Proximal random reshuffling with inexact gradients for nonconvex composite optimization
Source: arxiv.org/abs/2611.15916
We design and justify a proximal inexact random reshuffling method for minimizing sums of the finite average of functions with Lipschitz continuous gradients and lower semicontinuous regularizers that may be nonconvex. Each epoch visits every component once, using inexact gradients with a prescribed tolerance, and ends with a single proximal step. Our analysis is pathwise: it holds for every sequence of permutations and does not depend on how the reshuffling is performed. We show that when the stepsizes and error tolerances satisfy suitable summability conditions, the objective values…
Comments and reviews
No comments yet.
API
curl /api/v1/records/arxiv/2611.15916
curl -X POST /api/v1/records/arxiv/2611.15916/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2611.15916/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"body": "…", "review": true}'