Merge pull request #400 from smash-transport/krupczak/PCA

correct a mistake

Author: Hendrik1704

CHANGELOG.md

@@ -33,6 +33,7 @@ Date: 2026-XX-XX
 * Particle: Refactor class to use more efficient internal data structures to save memory.
 
 ### Fixed
+* PCAFlow: Fixed eigenvector selection from the sorted eigenvector array.
 * Failing import of `JetAnalysis` if `fastjet` is not installed or not compatible with the current numpy/python version. Now if the import fails, the user can still use all other features of SPARKX, but the `JetAnalysis` class will not be available.
 
 [Link to diff from previous version](https://github.com/smash-transport/sparkx/compare/v2.2.0...v2.3.0)

src/sparkx/flow/PCAFlow.py

@@ -441,7 +441,7 @@ def __compute_flow_PCA(self, bins: Union[List[float], np.ndarray]) -> None:
             for bin in range(number_bins):
                 for alpha in range(self.alpha_):
                     eval = sorted_eigenvalues[alpha]
-                    evec = sorted_eigenvectors[alpha]
+                    evec = sorted_eigenvectors[:, alpha]
                     if (eval >= 0.0) and (
                         self.sigma_multiplicity_total_[sub][bin] > 0.0
                     ):
@@ -472,7 +472,7 @@ def __compute_flow_PCA(self, bins: Union[List[float], np.ndarray]) -> None:
         for bin in range(number_bins):
             for alpha in range(self.alpha_):
                 eval = sorted_eigenvalues[alpha]
-                evec = sorted_eigenvectors[alpha]
+                evec = sorted_eigenvectors[:, alpha]
                 if (eval >= 0.0) and (self.bin_multiplicity_total_[bin] > 0.0):
                     self.Flow_[bin][alpha] = (
                         np.sqrt(eval)