The paper demonstrates that a surprisingly simple set of mathematical operations is enough to perform arbitrary computation. Specifically, the author shows that two operations, raising numbers to powers and taking logarithms, are sufficient to simulate any "tag system," which is a type of simple rule-based symbol manipulation process that is known to be capable of universal computation (meaning it can, in principle, simulate any computer program). The key trick is encoding the current state of the computation as a single integer, and writing a single mathematical expression that, when evaluated, produces the integer encoding the next step. Feeding the output back into the same expression then chains together an entire computation.
The clever insight that makes this work is exploiting a subtle property of the complex logarithm. When you take logarithms in the complex number system, the result "wraps around" periodically, similar to how a clock resets after 12 hours. The author repurposes this wraparound behavior to perform the kind of rounding, remainder, and selection operations that are normally needed in step-by-step computation. Earlier work on encoding computation into mathematical expressions typically relied on functions like the floor function, trigonometric functions like sine, or explicit remainder operations. Here, all of that is replaced by just powers and logarithms, with no additional constants needed beyond what appears in the expression itself.
One concrete consequence of this result is that a specific decision problem becomes provably unsolvable. Because the expression can simulate arbitrary computation, there is no general algorithm that can look at the expression and a threshold number and decide whether the sequence of values produced by repeatedly applying the expression will ever dip below that threshold. As a bonus result, the same mathematical machinery can also represent basic building blocks of computation like addition, exponential growth, and remainder operations, which means it can express any function in a well-known class called the Kalmar elementary functions, a broad and practically important family of computable functions.