRN, 2014)

Modφ convergence, with and .
We discuss a generalization of the notion of modGaussian convergence, which we call "modφ convergence". This applies to much more general model distributions than the Gaussian ones, and we show how, in this context, very simple conditions lead to a local limit theorem when modφ convergence holds. We derive many applications, from the winding number of planar brownian motion to models of random squarefree integers, and show that this framework suggests a strong form of the conjecture that the values of the Riemann zeta function on the critical line are dense in the complex plane. This paper is dedicated to the memory of Marc Yor. 
47 
Survey
(Association
Bourbaki, Astérisque,
20102012)

Sieve in expansion and Crible en expansion
This is a survey paper on the recent developments of sieve involving expander graphs, notably the "sieve in orbits" of Bourgain, Gamburd and Sarnak, with a few quick asides concerning geometric applications (e.g., to DunfieldThurston random 3manifolds). It is the written counterpart of a lecture at the Séminaire Bourbaki. The French version (translation) is essentially identical with the English one. 
46  Research paper (Compositio
Math., 2012)

Local spectral equidistribution for Siegel modular forms and applications, with and .
This is a first exploration, in the case of a group of rank 2 which is not a general linear group, of the philosophy that says (roughly) that good "families" of cusp forms should have the property that, for any fixed prime p, their pcomponent (in the sense of automorphic representations) should behave well, and in particular become equidistributed in a suitable space of Satake parameters, with respect to some measure, possibly depending on the prime (there is a semiphilosophical discussion in this talk that I gave at R. Bruggeman's 65th birthday conference in 2009). We study this local distribution question for Siegel cusp forms of genus 2, where the parameter growing to infinity is the weight. Our averages are performed subject to a specific weighting which allows us to apply results concerning Bessel models and a variant of Petersson's formula. We obtain for this family a quantitative local equidistribution result, and derive a number of consequences. In particular, we show that the computation of the density of lowlying zeros of the degree 4 spinor Lfunctions (for restricted test functions) gives global evidence for a wellknown conjecture of Böcherer concerning the arithmetic nature of Fourier coefficients of Siegel cusp forms, which involves central values of Lfunctions.. (Note that qualitative equidistribution statements can be proved more easily using the approximate orthogonality of Fourier coefficients of Poincaré series proved in our earlier paper, number 43 below). 
45  Research paper (Israel J. Math., 2012)

Splitting fields of characteristic polynomials of random elements in arithmetic groups, with and .
We discuss rather systematically the principle, implicit in our earlier paper 32 and in other works, that for a "random" element in an arithmetic subgroup of a (split, say) reductive algebraic group over a number field, the splitting field of the characteristic polynomial, computed using the any faitful representation, has Galois group isomorphic to the Weyl group of the underlying algebraic group. Besides tools such as the large sieve, which we had already used, we introduce some probabilistic ideas (large deviation estimates for finite Markov chains) and the general case involves a more precise understanding of the way Frobenius conjugacy classes are computed for such splitting fields (which is related to a map between regular elements of a finite group of Lie type and conjugacy classes in the Weyl group which had been considered earlier by Carter and Fulman for other purposes; we show in particular that the values of this map are equidistributed). 
44  Research paper (Duke Mathematical Journal, 2012)

Expander graphs, gonality and variation of Galois representations, with and .
We show that strong finiteness theorems concerning variation of Galois representations (e.g., with respect to maximal Galois action on torsion points of oneparameter families of abelian varieties) over number fields can be obtained using expansion properties of Cayley graphs attached to a family of coverings of the base curve, using lowerbounds for the gonality in such families that arise by combining a result of Li and Yau with the comparision principle of Brooks and Burger relating the first nonzero eigenvalue of the Riemannian and combinatorial Laplace operators for hyperbolic surfaces. Although expander families are natural targets for our methods, a weaker notion of expansion (which we call esperantism) is also sufficient. This surprising connection leads also to many intriguing questions. This paper is in some sense a continuation of the paper 34 below (also joint with ); additional discussion can be bound in this blog post of J. Ellenberg, and this other one. 
43  Research paper (Mathematika, 2011)

A note on Fourier coefficients of Poincaré series, with and .
We give a short and "soft" proof of the qualitative approximate orthogonality of Fourier coefficients of Poincaré series for classical and Siegel modular forms, in the limit when the weight goes to infinity. 
42  Research paper (Journal of the LMS, 2012)

