AI TrackerFrontier desk · AEST
Back to news
normalaitech

Bi-FORK: Generative Modeling of High-Dimensional Bifurcating Systems

arXiv cs.AI · 9 Oct 2026, 03:58 AEST

Bi-FORK is a generative framework designed to learn one-to-many solution maps in high-dimensional bifurcating systems. It effectively generates complete trajectories and recovers distinct solution branches, significantly advancing the field of generative modeling in physical systems.

Key facts

Read aloud

Bi-FORK is a generative framework designed to learn one-to-many solution maps in high-dimensional bifurcating systems. It effectively generates complete trajectories and recovers distinct solution branches, significantly advancing the field of generative modeling in physical systems. Bifurcations are ubiquitous in physical systems, from structural buckling to fluid and climate dynamics, yet they remain largely unexplored in deep learning. At a symmetry-breaking bifurcation, a single input admits multiple equally valid solutions, violating the one-to-one assumption underlying most learned physical surrogates. Bi-FORK generates complete trajectories through latent flow matching, preserving space and time coherence, and uses repulsion-guided sampling to recover distinct solution branches in a single amortized pass. We evaluate Bi-FORK on buckling beams, mechanical metamaterials, and Allen-Cahn phase separation, spanning continuous, discrete, and field-valued bifurcations with discretizations up to two hundred sixty thousand points. Bi-FORK recovers the multimodal solution structure while scaling several orders of magnitude beyond prior approaches, opening generative modeling to high-dimensional bifurcating physical systems.

Sources in this cluster