{"id":"arxiv/2611.03621","url":"https://arxiv.org/abs/2611.03621","tags":["math.NA","math.OC"],"score":0,"title":"Multi-Trajectory Parallel-in-Iteration Optimization","domain":"math.NA","source":"arxiv","authors":["Guilherme H. M. Araujo","Oliver A. Krzysik","Hans De Sterck"],"reviews":0,"subject":"math.NA","summary":"Previously, we developed a ``parallel-in-iteration'' optimization framework that uses multigrid reduction-in-time (MGRIT) to parallelize slowly converging optimization methods, such as gradient descent. Unlike time-dependent evolution problems, optimization requires only a final point with a sufficiently small gradient, rather than an accurate approximation of the entire trajectory of iterates. This motivates a different use of MGRIT: instead of repeatedly improving the approximation of a single long trajectory, we introduce restarted MGRIT (R-MGRIT), which performs one MGRIT iteration on a…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-11-02","reactions":{}}