allExperiences

Return all experiences in replay memory buffer

Since R2022b

collapse all in page

    Syntax

    experiences = allExperiences(buffer)
    experiences = allExperiences(buffer,Name=Value)

    Description

    experiences = allExperiences(buffer) returns all experiences stored in experience buffer buffer as individual experiences, each with a batch size of 1 and a sequence length of 1.

    example

    experiences = allExperiences(buffer,Name=Value) specifies the type and concatenation of the fields in experience using one or more name-value pair arguments. You can specify whether to return the experiences as dlarray objects or whether to store them in the GPU. You can also return experiences concatenated along the batch dimension or the sequence dimension.

    Examples

    collapse all

    Open Live Script

    Define observation specifications for the environment. For this example, assume that the environment has two observation channels: one channel with two continuous observations and one channel with a three-valued discrete observation.

    obsContinuous = rlNumericSpec([2 1],...
    LowerLimit=0,...
    UpperLimit=[1;5]);
obsDiscrete = rlFiniteSetSpec([1 2 3]);
obsInfo = [obsContinuous obsDiscrete];

    Define action specifications for the environment. For this example, assume that the environment has a single action channel with one continuous action in a specified range.

    actInfo = rlNumericSpec([2 1],...
    LowerLimit=0,...
    UpperLimit=[5;10]);

    Create an experience buffer with a maximum length of 5000.

    buffer = rlReplayMemory(obsInfo,actInfo,5000);

    Append a sequence of 10 random experiences to the buffer.

    for i = 1:10
    experience(i).Observation = ...
        {obsInfo(1).UpperLimit.*rand(2,1) randi(3)};
    experience(i).Action = {actInfo.UpperLimit.*rand(2,1)};
    experience(i).NextObservation = ...
        {obsInfo(1).UpperLimit.*rand(2,1) randi(3)};
    experience(i).Reward = 10*rand(1);
    experience(i).IsDone = 0;
end

append(buffer,experience);

    After appending experiences to the buffer, you extract all of the experiences from the buffer. Extract all of the experiences as individual experiences, each with a batch size of 1 and sequence size of 1.

    experience = allExperiences(buffer)
    experience=10×1 struct array with fields:
    Observation
    Action
    NextObservation
    Reward
    IsDone

    Alternatively, you can extract all of the experiences as a single experience batch.

    expBatch = allExperiences(buffer,ConcatenateMode="batch")
    expBatch = struct with fields:
        Observation: {[2×1×10 double]  [1×1×10 double]}
             Action: {[2×1×10 double]}
             Reward: [9.5751 9.1574 7.4313 8.2346 1.8687 1.6261 5.0596 2.5428 3.5166 5.6782]
    NextObservation: {[2×1×10 double]  [1×1×10 double]}
             IsDone: [0 0 0 0 0 0 0 0 0 0]

    Input Arguments

    collapse all

    Experience buffer, specified as one of the following replay memory objects.

    Name-Value Arguments

    collapse all

    Specify optional pairs of arguments as Name1=Value1,...,NameN=ValueN, where Name is the argument name and Value is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.

    Example: mode="batch"

    Concatenation mode specified as a one of the following values.

    • "none" — Return experience as N individual experiences, each with a batch size of 1 and a sequence length of 1.

    • "batch" — Return experience as a single batch with a sequence length of 1.

    • "sequence" — Return experience as a single sequence with a batch size of 1.

    Option to return output as deep learning array, specified as a logical value. When you specify ReturnDlarray as true the fields of experience are dlarray objects.

    Example: ReturnDlarray=true

    Option to return output as GPU array, specified as a logical value. When you specify ReturnGPUarray as true the fields of experience are stored in the GPU.

    Setting this option to true requires both Parallel Computing Toolbox™ software and a CUDA® enabled NVIDIA® GPU. For more information on supported GPUs see GPU Computing Requirements (Parallel Computing Toolbox).

    You can use gpuDevice (Parallel Computing Toolbox) to query or select a local GPU device to be used with MATLAB®.

    Example: ReturnGpuArray=true

    Output Arguments

    collapse all

    All N buffered experiences, returned as a structure array or structure. When mode is:

    • "none", experience is returned as a structure array of length N, where each element contains one buffered experience (batchSize = 1 and SequenceLength = 1).

    • "batch", experience is returned as a structure. Each field of experience contains all buffered experiences concatenated along the batch dimension (batchSize = N and SequenceLength = 1).

    • "sequence", experience is returned as a structure. Each field of experience contains all buffered experiences concatenated along the batch dimension (batchSize = 1 and SequenceLength = N).

    experience contains the following fields.

    Observation, returned as a cell array with length equal to the number of observation specifications specified when creating the buffer. Each element of Observation contains a DO-by-batchSize-by-SequenceLength array, where DO is the dimension of the corresponding observation specification.

    Agent action, returned as a cell array with length equal to the number of action specifications specified when creating the buffer. Each element of Action contains a DA-by-batchSize-by-SequenceLength array, where DA is the dimension of the corresponding action specification.

    Reward value obtained by taking the specified action from the observation, returned as a 1-by-1-by-SequenceLength array.

    Next observation reached by taking the specified action from the observation, returned as a cell array with the same format as Observation.

    Termination signal, returned as a 1-by-1-by-SequenceLength array of integers. Each element of IsDone has one of the following values.

    • 0 — This experience is not the end of an episode.

    • 1 — The episode terminated because the environment generated a termination signal.

    • 2 — The episode terminated by reaching the maximum episode length.

    Version History

    Introduced in R2022b

    See Also

    Functions

    Objects