Expand description
§solow-linalg
Pure-Rust dense linear algebra for the Solow statistical stack. No system LAPACK/BLAS is required: the decompositions are implemented here and validated against an authoritative reference.
Structs§
- Lu
- An LU factorization with partial (row) pivoting:
P A = L U.
Functions§
- cholesky
- Lower-triangular Cholesky factor
Lsuch thatA = L Lᵀ. - cholesky_
solve - Solve a symmetric-positive-definite system
A x = bvia Cholesky (A = L Lᵀ). - det
- Determinant of a square matrix (product of LU pivots times the swap sign).
- eigh
- Eigendecomposition of a real symmetric matrix via the cyclic Jacobi method.
- inv
- Inverse of a square matrix.
- lstsq
- Minimum-norm least-squares solution to
A x ≈ bvia the pseudoinverse. - lstsq_
qr - Fast least-squares for a tall, full-column-rank design via Householder QR.
- lu_
factor - Compute the LU factorization of a square matrix with partial pivoting.
- matrix_
rank - Numerical rank of a matrix using the reference convention
tol = max(s) * max(m, n) * eps. - pinv
- Moore–Penrose pseudoinverse via the SVD, together with the singular values.
- pinv_
rcond pinvwith an explicit relative cutoff.- qr
- Reduced (economy) Householder QR of a tall-or-square matrix
A(m ≥ n): returns(Q, R)withQof shapem × n(orthonormal columns) andRupper-triangular of shapen × n, such thatA = Q R. - solve
- Solve the square linear system
A x = b. - solve_
matrix - Solve
A X = Bfor a matrix right-hand side. - svd
- Singular value decomposition
A = U diag(s) Vᵀ(economy / thin form).