propulate
propulate copied to clipboard
Helmholtz-AI-Energy
Propulate is an asynchronous evolutionary optimization algorithm and software package for global optimization and hyperparameter search on high-performance computers.
Currently, the Gaussian processes used in `Propulate`'s surrogate models rely on the `GPy` package which repeatedly caused problems and errors, in particular in the context with `matplotlib`. According to SheffieldML/GPy#1034,...
This PR adds parameter embedding to the Individual class so distance based propagators can work with categorical parameters. It also adds a Nelder-Mead simplex optimisation inspired propagator. Since propulate requires...
A migration probability per island is difficult, since each worker has to decide whether to migrate independently of the other workers on the island to avoid synchronisation. Easiest fix is...
Title more flexibility and easier filtering
During testing output from the logger is printed multiple times. The number of prints depends on the number of tests already run. This is fixed in the sequential tests in...
Enhance the current test functionality to - Test CMA-ES with migration - Enable unit tests of all GA propagators In addition, we should try to remove all warnings generated during...
The fixture-decorated function `function_parameters` is used in multiple tests for requesting different combinations of benchmark function name and the corresponding global optimum: ``` @pytest.fixture( params=[ ("rosenbrock", 0.0), ("step", -25.0), ("quartic",...
Metadata
Owner
Metadata
Propulate is an asynchronous evolutionary optimization algorithm and software package for global optimization and hyperparameter search on high-performance computers.