Fast HOSVD via Randomized Preconditioning and Mixed Precision
Source: arxiv.org/abs/2611.21049
Computing dominant mode subspaces is a major cost in the higher-order singular value decomposition (HOSVD) and its sequentially truncated variant (STHOSVD). Randomized Tucker algorithms can reduce this cost by approximating the required subspaces from small sketches, but may introduce additional approximation errors. We exploit the extreme aspect ratios of tensor unfoldings, whose mode dimensions can remain manageable while being orders of magnitude smaller than the products of the remaining dimensions. Instead of directly approximating the desired subspaces, we sketch the entire unfolding…
Comments and reviews
No comments yet.
API
curl /api/v1/records/arxiv/2611.21049
curl -X POST /api/v1/records/arxiv/2611.21049/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2611.21049/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"body": "…", "review": true}'