🎨 Improve RL Action Handling (#401)

<!--- This file has been generated from an external template. Please do
not modify it directly. -->
<!--- Changes should be contributed to
https://github.com/munich-quantum-toolkit/templates. -->

## Description

This PR introduces a class structure to the RL compilation actions and
resolves #144.


## Checklist:

<!---
This checklist serves as a reminder of a couple of things that ensure
your pull request will be merged swiftly.
-->

- [x] The pull request only contains commits that are focused and
relevant to this change.
- [x] I have added appropriate tests that cover the new/changed
functionality.
- [x] I have added entries to the changelog for any noteworthy
additions, changes, fixes, or removals.
- [x] The changes follow the project's style guidelines and introduce no
new warnings.
- [x] The changes are fully tested and pass the CI checks.
- [x] I have reviewed my own code changes.

---------

Signed-off-by: Nils Quetschlich <nils.quetschlich@tum.de>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

Author: 4 people

CHANGELOG.md

@@ -9,6 +9,10 @@ This project adheres to [Semantic Versioning], with the exception that minor rel
 
 ## [Unreleased]
 
+### Changed
+
+- ✨ Improve RL action handling by using dataclasses ([#401]) ([**@nquetschlich**])
+
 ## [2.2.0] - 2025-02-02
 
 _📚 Refer to the [GitHub Release Notes](https://github.com/munich-quantum-toolkit/predictor/releases) for previous changelogs._
@@ -22,6 +26,9 @@ _📚 Refer to the [GitHub Release Notes](https://github.com/munich-quantum-tool
 
 <!-- Contributor -->
 
+[**@burgholzer**]: https://github.com/burgholzer
+[**@nquetschlich**]: https://github.com/nquetschlich
+
 <!-- General links -->
 
 [Keep a Changelog]: https://keepachangelog.com/en/1.1.0/

docs/Usage.rst

@@ -43,15 +43,10 @@ MQT Predictor Framework Setup
 To run ``qcompile``, the MQT Predictor framework must be set up. How this is properly done is described next.
 
 First, the to-be-considered quantum devices must be included in the framework.
-Currently, all devices supported by `MQT Bench <https://github.com/cda-tum/mqt-bench>`_ are natively supported.
-In case another device shall be considered, it can be added by using a similar format as in MQT Bench but it is not
-necessary to add it in the repository since it can be directly added to the MQT Predictor framework as follows:
+All devices supported by `MQT Bench <https://github.com/cda-tum/mqt-bench>`_ are natively supported.
+Furthermore, a custom device can be added to the framework as long as it is provided as a Qiskit Target object.
 
-- Modify in `mqt/predictor/rl/predictorenv.py <https://github.com/munich-quantum-toolkit/predictor/tree/main/src/mqt/predictor/rl/predictorenv.py>`_. the line where ``mqt.bench.devices.get_device_by_name`` is used.
-- Modify in `mqt/predictor/ml/predictor.py <https://github.com/munich-quantum-toolkit/predictor/tree/main/src/mqt/predictor/ml/predictor.py>`_. the lines where ``mqt.bench.devices.*`` are used.
-- Follow the same data format as defined in `mqt.bench.devices.device.py <https://github.com/cda-tum/mqt-bench/tree/main/src/mqt/bench/devices/device.py>`_
-
-Second, for each supported device, a respective reinforcement learning model must be trained. This is done by running
+Second, for each device, a respective reinforcement learning model must be trained. This is done by running
 the following command based on the training data in the form of quantum circuits provided as qasm files in
 `mqt/predictor/rl/training_data/training_circuits <https://github.com/munich-quantum-toolkit/predictor/tree/main/src/mqt/predictor/rl/training_data/training_circuits>`_:
 
@@ -72,6 +67,29 @@ the following command based on the training data in the form of quantum circuits
 This will train a reinforcement learning model for the ``ibm_falcon_27`` device with the expected fidelity as figure of merit.
 Additionally to the expected fidelity, also critical depth is provided as another figure of merit.
 Further figures of merit can be added in `mqt.predictor.reward.py <https://github.com/munich-quantum-toolkit/predictor/tree/main/src/mqt/predictor/reward.py>`_.
+Please note that there is a pre-configured set of available compilation passes that are supported.
+This is defined in `mqt.predictor.rl.actions <https://github.com/munich-quantum-toolkit/predictor/tree/main/src/mqt/predictor/rl/actions.py>`_ and can be easily extended.
+If another compilation pass from Qiskit, TKET, or BQSKit shall be added, this can be done using:
+
+.. code-block:: python
+
+    from mqt.predictor.rl.actions import (
+        CompilationOrigin,
+        DeviceIndependentAction,
+        PassType,
+        register_action,
+    )
+
+    my_custom_pass = ...  # Define your custom pass here, e.g., a Qiskit pass
+    action = DeviceIndependentAction(
+        name="test_action",
+        pass_type=PassType.OPT,
+        transpile_pass=[my_custom_pass],
+        origin=CompilationOrigin.QISKIT,
+    )
+    register_action(action)
+
+For other sources, defining a new ``CompilationOrigin`` is necessary as well as providing parsing methods to and from Qiskit's QuantumCircuit, since this is used as our internal representation of quantum circuits.
 
 Third, after the reinforcement learning models that are used for the respective compilations are trained, the
 supervised machine learning model to predict the device selection must be trained.