ou_strategy noise missing dt?

[zacwellmer]

I could be wrong but it seems like we need to add a dt parameter to ou_strategy.py's evolve_state function?

as per the openai baseline noise

def evolve_state(self):
    x = self.state
    dx = self.theta * (self.mu - x) * self.dt + self.sigma * np.sqrt(self.dt) * np.random.normal(size=self.action_space)
    self.state = x + dx
    return self.state

[dementrock]

Make sense. Would you like to submit a pull request?

[zacwellmer]

#197 submitted

[schneimo]

Another question regarding to the OU noise process here:
In baselines the previous state is set to a zero-array, wherefore here in rllab the state is set to an array filled with ones!?
I think that the baselines noise is correct, because so the noise is centered around 0, or not?

Baselines:

def reset(self):
        self.x_prev = self.x0 if self.x0 is not None else np.zeros_like(self.mu)

rllab:

def reset(self):
        self.state = np.ones(self.action_space.flat_dim) * self.mu