The paper studies what happens when you repeatedly apply mathematical transformations called endomorphisms to geometric objects called varieties. An endomorphism is a map from a mathematical object back to itself, and when you compose several such maps together, you get a semigroup, which is essentially a collection of maps you can combine by applying one after another. The central question is: how fast do certain numerical measures of complexity, specifically the "degree" and "height" of these maps, grow as you apply longer and longer compositions of the generators?
The authors focus on a special class of semigroups called "partially commutative," meaning the only rule governing how generators interact is that some pairs of them can be swapped in order without changing the result. This is a natural middle ground between fully free semigroups, where order always matters, and fully commutative ones, where it never does. The key technical achievement is a general theorem that connects the growth rate of a multiplicative function (one that behaves like multiplication under composition) to another nonnegative function, provided a condition called "Tate telescoping" holds. This condition is borrowed from the theory of arithmetic dynamics and essentially ensures that the degree or height behaves predictably under iteration.
On the applied side, the authors construct new explicit families of maps to which their theoretical results apply. To do this, they extend a classical tool called the "ping-pong lemma," which is normally used to show that individual maps generate a free group or semigroup by having them act on separate regions of a space. The authors generalize this to work with entire subsemigroups rather than individual elements, which gives them a more flexible way to verify that complicated collections of maps have the free or partially commutative structure needed for their growth rate formulas to kick in.