{"id":"arxiv/2606.13825","url":"https://arxiv.org/abs/2606.13825","tags":["math.OC","cs.LG"],"score":0,"title":"Scalable Deep Unfolding of Conic Optimizers","domain":"math.OC","source":"arxiv","authors":["Alex Oshin","Rahul Vodeb Ghosh","Evangelos A. Theodorou"],"reviews":0,"subject":"math.OC","summary":"Deep unfolding (DU) accelerates iterative optimizers by introducing learnable components and training them through unrolled iterations, but extending DU to the large-scale semidefinite programs (SDPs) common in robotics has remained limited. Unrolling a full-update conic solver such as COSMO exposes two obstacles that prior work on learned conic solvers has not: backpropagating through the per-iteration linear-system solve incurs memory quadratic in the problem size once the coefficient matrix is formed explicitly, and backpropagating through the positive semidefinite (PSD) cone projection…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-06-11","reactions":{}}