Kobi Felton

https://github.com/sustainable-processes/summit/blob/a1cecdd41df8119005173b46ac45fb22472628d6/summit/strategies/sobo.py#L297

BaumgartnerEmualtor is not loading the correct domain (eg., uses yield instead of yld) and also not allowing multiobjective cost.

Right now the code for transforms is complicated and also uses deprecated features in pandas like setting multi-index code in place. The code should be refactored to be simpler and...

- [ ] Check if constraints are violated by strategies - [ ] Switch to using runners to reduce complexity of the code - [ ] Make sure all functions...

The multitask strategy currently only works with num_experiments=1 because it uses ExpectedImprovement, which doesn't support batch optimization in BoTorch. The [qEI](https://botorch.org/api/acquisition.html#botorch.acquisition.monte_carlo.qExpectedImprovement) acquisition function should work better, but will require some...

This is a draft PR to remove the dependency on GPy as it is no longer maintained. Before we do that, we'll need to replace the current SOBO strategy with...

Sweep: Emulator should not be retrained on initialization

1

ExperimentalEmulator should not be retrained on reinitialization (see `summit/benchmarks/experimental_emulator.py`). See: https://github.com/sustainable-processes/summit/blob/8b987dee789eeb6e7e87184e935425330e2501f5/summit/benchmarks/experimental_emulator.py#L308-L317

Right now we don't really test mixed domain and categorical only domains.

Pypi

1

Can we put gryffin on pypi?