NEWS
lazymatrix 0.1.0 (2026-07-14)
Initial CRAN submission.
New Features
- Introduced lazy evaluation framework for working with normalized sparse matrices
- Represent large sparse matrices symbolically with the S4 object
LazyMatrix
- Fundamental matrix operations like matrix-vector multiplication and transpose matrix-vector multiplication supported for the
LazyMatrix object
- Added support for subsetting methods, mirroring matrix objects from base R and
Matrix. The symbolic equivalent of column vectors is called LazyColumn
- Implemented specific statistical algorithms, LSQR, truncated partial SVD, for showcasing how
lazymatrix can be implemented
- Optimized internal C++ loops for sparse matrix operations using
RcppArmadillo.
Improvements
- Optimized matrix multiplication for lazy objects to avoid materialization
- Added vignette demonstrating typical workflow with large sparse datasets
Bug Fixes
- No known bugs fixed in initial release
Documentation
- Added comprehensive
roxygen2 documentation for all exported functions
- Included example code in each function's documentation
- Wrote getting started vignette covering installation and basic usage
Internal Changes
- Utilized Matrix package for underlying sparse matrix representation
- Implemented partial SVD using
irlba