ModGaussian convergence and the value distribution of ζ(1/2+it) and related quantities, with .
A difficult question in the theory of the Riemann zeta function (first raised by Ramachandra) asks whether the set of values ζ(1/2+it), t real, is dense in the complex plane. We provide evidence for an affirmative answer (in a strong quantitative form) by relating the problem to modGaussian convergence (as defined in our paper 35 with J. Jacod) and the moment conjectures for the zeta function. This remains conditional, but we obtain results of independent interest concerning values of characteristic polynomials of random matrices in compact classical groups, and from this derive similar results for values of Lfunctions over finite fields  e.g. the central values of Lfunctions of Dirichlet characters over F_{p}[T] form a dense subset of the complex plane, where p runs over primes. (Version updated May 21, 2010). 
41  Research paper 
Moddiscrete expansions, with and .
This is a systematic exploration of the probabilistic consequences of conditions similar to modPoisson convergence, in terms of approximation (in Kolmogorov or total variation sense) of discrete random variables by Poisson variables, PoissonCharlier expansions, or similar constructions. In particular, this provides approximations in total variation norm for the number of prime factors of integers up to N, with error that can be an arbitrary inverse power of log log N. This is also partly motivated by earlier results of Hwang (based on probability generating functions). 
40  Research paper (Math. Proc. Cambridge
Phil. Soc., 2010)

On modular signs, with , and .
We consider in a fairly systematic way the properties of the sequence of signs of Hecke eigenvalues of classical primitive modular forms, obtaining results concerning the first sign change (improving strongly a result of Iwaniec, Kohnen and Sengupta), "multiplicity one" theorems for the sequence of signs (with small set of exceptions), and statistical upper and lower bounds for the number of forms with a given set of signs for the first few primes. 
39  Survey (Milan
Journal of Mathematics, 2010)

Some aspects and applications of the Riemann Hypothesis over finite fields.
This is a survey of some of the applications of the Riemann Hypothesis over finite fields, in the realm of analytic number theory, with a particular emphasis on presenting Deligne's Equidistribution Theorem and some of its applications. This is the written version of my lecture at the Verbania Conference on the 150th anniversary of the Riemann Hypothesis. 
38  Research paper
(IMRN, 2010)

ModPoisson convergence in probability and number
theory, with .
Building on our previous work with J. Jacod (paper 35), we consider in more detail the "modPoisson" convergence of random variables. Besides pointing out various instances in probability, we emphasize its presence in the distribution of the number of prime and irreducible factors of integers and polynomials over finite fields. In that second case, we show that the result amounts to a zerodimensional case of the KatzSarnak philosophy in the limit of large conductor. See this post for an explanation of the analogy involved. 
37  Research
paper (Archiv
der Mathematik, 2010)

Amplification arguments for large sieve inequalities.
This paper combines an earlier note giving an alternate proof of the arithmetic large sieve inequality, based on amplification ideas, and parts of the note on "modular signs", linked together by means of another result on the statistical paucity of primitive modular forms for which the first few Hecke eigenvalues have the same sign; this last result is deduced from a largesieve type inequality which has the feature of (partly) unifying the classical large sieve and the spectral large sieve inequalities for Fourier coefficients of cusp forms, and the proof depends on using the amplification method. 
36  Research paper (Experimental Mathematics, 2012)

The Chebotarev invariant of a finite group, with
We consider invariants of a finite group G which are related to the number of random (independent, uniformly distributed) conjugacy classes which are required to generate it; this is related, intuitively at least, to the number of primes required to determine the Galois group of an integral polynomial when computing Frobenius conjugacy classes (though the precise link between the model considered here and the actual arithmetic situations requires some care). This paper gives purely grouptheoretic expressions for the invariants, and uses them to investigate their values, both theoretically for certain families of groups which are easily accessible, and numerically using Magma. For a more informal description, see this blog post. 
35  Research paper (Forum
Mathematicum, 2011)

ModGaussian convergence: new limit theorems in probability and number theory, with and .
We show that certain theorems and conjectures in Random Matrix Theory can be interpreted as instances of a new type of limit theorems in probability theory, which in a sense "refine" standard results of convergence to the normal distribution. A similar phenomenon, with Poisson distribution instead, turns out to be present in the classical theorem of Erdös and Kác on the distribution of the number of prime factors of integers. Also arXiv:0807.4739. 
34  Research paper (Journal of the LMS, 2009)

