Introduces the sketched landing method using Gaussian or subsampling sketch matrices to lower per-iteration cost in orthogonality-constrained optimization while preserving convergence guarantees in expectation.
A randomized feasible algorithm for optimization with orthogonal constraints,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
The sketched landing method for large-scale optimization under orthogonality constraints
Introduces the sketched landing method using Gaussian or subsampling sketch matrices to lower per-iteration cost in orthogonality-constrained optimization while preserving convergence guarantees in expectation.