# `aug_path_length` — Path Length Augmenter _Group_: **Augmentation** · _Binding_: `n4m.sklearn.PathLengthAugmenter` · _C ABI_: `n4m_aug_path_length_*` ## Description Simulate multiplicative path-length variation. ### Parameters | Name | Type | Default | |------|------|---------| | `path_length_std` | `float` | `0.05` | | `min_path_length` | `float` | `0.1` | | `rng` | `Optional[PCG64]` | `None` | | `seed` | `int` | `0` | ## Explanations ### Bibliographic source _Standard spectroscopic operator — see the nirs4all preprocessing / augmentation handbook and the cited literature within the binding docstring._ ### Mathematical principle Simulate multiplicative path-length variation. ### Implementation C ABI `n4m_aug_path_length_*` in libn4m (create / apply / destroy lifecycle), wrapped by `n4m.sklearn.PathLengthAugmenter`. The same numerical kernel backs every language binding. ### Usage ```python from n4m.sklearn import PathLengthAugmenter op = PathLengthAugmenter() X_transformed = op.fit_transform(X) ``` ### Benchmarks Adaptive wall-clock per cell measured against [`full_matrix.csv`](../benchmarks/overview.md). Only backends that implement this method are listed; libraries without the method are omitted. **Verdict** · ✓ ref / ≈ ref / ~ shape mark a reference-gate pass at strict / relaxed / qualitative tolerance · ✓ bind = pls4all binding agrees with the C++ baseline · ✗ divergent · ⚠ error · — not run. The fastest backend per column is marked 🏆. **Reference gate**: strict — numeric equivalence (`rmse_rel_tol ≤ 1e-12`). ::::{tab-set} :class: parity-tabs :::{tab-item} 1 thread :sync: threads-1
| Backend | Parity | 50×250 (ms) | 250×50 (ms) |
|---|---|---|---|
| C++ native · libn4m | |||
pls4all.cpp.blas+omp | ✓ ref | — | — |
| Python · external | |||
nirs4all | source | — | — |