Update changelog for the N-dimensional generalization

Author: braniii

CHANGELOG.md

@@ -17,12 +17,22 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Added Features and Improvements 🙌:
 - Added submodule to work with MD data `fpsl.utils.md` together new with tutorial 🎉
 - Added submodule with periodic fourier expansion `fpsl.utils.fourier`
+- Generalized the Fokker-Planck energy regularization to arbitrary dimensions via a
+  stochastic finite-difference estimator, enabled through the new `v_sigma` parameter
+  of `FPSL` (falls back to the previous 1D exact-gradient estimator when `v_sigma=0`)
+- Added `fpsl.utils.fourier.get_fourier_func2d`, a 2D analog of `get_fourier_func`
+- Generalized the end-of-training Weights & Biases free-energy-landscape logging to
+  1D and 2D data (previously hardcoded to 1D)
+- Added a tutorial section on enforcing custom symmetries in the score network
 
 ### Other changes:
 - Use `cosine_decay_schedule` instead of `warmup_cosine_decay_schedule` if user selects `warmup_steps=0`
 
 ### Bugfix 🐛:
 - Exclude docs directory from build
+- Fixed `fpsl.utils.md.load_trajs` discarding the loaded pull-force trajectory
+- Fixed `ScorePeriodicMLP` silently dropping its sine Fourier features (falling back
+  to the same cosine-only behavior as `ScoreSymmetricPeriodicMLP`)
 
 
 ## [0.2.0] - 2025-08-06