Nonsimple abelian varieties in a family: geometric and analytic approaches, with , and .
We look at a single question using two quite different approaches, and obtain complementary answers... The question is to "count" how many fibers (in a number field) of a generically simple abelian variety remain simple, and the answer is always that not many do. The methods of arithmetic geometry (around monodromy groups) lead, using the theorem of Faltings, to the result that there are only finitely many exceptions, but not to quantitative information (in terms of the family). Sieve techniques (based on mixing the sieve for Frobenius and a version of Gallagher's "larger" sieve) give an estimate which grows to infinity with the permitted height of the exceptions... but which is effective and fairly wellcontrolled. So depending on the point of view, either may be more useful! A short discussion is in this blog post of J. Ellenberg. (See also a paper of Y. Zarhin, who finds many examples where the set of exceptions is actually empty). Also arXiv:0804.2166. 
33  Research paper (Acta
Arithmetica, 2011)

Averages of Euler products, distribution of singular series
and the ubiquity of Poisson distribution
This paper extends a wellknown result of Gallagher on the average value of the singular series for the prime ktuple conjecture in various directions. For instance, the (conditional) applications to the distribution of primes in intervals of "fair" length are extended to the distribution of any prime patterns in such intervals: if uniform versions of the BatemanHorn conjecture hold, there will be convergence to a Poisson distribution with the appropriate mean. In addition, the proofs are rather more transparent and intuitive. Also arXiv:0805.4682. 
32bis  Survey (Cambridge University Press, 2011)

A survey of algebraic exponential sums and some applications
This is an expanded version of a talk I gave at a (wonderful!) workshop on Motivic Integration and its interactions with model theory and nonarchimedean geometry, organized in 2008 by R. Cluckers, A. Macintyre, J. Nicaise and J. Sebag; it appears as a chapter in the proceedings volume. This survey of exponential sums over finite fields overlaps a bit  but surprisingly little  with the other one I wrote later for the Verbania conference (item 39); the main originality is the discussion of exponential sums over definable sets over finite fields, following the paper 26. (Note that this paper is numbered 32bis because I had forgotten to include it on the web page at the right time...) 
32  Research paper 
An explicit integral polynomial whose splitting field has
Galois group W(E_{8}),
with
and .
By exploiting ideas about the "generic" behavior of the splitting field of characteristic polynomials of elements of linear algebraic groups, we construct an explicit polynomial P of degree 240 with integral coefficients such that the Galois group of its splitting field is the Weyl group of the exceptional group of type E_{8}. (Apparently, the first such polynomial to be written down, though the existence has been known for a long time; the largest coefficients of P have about a hundred digits). This paper is dedicated to H. Cohen and appears in the special volume of the Journal de Théorie des Nombres de Bordeaux published in his honor. Also arXiv:0801.1733. 
31  Research paper (IMRN, 2008)

The large sieve,
monodromy and zeta functions of algebraic curves, II:
Independence of the zeros
Using the sieve for Frobenius, we show that there are "usually" no multiplicative or Qlinear relations between zeros of Lfunctions of algebraic curves over finite fields, which is an analogue of wellknown linear independence conjectures for zeros of Dirichlet Lfunctions and has consequences, for instance, for determining the distribution of the difference between the number of points of two curves over extensions of the base field (this is an analogue of the Chebychev bias... except that here there is typically no bias). In the case (of most interest) of multiplicative relations, we outline an alternative approach, based on the theory of Frobenius tori of Serre (this was suggested by N. Katz). Also arXiv:0807.2118. 
30  Book
(Cambridge
University Press, 2008)

The large sieve and its applications
This is the expanded version of the preprint "The principle of the large sieve" (see arXiv:math.NT/0610021). The general sieve framework developed in this book is also described in a guest post on T. Tao's blog. Here are the current corrections, and a short addition that explains how to derive the arithmetic large sieve inequality from the "dual" inequality in the general setting of the book. 
29  Book translation
(Princeton
University Press, 2007) 
Mathematics for physics... and for physicists!
(English translation of «Mathématiques pour la physique... et les physiciens !», by Walter Appel, published by Éditions H&K) 
28 
Survey
(Association Bourbaki,
Astérisque,
2007)

Écarts successifs
entre nombres premiers (d'après Goldston, Pintz,
Yıldırım) Séminaire Bourbaki, Exposé 959 (2006). 
27  Popular survey
(BelinÉchelles,
2006)

Poincaré et la
théorie analytique des nombres (in «L'Héritage scientifique d'Henri Poincaré», edited by É. Charpentier, É. Ghys et A. Lesne); an English translation is published by the AMS, and here is my own version. 
26  Research paper
(Israel
J. Math., 2007)

Exponential
sums over definable subsets of finite fields
Also arXiv: math.NT/0504316 
25  Research paper
(Archiv
der Mathematik, 2007)

Équirépartition adélique de mesures algébriques dans un groupe résoluble et sommes de Kloosterman 
24  Research paper
(Int. J. of
Number Theory, 2006)

