How to train new Rl Algorithm with Parallel Computing? I got the error:Parallel training is not supported for this agent

I successfully run the Algorithm called PIBB, buet I got the error when i want to run it parallel

错误使用 rl.agent.CustomAgent/checkAgentTrainOptionsCompatible_
Parallel training is not supported for this agent. Set 'trainingOptions.Parallelization' to 'none'.

出错 rl.agent.AbstractAgent/checkAgentTrainOptionsCompatible (第 128 行)
            trainingOptions = checkAgentTrainOptionsCompatible_(this, trainingOptions);

出错 rl.agent.AbstractAgent/train (第 75 行)
    trainingOptions = checkAgentTrainOptionsCompatible(this, trainingOptions);

出错 MyTrain (第 33 行)
trainResults = train(agent,env,opt);

Answer 1

[Sanjana]

Hi,

I understand that you're encountering difficulties while using parallel processing to train a custom defined RL agent in MATLAB.

To resolve the issue, I suggest that you follow these steps:

• Verify that you have installed the Parallel Computing Toolbox on your system. You can check this by typing "ver" in the Command Window, which will display all the installed products and their versions. Look for the Parallel Computing Toolbox in the list
• Ensure that the "trainingOptions" to train the agent are correctly defined, specifically, the "useParallel" option must be set to true. This will configure the training process to use parallel processing. Additionally, you can specify the options for parallel training using "ParallelizationOptions"
• For further assistance, I recommend referring to the following links:

https://www.mathworks.com/help/reinforcement-learning/ref/rl.option.rltrainingoptions.html#mw_3ee5a5bb-69a6-46bb-97a0-7428ccfdc5d6

https://www.mathworks.com/help/reinforcement-learning/ug/train-agents-using-parallel-computing-and-gpu.html

Hope this helps!