Automating the Derivation of Advanced Algorithms in Numerical Linear Algebra using Associative Algebras with Involution
Source: arxiv.org/abs/2611.14831
Standard algorithms of numerical linear algebra use their scalars through seven operations: addition, subtraction, multiplication, division, the adjoint, the $*$-square root and $(i,j)$ indexing. This thesis overloads these operations, as automatic differentiation overloads arithmetic, allowing scalars in associative algebras with involution. Many of the resulting decompositions and algorithms reduce to Hermitian eigendecomposition over these $*$-algebras. The decompositions are organised by doubly-signed quivers, whose vertices record real, complex or quaternionic scalars and the action of…
Comments and reviews
No comments yet.
API
curl /api/v1/records/arxiv/2611.14831
curl -X POST /api/v1/records/arxiv/2611.14831/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2611.14831/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"body": "…", "review": true}'