AI RESEARCH
A Unifying Framework for Parallelizing Sequential Models with Linear Dynamical Systems
arXiv CS.LG
•
ArXi:2509.21716v2 Announce Type: replace Harnessing parallelism in seemingly sequential models is a central challenge for modern machine learning. Several approaches have been proposed for evaluating sequential processes in parallel using iterative fixed-point methods, like Newton, Picard, and Jacobi iterations. In this work, we show that these methods can be understood within a common framework based on linear dynamical systems (LDSs), where different iteration schemes arise naturally as approximate linearizations of a nonlinear recursion.