pp_first_derivative — First derivative

Group: Preprocessing · Binding: n4m.sklearn.FirstDerivative · C ABI: n4m_pp_first_derivative_*

Description

np.gradient(X, delta, axis=1, edge_order=...) (shape-preserving).

Parameters

Name	Type	Default
delta	float	1.0
edge_order	int	2

Explanations

Bibliographic source

Standard finite-difference / gap derivative; see Savitzky & Golay (1964) and Norris & Williams (1984).

Mathematical principle

Approximates \(\mathrm{d}\mathbf{x}/\mathrm{d}\lambda\) by finite differences. The first derivative removes constant baseline offsets (additive scatter) and accentuates inflection points of overlapping bands.

Implementation

C ABI n4m_pp_first_derivative_* in libn4m (create / apply / destroy lifecycle), wrapped by n4m.sklearn.FirstDerivative. The same numerical kernel backs every language binding.

Usage

from n4m.sklearn import FirstDerivative
op = FirstDerivative()
X_transformed = op.fit_transform(X)

Benchmarks

Adaptive wall-clock per cell measured against full_matrix.csv. 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  ·  ⇄ cross-check = documented by-design selector/RNG/model, noncanonical API/facade convention, or secondary oracle  ·  ✗ divergent  ·  ⚠ error  ·  — not run. The fastest backend per column is marked 🏆.

Reference gate: strict — numeric equivalence (rmse_rel_tol ≤ 1e-12).

1 thread

Backend	Parity	50×250 (ms)	250×50 (ms)
C++ native · libn4m
pls4all.cpp.blas	✓ ref	0.01 ms	0.01 ms🏆
pls4all.cpp.blas+omp	✓ ref	0.01 ms	0.03 ms
pls4all.cpp.omp	✓ ref	0.03 ms	0.03 ms
pls4all.cpp.ref	✓ ref	0.01 ms🏆	0.01 ms
Python · pls4all
pls4all.python	✓ bind	0.01 ms	0.01 ms
pls4all.sklearn	✓ bind	0.01 ms	0.01 ms
Python · external
nirs4all	source	0.04 ms	0.05 ms
ref.python_numpy	source	—	—

---

See also: methods index · interactive dashboard