[D] Looking for arXiv endorsement (cs.LG) - PDE-based world model paper

Hi everyone, I'm a researcher looking for an arXi endorsement for cs. LG to submit my first paper. I've been working for about a year on FluidWorld, a world model where the prediction engine is a reaction-difffusion PDE instead of attention. The Laplacian diffusion handles spatial propagation, learned reaction terms do the nonlinear mixing, and the PDE integration itself produces the prediction. No attention, no KV-cache, O(N) complexity, 867K parameters total.