linear-optimization

This paper presents a deterministic algorithm for online inverse linear optimization with O(d) regret and O(d^2) time per round, marking the first efficient and proper bound of this kind, with the main result obtained using the Cogentic agentic framework and Gemini 3.1 Pro.

Modal explains how it reduces AI inference cold starts by 40x using cloud buffers, a custom filesystem, checkpoint/restore, and CUDA checkpoint/restore, framing cloud buffer management as a linear optimization problem solved with GLOP.