# Changelog
All changes to L0Learn will be documented here, starting from v1.1.0 onwards.

The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).


## [2.1.0] - 2023-03-04
### Added
- More detailed documentation.

## [2.0.9] - 2023-02-24
### Fixed
- Update install config script.
- Compiler warnings.

## [2.0.8] - 2023-02-22
### Fixed
- Remove C++11 requirement.

## [2.0.7] - 2023-02-20
### Fixed
- Use smaller examples in the documentation.

## [2.0.6] - 2023-02-20
### Fixed
- Compiler warnings.

## [2.0.5] - 2022-02-13
### Fixed
- More accurate objective computation in CDPSI for logistic regression. Thanks to Jiachang Liu for the fix.

## [2.0.4] - 2021-05-30
### Added
- More detailed description of sparse matrix support in the Vignette.

## [2.0.3] - 2021-03-28
### Fixed
- UBSAN warning.

## [2.0.2] - 2021-03-18
### Fixed
- Compilation warnings (-Wreorder and -Wunused-lambda-capture).

## [2.0.1] - 2021-03-17
### Fixed
- Memory handling (to fix CRAN warnings).

## [2.0.0] - 2021-03-01
### Added
- Efficient handling of sparse data matrices.
- Ability to enforce bounds (box constraints) on the learned coefficients.
- Faster implementation of local search for logistic regression.
- Unit tests.
### Fixed
- Improved convergence criteria.

## [1.2.0] - 2019-08-30
### Added
- Support for Classification (Logistic Regression and Smoothed SVM).
### Fixed
- An edge case when y is constant and intercept=FALSE. Thanks to Nick Koning for reporting this.

## [1.1.1] - 2019-05-24
### Fixed
- Memory handling, especially for problems with very large n. Thanks to Will Nicholson for reporting the issue.

## [1.1.0] - 2019-04-07
### Fixed
- A bug in cvfit when supp size becomes equal to p.