The paper investigates a special kind of rigidity in the geometry of certain mathematical objects called Calabi-Yau varieties, which are multi-dimensional shapes that appear in string theory and algebraic geometry. The central question is about "rank two attractors," which are special parameter values where the geometry of a family of these shapes becomes particularly constrained. The paper works in the setting of p-adic numbers, a number system built around a prime number p that gives a different notion of distance than ordinary real numbers. The authors study how a certain transformation called the Frobenius acts on families of Calabi-Yau varieties and ask: at which special parameter values does this transformation stay fixed, rather than moving things around?
The main technical result gives a concrete criterion, phrased in terms of cohomology (an algebraic tool for detecting geometric structure), that guarantees several things simultaneously. If a certain two-dimensional piece of the p-adic cohomology of a Calabi-Yau variety has the right stability properties, then a specific geometric line within that cohomology is preserved by the Frobenius, and remarkably precise numerical congruences called supercongruences hold. These supercongruences say that certain infinite series, when evaluated at special parameter values, agree with algebraic numbers modulo very high powers of p. The paper also connects several different constructions of canonical lifts, showing they should agree at exactly these special fixed points.
The authors then test their framework on concrete examples. Their main example is a particular family of Calabi-Yau threefolds called the Hulek-Verrill family, evaluated at a specific parameter value predicted to be a rank two attractor. Using an unproven but plausible conjecture due to Dummigan, they prove that the Frobenius fixes this parameter, the supercongruences hold, and several independently constructed canonical objects all coincide there. A second example involves a well-known family of K3 surfaces (two-dimensional Calabi-Yau shapes) with complex multiplication. The paper also settles a clean standalone result: a natural map on a simple p-adic geometric object called a split formal torus has essentially one fixed point in a natural domain, confirming a basic but useful rigidity property.