Getting Started

Use this page for the shortest path from installation to a first working spline. If you already know the workflow you want, jump to Common Tasks. If you want a guided walkthrough, go to Tutorials.

1. Install the package

Spline Core requires MATLAB R2024b or newer. Fitting workflows also depend on the Distributions package.

mpminstall("local/path/to/spline-core");

If you manage dependencies yourself, install without them explicitly:

mpminstall("local/path/to/spline-core", InstallDependencies=false);

For full install details, see Installation.

2. Choose the right starting point

If your data are… Start with Next page
exact and trusted InterpolatingSpline Spline Interpolation
noisy or constrained ConstrainedSpline Fitting Noisy Data
planar x(t), y(t) samples TrajectorySpline TrajectorySpline reference

If you only remember one distinction, use InterpolatingSpline for exact interpolation, ConstrainedSpline for fitting, and TrajectorySpline when you want a shared-parameter planar trajectory model.

3. Exact interpolation quickstart 

t = linspace(0, 1, 12)';
x = sin(2*pi*t);
f = InterpolatingSpline.fromGriddedValues(t, x, S=3);

tq = linspace(t(1), t(end), 200)';
xq = f(tq);
dxq = f.valueAtPoints(tq, D=1);

This gives you a reusable spline object for values and derivatives. For rectilinear grids, pass one grid vector per dimension:

F = InterpolatingSpline.fromGriddedValues({x, y}, V, S=[3 3]);
Vq = F(Xq, Yq);

4. Noisy fitting quickstart 

t = linspace(0, 1, 50)';
xObs = exp(t) + 0.05*randn(size(t));

noiseModel = NormalDistribution(0.05);
fit = ConstrainedSpline.fromData(t, xObs, S=3, splineDOF=12, distribution=noiseModel);

tq = linspace(t(1), t(end), 200)';
xFit = fit(tq);

From the same starting point you can add robust error models, local point constraints, and global shape constraints. For rectilinear grids in two or more dimensions, switch to ConstrainedSpline.fromGriddedValues({x, y, ...}, values, ...).

5. Planar trajectory quickstart 

t = linspace(0, 1, 24)';
x = cos(2*pi*t);
y = sin(2*pi*t);

trajectory = TrajectorySpline.fromData(t, x, y, S=3);

tq = linspace(t(1), t(end), 200)';
xq = trajectory.x(tq);
yq = trajectory.y(tq);
uq = trajectory.u(tq);
vq = trajectory.v(tq);

Use TrajectorySpline.fromData(...) for raw trajectory samples. The direct TrajectorySpline(t=t, x=xSpline, y=ySpline) constructor is the canonical solved-state path used for persisted restart and other bootstrap cases.

Where to go next

Table of contents