Skip to content

Releases: machines-in-motion/mim_solvers

Release 0.0.5

16 Aug 15:40
v0.0.5
5ee535f
Compare
Choose a tag to compare
  • Added callbacks (this changes the API, see the examples)
  • Added vectorization and malloc check options in CMakeLists
  • Added Github CI
  • Fixed bindings and renamed merit function parameters (mu_dynamic and mu_constraint)
  • Code optimization
  • Now checking stopping criteria for QP every 25 iterations in SolverCSQP (as a result, it is now more efficient to use SolverSQP for unconstrained problems)

Full Changelog: v0.0.4...v0.0.5

Release 0.0.4

03 Apr 18:44
v0.0.4
Compare
Choose a tag to compare
  • Fixed rho initialization in CSQP
  • Added callbacks in QP solvers
  • Replaces warm_start_y by reset_y.
  • By default, y and rho are not reset.

Full Changelog: v0.0.3...v0.0.4

Release 0.0.3

27 Jan 00:24
v0.0.3
Compare
Choose a tag to compare

What's Changed

  • Added unittest infrastructure in python and C++
  • Added C++ benchmarks of the CSQP and SQP solvers
  • Optimized code speed for CSQP and SQP solvers
  • Use systematically KKT residual stopping criteria in all solvers
  • Deprecation warning of DDP and FDDP solvers
  • Inform user not to use equality constraint API of Crocoddyl
  • Updated python implementations of the solvers
  • Updated regularization variables consistently with Crocoddyl 2.0.2
  • Fixed bug when relative residual tolerance is set to 0 in stagewise QP solver
  • Refactored python code (now separated from python bindings)

Full Changelog: v0.0.2...v0.0.3

Release 0.0.2

10 Dec 14:48
v0.0.2
4d9e409
Compare
Choose a tag to compare

What's Changed

New Contributors

Full Changelog: v0.0.1...v0.0.2

mim_solvers v0.0.1

30 Nov 11:14
v0.0.1
Compare
Choose a tag to compare

Implementation of numerical solvers used in the Machines in Motion Laboratory