The paper resolves a conjecture about certain mathematical objects called modular forms that appeared in a recent study of a family of geometric shapes known as K3 surfaces. Modular forms are functions with rich symmetry properties that show up across number theory, geometry, and even theoretical physics. The conjecture, made by Bönisch, Duhr, and Maggio, said that three specific modular forms called C4, C6a, and C6b are "magnetic," meaning that when you expand them as power series and look at the coefficients, dividing each coefficient by a certain power of its index always gives an integer. The paper proves something even stronger: the denominators are exactly one, not just bounded by some finite number. Roughly speaking, the coefficients are more divisible than the conjecture required.
The proof works differently for the two weights involved. For the weight-four form C4, the authors find a clever change of variables, essentially a substitution that transforms one natural coordinate into another, under which the divisibility becomes a direct consequence of a well-known property of binomial coefficients. For the two weight-six forms C6a and C6b, the key insight is identifying them with classical objects tied to complex multiplication, a special algebraic structure that makes certain modular forms particularly well-behaved. Once this identification is made, a general theorem about how modular forms of half-integer weight interact with integer-weight forms, via a construction called a theta lift, delivers the required divisibility at all odd primes essentially for free.
The remaining challenge is handling divisibility by the prime 2 separately, since it does not follow from the general machinery. Here the authors introduce a specific infinite family of power series and prove that a certain averaging operator, called the U2 operator, multiplies divisibility by 2 very rapidly as you pass to higher powers of 2. The technical result is that the 2-adic valuation of the coefficients grows at least five times as fast as the index of the power of 2, which is much faster than the growth rate that magneticity requires. Combining all three pieces, divisibility at odd primes via the theta lift, divisibility at 2 via the power series contraction, and the weight-four binomial argument, the authors conclude that the full collection of modular forms associated to this K3 family is magnetic with no denominators at all.