lapack_eig_f32

void lapack_eig_f32(const float * A, int n, float * wr, float * wi, float * VR, int * info)

source edit

General (non-symmetric) eigendecomposition (float32).

Computes eigenvalues of a general real matrix; complex eigenvalues are returned as separate real (wr) and imaginary (wi) parts. Right eigenvectors (columns of $V$ ) are written to VR if non-null.

Math

$A v_j = \lambda_j v_j, \quad \lambda_j = \mathrm{wr}_j + i \, \mathrm{wi}_j$

References

LAPACK sgeev_.

Parameters

Aconst float*

Row-major

n \times n

input.

nint

Side length.

wr, wifloat*

Real / imaginary parts of the eigenvalues, length

n

VRfloat*

Right-eigenvector matrix (row-major) or nullptr to skip vectors.

infoint*

LAPACK status (> 0 = QR iteration failed to converge).