The paper studies a specific family of algebraic curves, which are geometric objects defined by polynomial equations relating two variables. The particular curves here have the form y cubed equals x to a certain power plus x, where the exponent and the field of definition depend on a prime number p and a parameter r. The authors work over finite fields, meaning fields that contain only finitely many numbers, and they count exactly how many points on the curve have coordinates in a specific finite field. Getting exact counts like this is a central problem in arithmetic geometry with connections to cryptography and coding theory.
The main result is a precise formula for the number of points on these curves. What makes this formula striking is that it places the curves at one of two extremes: when r is odd, the curve is "maximal," meaning it has as many points as theoretically possible given its complexity (measured by a quantity called the genus), and when r is even, it is "minimal," meaning it has as few points as possible. Maximal curves are especially prized because they are analogous to the famous Hermitian curve and have applications in constructing good error-correcting codes.
The proof works by transforming the curve into a more tractable form called a Kummer model, then counting points using classical number-theoretic tools called Jacobi sums and Gauss sums. These are certain weighted sums involving roots of unity that encode deep arithmetic information about the curve. A key step is connecting the Gauss sums back to the much older and better-understood Fermat cubic curve, which allows the authors to pin down a crucial sign ambiguity. The simplest concrete example of their result is the curve y cubed equals x to the 13th plus x over a field with 5 to the 6th power elements, which turns out to be maximal, and this is just the first instance of an infinite family sharing the same extremal property.