Vandermonde Node Conditioning

Vandermonde Node Conditioning Theory

A Vandermonde matrix is defined by a set of nodes or points x_0, x_1, …, x_(n-1). For a row-node configuration, the i-th row corresponds to the powers of the node x_i, such that V_(i,j) = x_i^j. For a 3 × 3 matrix: V = [[1, x_0, x_0²], [1, x_1, x_1²], [1, x_2, x_2²]]

Worked Example

Consider the standard separated base nodes: x = (-1, 0, 2), with shift 0 and scale 1. The Vandermonde matrix V is:

V = [
  [1, -1,  1],
  [1,  0,  0],
  [1,  2,  4]
]

The determinant of V is given by the product of the differences between nodes: det(V) = (x_2 - x_1)(x_2 - x_0)(x_1 - x_0) = (2 - 0)(2 - (-1))(0 - (-1)) = 2 × 3 × 1 = 6.

Experiment

You can interact with the Vandermonde Node Conditioning module to see how clustering and shifting nodes affect the condition number.

Change the Nodes to Clustered. Notice the nodes bunch together near 0.5. The determinant decreases to 0.25, and the condition number increases to 24.
Change the Shift to 1 (for Separated nodes). The nodes shift to (0, 1, 3). The determinant remains 6.0, but the condition number increases to 39.
Select the Repeated nodes configuration. The nodes become (0, 0, 1), and the determinant drops to 0.0, making the matrix singular (rank 2) with an unavailable finite condition number.

Limitations

The condition number \kappa_∞ = ‖V‖_∞ ‖V⁻¹‖_∞ is highly sensitive to the scale and shift of the nodes. While a small determinant is often associated with ill-conditioning, it is not by itself a condition number—a matrix can have a small determinant but still be well-conditioned if scaled appropriately. This module restricts exploration to a closed domain of 3 × 3 matrices to provide a tractable exact inverse calculation and clear visual representation. Singular matrices do not yield a finite condition number, and polynomial interpolation ceases to be uniquely defined.

References

For further reading on the properties and applications of Vandermonde matrices, including detailed discussions on their condition numbers, you can consult the following primary resource:

What Is a Vandermonde Matrix? - Nick Higham: An insightful article by Nick Higham that explains the fundamentals of Vandermonde matrices and the impact of their node properties.

Vandermonde Node Conditioning