2.0.0

Gym-ANM v2.0 is here 🎉

What's Changed

From gym to gymnasium

gym-anm now fully runs on top of gymnasium, instead of the older gym standard. This means there have been a few API changes to match the latest gymnasium framework:

Change 1: reset()

Seeding the environment now happens within reset().

## gym-anm v1.x
def seed(self, seed):  # used to seed the env
    ...

def reset():  # used to reset the env
    ...

## gym-anm v2.0
def reset(self, *, seed, options = None):  # used to both seed and reset the env
    ...

Change 2: step()

The step() function returns a new extra parameter truncated: bool (which is always False in gym-anm), and done has been renamed terminated to match the gymnasium terminology.

## gym-anm v1.x
def step(...):
    ...
    return obs, r, done, info

## gym-anm v2.0
def step(...):
    ...
    return obs, r, self.terminated, truncated, info

Dependency updates

A few dependencies were bumped up and updated:

• numpy: 1.x -> >= 2.1
• pandas: 1.x -> >= 2.0

---

If you have any feedback, or are having trouble installing and running v2.0, definitely reach out! 🙂