Radial Kernel Attention for Polymer ML Force Fields
What it is
This paper introduces Radial Kernel Attention (RKA), a method for long-range modeling in polymer machine learning force fields that achieves linear computational complexity. It maps interatomic distances to learned radial features and combines them with content-dependent queries and keys to enable global feature exchange without quadratic costs.
The problem it solves
Existing local neighbor graph approaches limit communication across polymer chains, while explicit all-pair attention incurs prohibitive quadratic computational costs for large systems.
What you could use it for
- Simulating large polymer systems requiring long-range interactions
- Training machine learning force fields with reduced computational overhead
- Improving energy and force prediction accuracy in polymer dynamics
Why it matters
Enables efficient, accurate simulation of large-scale polymer systems by overcoming the quadratic cost of standard attention mechanisms while maintaining physical consistency.
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.