On
the rank of quadratic twists of elliptic curves over function
fields
Also arXiv: math.NT/0503732 
23  Research paper
(American
Math. Monthly, 2006)

Bernstein
polynomials and Brownian motion
Also see Problem 11155 in the May, 2005, issue of the Monthly (solution by T. Rivoal). Proposition 6.1 on zeros of random Bernstein polynomials is false! This was pointed out by Johannes Ruf; I was misremembering some properties of zeros of Brownian motion... 
22  Research paper
(Crelle,
2006)

The
large sieve, monodromy and zeta functions of curves
Also arXiv: math.NT/0503714. Here are a few small corrections (also contained in the forthcoming book "The large sieve and its applications"). 
21  Research paper
(Journal
of the L.M.S, 2006)

Weil
numbers generated by other Weil numbers and torsion fields of abelian
varieties
Also arXiv: math.NT/0504042. Note that Remark 3.9 (which is Remark 6 in Section 3 in the printed version) refers to a question (on the behavior of the splitting type of abelian varieties when reducing modulo primes) which was supposedly mentioned in an earlier paper (number 12), but in fact I had commentedout this remark during final corrections. 
20  Research paper
(Archiv
der Mathematik, 2005)

Variations of recognition problems for modular forms 
19  Book
(American
Math. Soc., 2004)

Analytic
Number Theory,
with
See also the Current list of corrections and (forthcoming) Supplementary material. There is also a Russian translation (published by MCCME, 2014). 
18  Survey
(Cambridge
University Press, 2007)

Elliptic curves, rank in families and random matrices
(Typed version of a survey lecture given at the Newton Institute Workshop and random matrices and Lfunctions in July 2004, together with a survey lecture on the variation of the rank of elliptic curves in families at the AIM/Princeton University workshop on the Birch and SwinnertonDyer Conjecture in November 2003) 
17  Research paper 
Small gaps in coefficients of Lfunctions and Bfree
numbers in short intervals,
with
and
Also arXiv: math.NT/0507001. Updated in November 2019, to correct a minor mistake in Lemma 2.2 pointed out by Narasimha Kumar; this does not affect the main results. 
16  Research note
(American
Math. Monthly, 2004)

On the reducibility of arctangents of integers
See also Sequence A002312 in the Encyclopedia of Integer Sequences. (This is a short application of the result of Duke, Friedlander and Iwaniec proved in the graduate course below). 
15  Book
(SMF
Cours Spécialisés, 2004)

Un cours de théorie analytique des nombres
(DEA/graduate course in Bordeaux, 20012002). 
14  Book chapters
(Birkhäuser,
2003)

Automorphic forms, Lfunctions and number theory: 3
lectures
See some corrections to the published version. 
13  Preprint (2003)  Dependency on the group in automorphic Sobolev inequalities 
12  Research paper
(Manuscripta
Math., 2003)

Some
LocalGlobal Applications of Kummer Theory
Note that Proposition 6.11 is not in the published version 
11  Research paper
(Pacific J. Math.,
2002).

Zeros of families of automorphic Lfunctions close to 1, with 
10  Research paper
(J. Ramanujan
Math. Society, 2006)

Analytic
problems for elliptic curves
Also arXiv: math.NT/0510197 
9  Research paper
(Duke Math. J.,
2002)

RankinSelberg Lfunctions in the level aspect, with and 
8  Research paper
(Invent.
math., 2000)

Mollification of the fourth moment of automorphic Lfunctions and arithmetic applications, with and 
7  Published
(Israel J.
Math., 2000)

Explicit upper bound for the (analytic) rank of J_{0}(q), with 
6  Research paper
(Manuscripta
Math., 2001)

Deux théorèmes de nonannulation de valeurs spéciales de fonctions L, with 
5  Appendix
(Duke Math. J.,
2001)

Vérification
de l'hypothèse H_{p}(chi)
pour p grand,
with
Appendix to a paper by L. Merel 
4  Research paper
(Crelle,
2000)

Nonvanishing of high derivatives of automorphic Lfunctions at the center of the critical strip, with and 
3  Research paper
(Acta
Arith., 2000)

A lower bound for the rank of J_{0}(q), with 
2  Research paper
(Duke Math. J.,
2000)

The analytic rank of J_{0}(q) and zeros of automorphic Lfunctions, with 
1  Research paper
(Invent.
math., 2000) 
A problem of Linnik for elliptic curves and mean value estimates for automorphic representations, with and 
Last update 29.3.2020 by E. Kowalski