mescla.emma.projection.subspace_distance¶
- mescla.emma.projection.subspace_distance(points, standardizer, eigenvectors, k)[source]¶
Euclidean distance of each water from the mixing subspace, in standardised units.
For a candidate end-member this measures how well it belongs to the same mixing system as the samples: a large distance means the candidate carries chemistry the mixture does not contain. Tubau et al. (2014) used exactly this to decide how tightly to constrain each end-member (their W1 sat close to the subspace and was given small variances; D2 sat far off and was given large ones).
- Parameters:
points (Any)
standardizer (Standardizer)
eigenvectors (ndarray)
k (int)
- Return type: