Frobenius, History, and the Resolution of Singularities
This presentation explores a new framework for resolving singularities in positive characteristic, where classical numerical invariants fail. The authors replace single-number descent with a rich local packet incorporating differential, Frobenius, and exceptional-divisor data, then prove termination through well-founded descent on historically addressed occurrences. The result is a claimed canonical, functorial resolution algorithm that preserves geometric structure throughout.Script
In positive characteristic, the classical approach to resolution breaks down because a single numerical invariant cannot capture the behavior of singularities under Frobenius operations and inseparable phenomena. This paper offers a bold alternative: replace numerical descent with historical descent, tracking the ancestry of exceptional divisors and using a structured local packet to guide termination.
The local theory starts with differential integral saturation, which enriches a marked Rees algebra with total Hasse operators and coefficient cubes. This makes the invariant robust under the purely inseparable behavior and vanishing derivatives that plague characteristic p, capturing information invisible to ordinary order.
The authors construct Frobenius Hasse towers that organize singularities into a finite hierarchy of heights. Semilinear sources encode the interaction between additive structure and differential data, allowing the algorithm to expose defects at the appropriate level rather than forcing everything into a single scalar.
Exceptional history becomes a first-class invariant. Each divisor carries finite ancestry data, including its owner, parent, and trace information, which survives transformations and prevents the algorithm from losing track of configurations that reopen after blowup. This is not metadata; it is the state of the resolution itself.
Defect complexes are routed to specialized backends: surface, toroidal monomial, binomial, and additive type. Paid handoffs ensure that transferring a defect between backends is tracked and that no reclassification evades the global descent mechanism, preventing modularity from undermining termination.
Termination follows from well-founded descent on addressed occurrences, not from a pointwise scalar. Each completed macroblock replaces active parent occurrences by strict descendants in a dependent order, and the six realization components ensure the certificate survives every algorithmic transition. To dive deeper into resolution in positive characteristic and create your own explainer videos, visit EmergentMind.com.
