Fix TensorFlow v2.17 compatibility by removing optimizer caching

[zeyueN]

User description

Context:

Description of the Change:

Benefits:

Possible Drawbacks:

Related GitHub Issues:

---

PR Type

bug_fix

---

Description

• Removed the caching mechanism for the Euclidean optimizer in backend_manager.py to ensure compatibility with TensorFlow v2.17.
• The Euclidean optimizer is now always instantiated when accessed, preventing potential issues with stale or incompatible cached objects.

---

Changes walkthrough 📝

	Relevant files
Bug fix	backend_manager.py Remove caching for Euclidean optimizer to fix compatibility mrmustard/math/backend_manager.py Removed caching of the Euclidean optimizer object. Always create a new Euclidean optimizer instance. +1/-2

---

💡 PR-Agent usage: Comment /help on the PR to get a list of all available PR-Agent tools and their descriptions

[codiumai-pr-agent-pro[bot]]

PR-Agent was enabled for this repository. To continue using it, please link your git user with your CodiumAI identity here.

PR Reviewer Guide 🔍

⏱️ Estimated effort to review: 2 🔵🔵⚪⚪⚪

🏅 Score: 95

🧪 No relevant tests

🔒 No security concerns identified

⚡ Key issues to review Performance Concern Creating a new optimizer instance on every property access might impact performance if called frequently

[codiumai-pr-agent-pro[bot]]

PR-Agent was enabled for this repository. To continue using it, please link your git user with your CodiumAI identity here.

PR Code Suggestions ✨

Category	Suggestion	Score
Performance	Initialize the optimizer in the constructor instead of the property getter ___ **Consider initializing self._euclidean_opt in the class constructor instead of in the property getter. This would ensure that the optimizer is created only once when the object is instantiated, rather than every time the property is accessed.** [mrmustard/math/backend_manager.py [1316-1320]](https://github.com/XanaduAI/MrMustard/pull/489/files#diff-272a316cdf15707ddb7384f13258135f593fc560e21dc19cf17c16cfb5d0aa1cR1316-R1320) ```diff +def __init__(self, *args, **kwargs): + super().__init__(*args, **kwargs) + self._euclidean_opt = self.DefaultEuclideanOptimizer() + @property def euclidean_opt(self): r"""The configured Euclidean optimizer.""" - self._euclidean_opt = self.DefaultEuclideanOptimizer() return self._euclidean_opt ``` - [ ] **Apply this suggestion** Suggestion importance[1-10]: 8 Why: This suggestion improves performance by ensuring the optimizer is created only once during object instantiation, rather than every time the property is accessed. This change aligns with typical object-oriented design practices.	8
Best practice	✅ Rename property to method if new instance creation is intended ___ **If creating a new optimizer instance on every access is intentional, consider renaming the property to a method (e.g., get_euclidean_opt()) to better reflect its behavior of always returning a new instance.** [mrmustard/math/backend_manager.py [1316-1320]](https://github.com/XanaduAI/MrMustard/pull/489/files#diff-272a316cdf15707ddb7384f13258135f593fc560e21dc19cf17c16cfb5d0aa1cR1316-R1320) ```diff -@property -def euclidean_opt(self): - r"""The configured Euclidean optimizer.""" - self._euclidean_opt = self.DefaultEuclideanOptimizer() - return self._euclidean_opt +def get_euclidean_opt(self): + r"""Get a new instance of the configured Euclidean optimizer.""" + return self.DefaultEuclideanOptimizer() ``` `[Suggestion has been applied]` Suggestion importance[1-10]: 7 Why: Renaming the property to a method clarifies the behavior of returning a new instance each time, which improves code readability and aligns with the intended functionality. However, it may not be necessary if the behavior is not intentional.	7