Getting Started#

Installation#

SplineBox can be installed using pip:

pip install splinebox

This will install the minimal dependencies. To install SplineBox with the dependencies required for running examples, use:

pip install "splinebox[examples]"

Similarly, you can install dependencies for building the documentation or running the test suite:

pip install "splinebox[test]"

or

pip install "splinebox[docs]"

To install all dependencies:

pip install "splinebox[all]"

Note: If you are unfamiliar with spline terminology, such as knots and control points, refer to our theory introduction.

Constructing a spline with SplineBox is straightforward. Decide on the following:

The number of knots/control points your spline should have.
The type of spline you want to construct, i.e., choose a basis function.
Whether your spline should be closed or open (i.e., whether the two ends are connected to form a loop).

For this example, we will construct a cubic B-spline with 5 knots that is not closed.

import splinebox
spline = splinebox.Spline(M=5, basis_function=splinebox.B3(), closed=False)

At this point, we haven’t set any control points or knots. To shape the spline in ndim dimensions, you have three options:

Directly provide the knots:
```
spline.knots = np.random.rand(5, ndim)
```
Directly provide the control points:
```
spline.control_points = np.random.rand(7, ndim)
```
Note that 7 control points are needed (instead of 5) because splines are padded at the end.
Approximate data with a least squares fit:
```
spline.fit(np.random.rand(100, ndim))
```
Learn more about how fit works in the theory section on Data approximation or by checking the API splinebox.spline_curves.Spline.fit().

Knots are conventionally placed at integer values of the parameter t starting from zero. To sample an open spline with M knots:

t = np.linspace(0, M-1, 100)
vals = spline(t)

For a closed spline, continue past the last knot to sample all the way around:

t = np.linspace(0, M, 100)
vals = spline(t)