template<typename Scalar_, int Rows_, int Cols_>
class Eigen::Cauchy< Scalar_, Rows_, Cols_ >
An m x n Cauchy matrix represented by its two node vectors.
A Cauchy matrix has entry (i,j) equal to \( 1/(x_i - y_j) \); the class stores only the m + n nodes. Products are evaluated directly at O(mn) operations – the same cost as a dense product, but with O(m+n) storage and without ever forming the matrix. (Fast approximate products via multipole expansions are out of scope.)
The class is closed under transposition: \( C(x,y)^T = C(-y,-x) \), and the determinant of a square Cauchy matrix has the classical closed form \( \prod_{i<j}(x_j-x_i)(y_i-y_j) \big/ \prod_{i,j}(x_i-y_j) \).
Because operator* returns an Eigen product expression, a Cauchy also drops into the matrix-free iterative solvers, and it can be assigned to a dense matrix when an explicit representation is needed. As with any matrix-free operator, the iterative solvers must be instantiated with IdentityPreconditioner (e.g. GMRES<Cauchy<double>,IdentityPreconditioner>): the default preconditioners read individual coefficients through col() or InnerIterator, which the structured operators do not expose.
Square systems are solved in O(n^2) by CauchyLU, a partially pivoted LU factorization computed through the displacement structure (Gohberg-Kailath-Olshevsky): unlike the Toeplitz/Levinson world, Cauchy structure survives row permutations, which is what makes fast pivoted factorization possible. The Hilbert matrix is the Cauchy matrix with x_i = i+1, y_j = -j.
All nodes x_i must differ from all nodes y_j (entries would be infinite otherwise); this is not checked.
- Template Parameters
-
| Scalar_ | the scalar type, real or complex. |
| Rows_ | the number of rows at compile time, or Dynamic (the default). |
| Cols_ | the number of columns at compile time, or Dynamic (the default). |
- Examples
- a discretized Cauchy integral, polynomial least squares.
- See also
- class CauchyLU, makeCauchy()