Question regarding PETS & Discrete action spaces

[aralcimcim]

Question

Using custom Gymnasium environments with the PETS algorithm

[This is not a feature request but rather a question]
I've been trying to adapt a custom Gymnasium environment to work with the PETS algorithm, the only difference is that my custom environment has a discrete action space which does not work with the CEM optimizer since it is built on upper and lower bounds. I've looked into the documentation but have not found anything related to discrete action spaces. Would any of the provided algorithms in the library work out-of-the-box for custom Gymnasium environments with discrete action spaces?