BUG: fix NaN in ND linear interpolation outside convex hull

[YuriCastroDev]

Pull request type

• Code changes (bugfix, features)
• Code maintenance (refactoring, formatting, tests)

Checklist

• Tests for the changes have been added (if needed)
• Lint (black rocketpy/ tests/) has passed locally
• All tests (pytest tests -m slow --runslow) have passed locally
• CHANGELOG.md has been updated (if relevant)

Current behavior

Closes #926.

When using Function(..., interpolation="linear") with N-D data,
querying a point inside the axis-aligned bounding box but outside
the convex hull silently returns nan with no warning.

data = [
    [0.0, 0.0, 0.000],
    [0.0, 0.1, 0.100],
    [0.0, 0.4, 0.400],
    [0.3, 0.2, 0.150],
]
f = Function(data, interpolation="linear")
print(f(0.3, 0.3))  # nan

New behavior

f(0.3, 0.3) no longer returns nan. Points outside the convex
hull are correctly identified and routed to the extrapolation method.

Breaking change

• Yes
• No

Additional information

__get_value_opt_nd used per-axis min/max (bounding box) to decide
between interpolation and extrapolation. LinearNDInterpolator
operates on the Delaunay triangulation (convex hull), not the
bounding box — points in the gap between the two were silently
sent to the interpolator and came back as nan.

Fix: compute Delaunay(self._domain) once at setup, pass it to
LinearNDInterpolator (reuses the triangulation at no extra cost),
and store it as self._nd_triangulation. In __get_value_opt_nd,
replace the bounding box check with find_simplex(args) < 0.
Shepard, RBF and other ND interpolators are unaffected.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.12%. Comparing base (9cf3dd4) to head (f13daad).
⚠️ Report is 74 commits behind head on develop.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop     #969      +/-   ##
===========================================
+ Coverage    80.27%   81.12%   +0.85%     
===========================================
  Files          104      113       +9     
  Lines        12769    14560    +1791     
===========================================
+ Hits         10250    11812    +1562     
- Misses        2519     2748     +229     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Gui-FernandesBR]

LGTM

[Copilot]

Pull request overview

This PR fixes a bug in rocketpy.mathutils.Function where N-D interpolation="linear" could return NaN for query points that are inside the axis-aligned bounding box but outside the Delaunay convex hull, by using the triangulation to correctly route those points to extrapolation.

Changes:

• Build and cache a scipy.spatial.Delaunay triangulation for N-D linear interpolation and reuse it for convex-hull membership checks.
• Update N-D fast evaluation to detect “outside hull” points via find_simplex(...) < 0 instead of bounding-box checks.
• Add a regression unit test for the NaN case and document the fix in the changelog.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
rocketpy/mathutils/function.py	Cache Delaunay triangulation for N-D linear interpolation and use it to decide interpolation vs extrapolation.
tests/unit/mathutils/test_function.py	Adds regression test ensuring no NaN is returned for bounding-box-but-outside-hull queries.
CHANGELOG.md	Notes the bug fix under Unreleased/Fixed.

[phmbressan]

Good implementation. Much nicer than my initial idea of computing the interpolation and, with a mask, forwarding any existing NaNs to extrapolate. This is more mathematically sound and likely faster.

About the linter issue, it seems a false positive so I would simply disable that line:

from scipy.spatial import Delaunay  # pylint: disable=no-name-in-module

[YuriCastroDev]

I'm available to keep contributing, if you have any open issues or tasks that need help, I'd appreciate any guidance on where to focus next!