Releases: JuliaManifolds/Manopt.jl
Releases · JuliaManifolds/Manopt.jl
v0.5.17
Manopt v0.5.17
Added
- Introduce a
StopWhenCostChangeLess
stopping criterion that stops when the cost function changes less than a given value.
Merged pull requests:
- CompatHelper: bump compat for Documenter to 1 for package docs, (keep existing compat) (#472) (@github-actions[bot])
- CompatHelper: bump compat for RipQP in [weakdeps] to 0.7, (keep existing compat) (#474) (@github-actions[bot])
- CompatHelper: bump compat for RipQP to 0.7 for package docs, (keep existing compat) (#475) (@github-actions[bot])
- add a new stopping criterion. (#476) (@kellertuer)
v0.5.16
Manopt v0.5.16
Fixed
- fixes a bug in the
LineSearches.jl
extension, where two (old)retract!
s were still
present; they were changed toretact_fused!
.
Merged pull requests:
- Inplace version of vectorize and reshape (#468) (@blegat)
- Fix retracts in LineSearches extension (#471) (@kellertuer)
Closed issues:
- Non-array manifold with JuMP/MOI (#274)
v0.5.15
Manopt v0.5.15
Merged pull requests:
- Fix CMA-ES with nonpositive covariance eigenvalues (#467) (@mateuszbaran)
v0.5.14
Manopt v0.5.14
Merged pull requests:
- Fixes printing of JuMP models containg Manopt solver (#462) (@blegat)
- add linear subsolver to LevenbergMarquardt API (#463) (@mateuszbaran)
- Adapt to the new
default_basis
function. (#464) (@kellertuer) - Switch to using QuartoNotebookRunner (#465) (@kellertuer)
v0.5.13
Manopt v0.5.13
Merged pull requests:
- Add support for Riemanian objective through JuMP (#448) (@blegat)
- Remove leftover from merge (#452) (@blegat)
- Use scaled objective (#453) (@blegat)
- Wrap objective in JuMP wrapper (#455) (@blegat)
- Do not scale the cost function twice. (#457) (@kellertuer)
- fixing issues in quasi-Newton docs (#458) (@mateuszbaran)
- Fix a typo. (#459) (@kellertuer)
- Refactor Tests a bit (#460) (@kellertuer)
- Fix a small bug in Quasi Newton initialisation (#461) (@kellertuer)
Closed issues:
v0.5.12
Manopt v0.5.12
Added
- a
ScaledManifoldObjective
to easier build scaled versions of objectives,
especially turn maximisation problems into minimisation ones using a scaling of-1
. - Introduce a
ManifoldConstrainedSetObjective
- Introduce a
projected_gradient_method
Merged pull requests:
- Help with a small objective wrapper to scale objectives (#449) (@kellertuer)
- minor improvements for scaled objective (#450) (@mateuszbaran)
- Add a projected gradient method. (#451) (@kellertuer)
v0.5.11
Manopt v0.5.11
Added
- Configurable subsolver for the linear subproblem in Levenberg-Marquardt. The default subsolver is now also robust to numerical issues that may cause Cholesky decomposition to fail.
Merged pull requests:
- Customizable subsolver for Levenberg-Marquardt (#447) (@mateuszbaran)
Closed issues:
- Small example of inaccurate tracing retraction. (#446)
v0.5.10
Manopt v0.5.10
Fixed
- a proper implementation of the preconditioning for
quasi_Newton
, that can be used instead
of or in combination with the initial scaling.
Merged pull requests:
- Fix preconditioning for QN. (#445) (@kellertuer)
v0.5.9
Manopt v0.5.9
Added
- add a
PreconditionedDirection
variant to thedirection
gradient processor
keyword argument and its correspondingPreconditionedDirectionRule
- make the preconditioner available in quasi Newton.
- in
gradient_descent
andconjugate_gradient_descent
the rule can be added anyways.
Fixed
- the links in the AD tutorial are fixed and moved to using
extref
Merged pull requests:
- Fix links and move them to extref in the AD tutorial. (#443) (@kellertuer)
- Add a
PreconditionedDirection
direction rule (#444) (@kellertuer)
Closed issues:
- The link back to
change_representer
is not updated in use automatic differenentiatio tutorial (#442)
v0.5.8
Manopt v0.5.8
Fixed
- fixed a small bug in the
NonmonotoneLinesearchStepsize
hwn the injectivity radius is an irrational number. - fixed a small bug in
check_gradient
whereeps
might have been called on complex types. - fixed a bug in several gradient based solvers like
quasi_newton
, such that they properly work with the combined cost grad objective. - fixes a few typos in the docs.
Merged pull requests:
- Fix two small bugs. (#440) (@kellertuer)
- Fix a bug where QN did not accept CombinedCostGrad objectives. (#441) (@kellertuer)