Many use cases require a continuous action space. For example, if decision point is a knob that can turn 360 degrees, a continuous action space is more natural than a discrete action space.

Step 1 - In the "Actions" field within the Pathmind Helper properties, define your action space as continuous by using the @Continuous annotation.

@Continuous defines your action space as continuous.

low and high defines the range of values. In the screenshot above, it's been set to low = 0 and high = 360 which correspond with a knob being able to turn 360 degrees.

shape indicates the total number of decision points in your model. For example, if your simulation contains three knobs, then you would set shape = 3.

Step 2 - Update your doAction function to accept continuous actions. Please revisit this article for instructions on how to do that.

Did this answer your question?