DDPG Not getting any reward? Using the example from MathWorks

I am using the water pump example from Mathworks for the DDPG Reinforcement example, but I changed the pump model slightly and I am not getting any reward when I run the program, it is always zero. I changed the reward to always be 10 just to see if maybe it just wasnt learning, but it still only shows a reward of 0. Does anyone have any ideas? This is my first post so sorry if I am not submitting it correctly. Thanks!

[Walter Roberson]

It would help if you could attach your files so people could replicate the problem

[Jun]

Sorry, here you go. I am trying to have it change a resistor for a voltage divider network as a start for a more advanced circuit. But I am not an expert at reinforced learning so I have been adjusting the example provided by Matlab.

Answer 1

[Emmanouil Tzorakoleftherakis]

Hi Jun,

The IsDone input to the agent block is always true, so all episodes end prematurely. This is why you don't see any change in the reward. Change the conditions that set the IsDone flag to be true or set it to be false and the training will resume.