Set theory studies infinite collections and the different "sizes" of infinity. Beyond the countable infinity and the uncountable infinity, mathematicians have discovered that certain natural collections of functions or sets can have intermediate sizes that are not pinned down by the standard axioms of mathematics. These sizes are called cardinal characteristics, and understanding how they relate to each other is a central goal in set theory. Two important ones are called "d" and "a". The characteristic "d" measures how many functions on the natural numbers you need so that every other such function is eventually overtaken by one of them. The characteristic "a" measures the smallest size of a maximal almost disjoint family, which is a collection of infinite sets of natural numbers where any two sets in the collection overlap in only finitely many elements, and no new infinite set can be added while keeping that property.
Since the 1970s, a question posed by Judith Roitman had remained open: does having "d" equal the smallest uncountable infinity force "a" to equal that same infinity? The concern was that "a" might somehow be larger than "d", meaning you need more sets to build a maximal almost disjoint family than you need to dominate all functions. This paper proves, using only the standard axioms of mathematics, that the answer to Roitman's question is yes: if "d" is as small as possible among uncountable cardinals, then "a" must be equally small. This also answers a related question from Shelah by showing that a particular kind of model where "d" is strictly less than "a" cannot exist unless the total size of the real numbers is at least the third uncountable infinity.
The proof achieves more than just settling the inequality. It constructs a maximal almost disjoint family that is "indestructible" in a precise sense: the family stays maximal even after you extend the mathematical universe using certain forcing techniques, which are tools set theorists use to build new models of mathematics. This indestructibility holds whenever the dominating family that drives the construction remains dominating in the extended universe. As a bonus, the family can be chosen to have additional robustness properties, resolving several further open questions about how maximal almost disjoint families behave under forcing extensions.