Modular hyperbolas in context

This book studies the lattice sets $$ A_{N,a}=\{(x,y)\in \{1,\ldots,N-1\}^2 : xy \equiv a \pmod N\} $$ and the convex hulls they determine. For fixed modulus and fixed residue, such a set is already a familiar object in the number-theory literature: it is usually called a modular hyperbola.

We will use that phrase when it helps connect MTMN to earlier work. At the same time, we will keep the name MTMN for the project-wide point of view of this book, because the present subject is not only one congruence class at a time. It is the geometry of the whole residue family $$ \{A_{N,0},A_{N,1},\ldots,A_{N,N-1}\} $$ inside one multiplication table.

This chapter is only a guide to mathematical placement. It is not meant to be a full survey.

What earlier work already studies

There is already a substantial literature on modular hyperbolas. A convenient entry point is Igor E. Shparlinski's survey Modular hyperbolas [@shparlinski2012], which collects many of the main themes and references. At a broad level, that literature studies how the points satisfying $$ xy \equiv a \pmod m $$ are distributed, how they cluster inside boxes, how they behave geometrically, and how their behavior changes in higher-dimensional analogues. In that literature the modulus is often written as $m$ (and sometimes later as $n$); throughout this book it is the same modulus that we denote by $N$.

Some of those questions are close to the concerns of this book. Convex-hull questions, for example, are not new here. Konyagin and Shparlinski [@konyaginshparlinski2010] study the number of vertices of convex hulls of points on modular hyperbolas. Ford, Khan, and Shparlinski [@fordkhanshparlinski2010] study geometric properties of the unit modular hyperbola $$ xy \equiv 1 \pmod n $$ and obtain bounds connected with extremal points and convex-hull behavior. Other work studies concentration of points in small regions [@cilleruelogaraev2011], distances or coordinate differences [@shparlinskiwinterhof2008], and multidimensional modular hyperbolas [@shparlinski2007].

Read through the lens of Shparlinski's survey, the standard output of this literature is usually asymptotic rather than exact [@shparlinski2012]. One fixes a single congruence class, most often in the coprime regime $\gcd(a,m)=1$, and then asks for point counts in boxes, discrepancy estimates, concentration bounds, or upper and lower bounds for geometric statistics such as vertex counts, coordinate differences, or distances. The characteristic tools are exponential sums, especially Kloosterman sums, together with discrepancy methods and other analytic estimates. Even when the language is geometric, the aim is typically to understand the large-scale distribution of one modular hyperbola rather than to write exact formulas for a whole residue family.

There is also nearby literature on primitive or visible points, meaning lattice points with coprime coordinates and hence visible from the origin, for congruence-defined curves over finite fields [@shparlinskivoloch2007]. Those papers do not always use exactly the same MTMN setting, but they show that coprimality already carries its own arithmetic-geometric language in the surrounding literature.

The usual geometric realization

Most of this literature fixes a single congruence class and studies it in the usual positive residue window, typically inside a box such as $$ \{0,1,\ldots,m-1\}^2 $$ or $$ \{1,\ldots,m-1\}^2. $$ That is a natural first realization, and it is exactly the positive lattice window used in the main chapters of this book.

Another common feature is the emphasis on the coprime, or invertible, case $$ \gcd(a,m)=1. $$ In that regime both coordinates are units modulo $m$, and a large body of analytic and combinatorial machinery becomes available. This case is mathematically important, and it remains part of the background of MTMN as well.

How the present book differs

The distinctive emphasis of this book lies elsewhere. Rather than isolating one congruence class and one geometric statistic, it keeps the full residue family in view and compares classes residue by residue. Its main numerical quantities are the convex-hull areas $$ S(N,a)=\operatorname{Area}(\operatorname{conv}(A_{N,a})) $$ and $$ S(N)=\sum_{a=0}^{N-1} S(N,a). $$ This already shifts the question from the geometry of one modular hyperbola to the geometry of the whole multiplication table.

The book also gives special attention to the zero class and to non-coprime behavior. On that side, factorization and zero divisors create geometry that is invisible if one looks only at the unit case. The divisor-rectangle description of the zero class is the clearest example: there the hull can be read directly from the divisors of $N$, so the arithmetic of compositeness becomes part of the visible Euclidean structure.

The difference is therefore mainly one of emphasis. Much of the earlier literature studies one usually coprime residue class at a time and extracts asymptotic information from exponential-sum estimates. MTMN focuses on exact discrete geometric formulas in the fixed positive lattice window, treats all residues simultaneously, and includes the zero-divisor side from the start. In particular, exact area formulas, exact boundary-layer models, and the divisor-controlled hull of $A_{N,0}$ are not the usual targets of the analytic modular-hyperbola program even though they begin from the same congruence-defined point sets.

Later chapters do not replace that window by a competing one. Instead they use the same positive embedding to study divisor envelopes, support functions, and boundary layers.

So the purpose of this book is not to reintroduce a known object under a new name. Its purpose is to develop a residue-family, area-centered, and composite-sensitive program around an object that earlier literature has already studied from other angles.