BtoF

[ziofil]

User description

BtoF - will update description

---

PR Type

enhancement, bug_fix, tests

---

Description

• Refactored the PolyExpAnsatz class to improve its functionality, including new methods for ansatz generation, decomposition, and evaluation.
• Replaced polynomial_shape with num_DV_vars across multiple files for consistency and correctness.
• Renamed reorder_abc to reorder_Abc and updated related tests and functions.
• Fixed formatting issues in configuration and training files.
• Enhanced tests to align with the updated class and function implementations.

---

Changes walkthrough 📝

	Relevant files
Configuration changes	1 files conf.py Update configuration for mathjax and theme options              doc/conf.py Updated mathjax_path formatting for consistency. Fixed string concatenation in html_theme_options. +4/-3
Bug fix	4 files dm.py Update condition check to use `num_DV_vars`                            mrmustard/lab_dev/states/dm.py Replaced polynomial_shape with num_DV_vars for condition check. +1/-1      ket.py Update condition check to use `num_DV_vars`                            mrmustard/lab_dev/states/ket.py Replaced polynomial_shape with num_DV_vars for condition check. +1/-1      representations.py Update condition checks in representations                              mrmustard/physics/representations.py Replaced polynomial_shape with num_DV_vars for condition check. +2/-2      trainer.py Correct formatting in `ray.wait` call                                        mrmustard/training/trainer.py Fixed indentation and formatting in ray.wait call. +2/-2
Enhancement	2 files polyexp_ansatz.py Refactor and enhance `PolyExpAnsatz` class functionality  mrmustard/physics/ansatz/polyexp_ansatz.py Refactored PolyExpAnsatz class with new methods and properties. Added handling for num_derived_vars and num_DV_vars. Improved ansatz decomposition and simplification logic. Enhanced evaluation and partial evaluation methods. +392/-422 gaussian_integrals.py Refactor and rename functions for consistency                        mrmustard/physics/gaussian_integrals.py Renamed reorder_abc to reorder_Abc. Updated join_Abc method to use math.concat. +14/-14
Tests	2 files test_polyexp_ansatz.py Update tests for `PolyExpAnsatz` with new properties          tests/test_physics/test_ansatz/test_polyexp_ansatz.py Updated tests to use num_DV_vars instead of polynomial_shape. +4/-6      test_gaussian_integrals.py Update tests for renamed function `reorder_Abc`                    tests/test_physics/test_gaussian_integrals.py Updated tests to use reorder_Abc instead of reorder_abc. +7/-7

---

💡 PR-Agent usage: Comment /help "your question" on any pull request to receive relevant information

[codiumai-pr-agent-pro]

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

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 4 🔵🔵🔵🔵⚪

🏅 Score: 85

🧪 PR contains tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Possible Bug The _partial_eval_single method is called with incorrect number of arguments in _partial_eval method. The indices parameter is missing when calling _partial_eval_single. Performance Issue Multiple concatenations in join_Abc could be combined into a single operation to improve performance

[codiumai-pr-agent-pro]

The _partial_eval method should pass the indices parameter when calling _partial_eval_single. Change line 531 to: Abc.append(self._partial_eval_single(Ai, bi, ci, z, indices)) [important]

[codiumai-pr-agent-pro]

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

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Score
Best practice	Replace string concatenation with a single string literal to avoid potential whitespace issues The string concatenation using "+" operator for the copyright text is error-prone and could lead to unexpected whitespace. Use a single string literal instead. doc/conf.py [144] -"TensorFlow, the TensorFlow logo, and any related marks are trademarks " "of Google Inc." +"TensorFlow, the TensorFlow logo, and any related marks are trademarks of Google Inc." Apply this suggestion Suggestion importance[1-10]: 5 Why: The suggestion correctly identifies a potential issue with string concatenation that could lead to unwanted whitespace. While valid, the impact is minor as the current code still works correctly.	5
Possible issue	Add shape validation before array concatenation Add validation to ensure input arrays have compatible shapes before concatenation. mrmustard/physics/gaussian_integrals.py [265-271] +if A1Z.shape[-1] != ZA2.shape[-1]: + raise ValueError(f"Incompatible shapes for concatenation: {A1Z.shape}, {ZA2.shape}") A = math.concat([A1Z, ZA2], axis=-2) A = math.concat( [ A[:, :n1, :], A[:, nA1 : nA1 + n2, :], ], axis=-2, ) Apply this suggestion Suggestion importance[1-10]: 4 Why: Adding shape validation before concatenation could help catch potential errors earlier with clearer error messages, though the math.concat operation would raise an error anyway if shapes are incompatible.	4
Maintainability	Place inline code comment at the correct line to properly disable the intended lint warning Move the pylint disable comment to the line where it's actually needed, as it's currently misplaced and disabling the wrong line. mrmustard/training/trainer.py [397-399] -results, running_tasks = ray.wait( +results, running_tasks = ray.wait( # pylint: disable=unused-variable promises, num_returns=len(promises) -) # pylint: disable=unused-variable +) Apply this suggestion Suggestion importance[1-10]: 4 Why: The suggestion correctly points out that the pylint disable comment should be moved to the line where the unused variable is declared. While this improves code clarity, it's a minor maintainability enhancement.	4
Performance	Cache computed values to avoid redundant calculations Cache the computed exponential part to avoid recalculating it when combining with polynomial part. mrmustard/physics/ansatz/polyexp_ansatz.py [466-472] exp_sum = self._compute_exp_part(z) # shape (batch_size, k) if self.num_derived_vars == 0: ret = math.einsum("ik,i...->k...", exp_sum, self.c) else: poly = self._compute_polynomial_part(z) # shape (batch_size, k, *derived_shape) ret = self._combine_exp_and_poly(exp_sum, poly) +self._cached_exp_sum = exp_sum # Cache for potential reuse Apply this suggestion Suggestion importance[1-10]: 3 Why: While caching could improve performance in some cases, the benefit is likely minimal as the cached value would only be useful if the same z input is used multiple times, which is not a common use case.	3

💡 Need additional feedback ? start a PR chat