Releases: JuliaManifolds/Manopt.jl
Releases · JuliaManifolds/Manopt.jl
v0.4.54
Manopt v0.4.54
Added
convex_bundle_method
optimization algorithm for non-smooth geodesically convex functionsproximal_bundle_method
optimization algorithm for non-smooth functions.StopWhenSubgradientNormLess
,StopWhenLagrangeMultiplierLess
, and stopping criteria.
Fixed
- Doc strings now follow a vale.sh policy. Though this is not fully working,
this version however improves a lot of the doc strings concerning wording and spelling.
Merged pull requests:
- Introduce text checks (vale.sh) on Docstrings and comments (#353) (@kellertuer)
- Implementation of two bundle methods (#357) (@hajg-ijk)
v0.4.53
Manopt v0.4.53
Fixed
- fixes two storage action defaults, that accidentally still tried to initialize a
:Population
(as modified back to:Iterate
0.4.49). - fix a few typos in the documentation and add a reference for the subgradient menthod.
Merged pull requests:
- Kellertuer/fix storages (#356) (@kellertuer)
Closed issues:
- Add Warning if gradient larger than injectivity radius. (#346)
v0.4.52
Manopt v0.4.52
Added
- introduce an environment persistent way of setting global values with the
set_manopt_parameter!
function using Preferences.jl. - introduce such a value named
:Mode
to enable a"Tutorial"
mode that shall often provide more warnings and information for people getting started with optimisation on manifolds
Merged pull requests:
- Bump codecov/codecov-action from 3 to 4 (#355) (@dependabot[bot])
v0.4.51
Manopt v0.4.51
Added
- A
StopWhenSubgradientNormLess
stopping criterion for subgradient-based optimization.
Merged pull requests:
- Allow
d.msg
in DebugIfEntry to contain an identifier to print the field (#351) (@kellertuer) - Add StopWhenSubgradientNormLess stopping criterion (#352) (@hajg-ijk)
v0.4.50
Manopt v0.4.50
Fixed
- Fix Quasi Newton on complex manifolds.
Merged pull requests:
- Bump actions/cache from 3 to 4 (#348) (@dependabot[bot])
- Fix Quasi Newton for complex Manifolds (#350) (@kellertuer)
v0.4.49
Manopt v0.4.49
Added
A StopWhenEntryChangeLess to be able to stop on arbitrary small changes of specific fields
generalises StopWhenGradientNormLess to accept arbitrary norm= functions
refactor the default in particle_swarm to no longer “misuse” the iteration change check,
but actually the new one one the :swarm entry
Merged pull requests:
- Introduced an entry change stopping criterion and generalise the gradient norm SC. (#345) (@kellertuer)
v0.4.48
Manopt v0.4.48
Fixed
- fixes an imprecision in the interface of
get_iterate
that sometimes led to the swarm ofparticle_swarm
being returned as the iterate. - refactor
particle_swarm
in naming and access functions to avoid this also in the future.
To access the whole swarm, one now should useget_manopt_parameter(pss, :Population)
Merged pull requests:
- Differentiate more precisely bewteen swarm and iterate in PSO more stricly (#344) (@kellertuer)
v0.4.47
Manopt v0.4.47
Fixed
- fixed a bug, where the retraction set in
check_Hessian
was not passed on to the optional innercheck_gradient
call, which could lead to unwanted side effects, see #342.
Merged pull requests:
- Fix a bug where the retraction method in
check_Hessian
was not passed on (#343) (@kellertuer)
Closed issues:
v0.4.46
Manopt v0.4.46
Changed
- An error is thrown when a line search from
LineSearches.jl
reports search failure. - Changed default stopping criterion in ALM algorithm to mitigate an issue occurring when step size is very small.
- Default memory length in default ALM subsolver is now capped at manifold dimension.
- Replaced CI testing on Julia 1.8 with testing on Julia 1.10.
Fixed
- A bug in
LineSearches.jl
extension leading to slower convergence. - Fixed a bug in L-BFGS related to memory storage, which caused significantly slower convergence.
Merged pull requests:
- Fixes from benchmark (#341) (@mateuszbaran)
v0.4.45
Manopt v0.4.45
Manopt.jl 0.4.45
Added
- Introduce
sub_kwargs
andsub_stopping_criterion
fortrust_regions
as noticed in #336
Changed
WolfePowellLineSearch
,ArmijoLineSearch
step sizes now allocate lesslinesearch_backtrack!
is now available- Quasi Newton Updates can work inplace of a direction vector as well.
- Faster
safe_indices
in L-BFGS.
Merged pull requests:
- Faster
safe_indices
(#334) (@mateuszbaran) - Improve Linesearch and Quasi Newton allocations (#335) (@kellertuer)
- Introduce
sub_kwargs
also for Trust Regions (#337) (@kellertuer)
Closed issues:
- Improving performance of L-BFGS (#333)