Decomposing modular tensor products: `Jordan partitions', their parts and p-parts

TL;DR

This paper investigates the structure of Jordan partitions resulting from tensor products of Jordan blocks, proving conjectures about their p-parts and providing explicit formulas for certain cases, with implications for algebraic representation theory.

Contribution

It proves two longstanding conjectures about p-parts of Jordan partitions and establishes new relationships between their least common multiples and greatest common divisors.

Findings

Proved McFall's conjectures on p-parts of Jordan partitions.

Established that lcm(r,s) and gcd of the parts have equal p-parts.

Derived explicit formulas for Jordan partitions when p=2.

Abstract

Determining the Jordan canonical form of the tensor product of Jordan blocks has many applications including to the representation theory of algebraic groups, and to tilting modules. Although there are several algorithms for computing this decomposition in literature, it is difficult to predict the output of these algorithms. We call a decomposition of the form $J_r\otimes J_s=J_{\lambda_1}\oplus\cdots\oplus J_{\lambda_b}$ a `Jordan partition'. We prove several deep results concerning the $p$-parts of the $\lambda_i$ where $p$ is the characteristic of the underlying field. Our main results include the proof of two conjectures made by McFall in 1980, and the proof that ${\rm lcm}(r,s)$ and $\gcd(\lambda_1,\dots,\lambda_b)$ have equal $p$-parts. Finally, we establish some explicit formulas for Jordan partitions when $p=2$.