Example programs¶

The repository contains executable Python and C++ examples. The Python scripts were checked as programs, not only imported as modules, so their __main__ paths and plotting setup are exercised.

Run Python examples¶

From a development checkout:

uv sync
uv run python examples/cubic_spline_ex.py

Plots open interactively under a normal Matplotlib backend. For a noninteractive smoke run:

MPLBACKEND=Agg uv run python examples/double_s_ex.py

Scalar splines¶

Script	Demonstrates
`cubic_spline_ex.py`	Cubic interpolation and endpoint velocities
`c_s_smoothing_ex.py`	Several smoothing values and infinite endpoint weights
`c_s_smoot_search_ex.py`	Binary search for a prescribed maximum error
`c_s_with_acc1_ex.py`	Virtual-waypoint acceleration constraints and applications
`c_s_with_acc2_ex.py`	`SplineParameters` and quintic boundary segments

B-splines¶

Script	Demonstrates
`b_spline_ex.py`	Basis values, derivatives, knots, and 2D plotting
`b_spline_cubic_ex.py`	Cubic 3D interpolation with chord-length parameters
`b_spline_interpolate_ex.py`	Degree ¾/5 interpolation and derivative constraints
`b_spline_approx_ex.py`	Approximation with fewer control points and refinement
`b_spline_smooth_ex.py`	3D curve smoothing and error measures

Motion profiles¶

Script	Demonstrates
`double_s_ex.py`	Jerk-limited motion, reverse moves, phase timing, and factory use
`trapezoidal_ex.py`	Velocity- or duration-constrained segments and via points
`polynomials_ex.py`	Cubic, quintic, seventh-order, and multipoint polynomials
`lin_poly_parabolic_ex.py`	Linear paths joined by parabolic blends
`linear_ex.py`	Scalar and vector constant-velocity interpolation

Quaternion interpolation¶

Script	Demonstrates
`quat_visualization_ex.py`	SLERP/SQUAD comparison and quaternion plots
`squad_c2_ex.py`	SLERP, SQUAD, and `SquadC2` comparison
`log_quat_new_ex.py`	LQI and mLQI interpolation and diagnostics
`lqi_spiral_frenet_ex.py`	LQI fitted to Frenet orientations on a helix
`mlqi_spiral_frenet_ex.py`	mLQI fitted to Frenet orientations on a helix

The logarithmic examples intentionally use some implementation-level state for diagnostic plots. Run them with the Python backend when inspecting those details:

INTERPOLATEPY_NO_CPP=1 uv run python examples/log_quat_new_ex.py

Paths and generic interfaces¶

Script	Demonstrates
`simple_paths_ex.py`	Linear/circular geometry combined with polynomial time laws
`frenet_frame_ex.py`	Frenet and tool frames for circles and helices
`protocols_ex.py`	Generic sampling through runtime-checkable protocols
`main.py`	Minimal installed-version smoke check

Run every Python example headlessly¶

Use the repository checker to run every script in its own process:

uv run python -m scripts.check_python_examples --backend python

After building and copying the native extension into interpolatepy/, replace python with native to verify the adapter-backed API. The checker fails if native mode is requested but HAS_CPP is false.

This can take longer than the unit suite because several scripts construct many figures. Matplotlib can warn that many figures are open during a headless run; that warning is not a numerical failure.

C++ examples¶

Enable the CMake example target, then build it:

cmake -S cpp -B build/cpp-examples \
  -DINTERPOLATECPP_BUILD_EXAMPLES=ON \
  -DINTERPOLATECPP_BUILD_TESTS=OFF
cmake --build build/cpp-examples --parallel

The generated executables correspond to the files under cpp/examples/:

cubic, smoothing, and acceleration-constrained splines;
base, interpolation, approximation, and smoothing B-splines;
trapezoidal, polynomial, Double-S, and parabolic-blend motion;
quaternion interpolation;
paths and C++20 concepts.

Their exact output directory depends on the CMake generator and build configuration. For example, a single-configuration build commonly produces build/cpp-examples/examples/cubic_spline_example.

Writing a new example¶

make the file directly executable with an if __name__ == "__main__" block;
import supported algorithms from interpolatepy unless the example is explicitly about implementation internals;
keep data deterministic or seed random generators;
separate numerical construction from visualization when practical;
ensure the program completes with MPLBACKEND=Agg;
add it to this page and to the relevant tutorial.