can't convert complex to float

[chauvinSimon]

Thank you for this great code!

Just one question regarding the line 16 of prioritized_memory.py: Should not the absolute value of the TD-error be taken in return (error + self.e) ** self.a ?

This is what is done in the original paper. In addition, if not using abs(), I get the error can't convert complex to float due to (error + self.e) being negative.

[being-aerys]

You are correct in saying that the priority value = absolute td error. However, I think the passed error in return (error + self.e) ** self.a ? is already an absolute value. Or at least that is what u are supposed to do if not already.