Releases: kul-optec/alpaqa
Releases · kul-optec/alpaqa
1.0.0a9
- Improved CUTEst problem loader and adapter
- Return upper-triangular Hessian CUTEst adapter instead of lower (API break)
- QPALM solver interface, added to
alpaqa-driver
- Sparse logistic regression example
- Improve dynamic loading API, support exception propagation (ABI break)
- Configurable float types for smaller binaries, faster compile times
- Support modifying
alpaqa::DefaultConfig
- Add optional in/out parameter penalties ALM
- More PANTR statistics
- Improved
UnconstrProblem
, expose in Python API - Add
alpaqa-gradient-checker
utility - Improved documentation
- Updated installation instructions
- Improved CI with Valgrind testing
- More testing for PANOC with Anderson acceleration
- Improved Python packaging and metadata
Full Changelog: 1.0.0a8...1.0.0a9
1.0.0a8
- Intermediate development release
- CUTEst interface
- Change sparse matrix API
- Use cibuildwheel for packaging
- Add .deb package dependencies
Full Changelog: 1.0.0a7...1.0.0a8
1.0.0a7
- Intermediate development release
Full Changelog: 1.0.0a6...1.0.0a7
1.0.0a6
- New type-erased problem formulation
- PANOC solver specifically for optimal control problems
- Gauss-Newton accelerator
Full Changelog: 1.0.0a5...1.0.0a6
1.0.0a5
- Python: Support releasing the GIL during solve, allowing thread parallelism and handling KeyboardInterrupt.
Full Changelog: 1.0.0a4...1.0.0a5
1.0.0a4
- Complete ProblemTrampoline Python interface
- Flush after printing solver progress
Full Changelog: 1.0.0a3...1.0.0a4
1.0.0a3
- Change default value of
ALMParams::max_*_num_retries
to zero - Add
final_γ
statistic to PANOC
Full Changelog: 1.0.0a2...1.0.0a3
1.0.0a2
Full Changelog: 1.0.0a1...1.0.0a2