Releases: aimclub/iOpt
Releases · aimclub/iOpt
v0.4.0
We're making a new release with a number of improvements.
Here is the list of the main changes:
- Correct the output of the optimal solution in problems with restrictions
- Improve save progress
- Fix test for asynchronous parallel circuit
- Implement asynchronous parallel scheme
- Add grid search
- Fix a bug in the calculator destructor
v0.3.0
We're making a new release with a number of improvements.
Here is the list of the main changes:
- Improve the description of the algorithms.
- Fix a bug with running a local method for tasks with partially discrete parameters.
- Add several problems with partially discrete parameters and constraints.
- Develop a benchmark for conducting comparative experiments on the iOpt, HyperOpt, Optuna frameworks.
v0.2.2
We're making a new minor release with a number of improvements.
Here is the list of the main changes:
- The naming of the framework elements has been brought into line with the PEP8 rules.
- Now the completion of the search process according to the specified timeout works correctly on the first iteration.
- The creation of process pool has been changed. The new method allows using function pointers in the problem statement.
- Fix the minor bug at the first iteration when solving mixed integer problems.
v0.2.1
This week we released a new major version of iOpt - 0.2.1.
Here is the list of the main changes:
- Discrete (categorical) parameters are now supported.
- Now it is possible to save the current state of the optimization process, load and continue the calculation using the saved data.
- Now it is possible to set the starting point from which the optimization process starts.
- The ability to solve problems with a criterion that is computable not in the entire search domain was added.
- Stop condition through time limit is now supported.
- Major improvements in the visualization module were implemented.