KL-Guided Support Selection and Subspace Cubic Newton Methods for Sparse Optimization
Source: arxiv.org/abs/2611.22940
We study sparse composite optimization with a smooth loss and a possibly nonconvex, nonsmooth regularizer. The objective is locally smooth on each fixed support. Our method combines full-space proximal-gradient steps with inexact cubic-regularized Newton steps on a subspace selected by thresholding. If the iterates are bounded and the objective satisfies the Kurdyka--Łojasiewicz (KL) property, the entire sequence converges to a first-order stationary point. A threshold based on the KL exponent ensures that the selected coordinate set coincides with the support of the limit after finitely…
Comments and reviews
No comments yet.
API
curl /api/v1/records/arxiv/2611.22940
curl -X POST /api/v1/records/arxiv/2611.22940/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2611.22940/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"body": "…", "review": true}'