Linear algebra (scipy.linalg)

Linear algebra functions.

See also

numpy.linalg for more linear algebra functions. Note that identically named functions from scipy.linalg may offer more or slightly differing functionality.

Basics

inv(a[, overwrite_a, check_finite, ...])	Compute the inverse of a matrix.
solve(a, b[, lower, overwrite_a, ...])	Solve the equation a @ x = b for x, where a is a square matrix.
solve_banded(l_and_u, ab, b[, overwrite_ab, ...])	Solve the equation a @ x = b for x, where a is the banded matrix defined by ab.
solveh_banded(ab, b[, overwrite_ab, ...])	Solve the equation a @ x = b for x, where a is the Hermitian positive-definite banded matrix defined by ab.
solve_circulant(c, b[, singular, tol, ...])	Solve the equation C @ x = b for x, where C is a circulant matrix defined by c.
solve_triangular(a, b[, trans, lower, ...])	Solve the equation a @ x = b for x, where a is a triangular matrix.
solve_toeplitz(c_or_cr, b[, check_finite])	Solve the equation T @ x = b for x, where T is a Toeplitz matrix defined by c_or_cr.
matmul_toeplitz(c_or_cr, x[, check_finite, ...])	Efficient Toeplitz Matrix-Matrix Multiplication using FFT
det(a[, overwrite_a, check_finite])	Compute the determinant of a matrix
norm(a[, ord, axis, keepdims, check_finite])	Matrix or vector norm.
lstsq(a, b[, cond, overwrite_a, ...])	Compute least-squares solution to the equation a @ x = b.
pinv(a, *[, atol, rtol, return_rank, ...])	Compute the (Moore-Penrose) pseudo-inverse of a matrix.
pinvh(a[, atol, rtol, lower, return_rank, ...])	Compute the (Moore-Penrose) pseudo-inverse of a Hermitian matrix.
khatri_rao(a, b)	Khatri-Rao product of two matrices.
orthogonal_procrustes(A, B[, check_finite])	Compute the matrix solution of the orthogonal (or unitary) Procrustes problem.
matrix_balance(A[, permute, scale, ...])	Compute a diagonal similarity transformation for row/column balancing.
subspace_angles(A, B)	Compute the subspace angles between two matrices.
bandwidth(a)	Return the lower and upper bandwidth of a 2D numeric array.
issymmetric(a[, atol, rtol])	Check if a square 2D array is symmetric.
ishermitian(a[, atol, rtol])	Check if a square 2D array is Hermitian.
LinAlgError	Generic Python-exception-derived object raised by linalg functions.
LinAlgWarning	The warning emitted when a linear algebra related operation is close to fail conditions of the algorithm or loss of accuracy is expected.

Eigenvalue Problems

eig(a[, b, left, right, overwrite_a, ...])	Solve an ordinary or generalized eigenvalue problem of a square matrix.
eigvals(a[, b, overwrite_a, check_finite, ...])	Compute eigenvalues from an ordinary or generalized eigenvalue problem.
eigh(a[, b, lower, eigvals_only, ...])	Solve a standard or generalized eigenvalue problem for a complex Hermitian or real symmetric matrix.
eigvalsh(a[, b, lower, overwrite_a, ...])	Solves a standard or generalized eigenvalue problem for a complex Hermitian or real symmetric matrix.
eig_banded(a_band[, lower, eigvals_only, ...])	Solve real symmetric or complex Hermitian band matrix eigenvalue problem.
eigvals_banded(a_band[, lower, ...])	Solve real symmetric or complex Hermitian band matrix eigenvalue problem.
eigh_tridiagonal(d, e[, eigvals_only, ...])	Solve eigenvalue problem for a real symmetric tridiagonal matrix.
eigvalsh_tridiagonal(d, e[, select, ...])	Solve eigenvalue problem for a real symmetric tridiagonal matrix.

Decompositions

lu(a[, permute_l, overwrite_a, ...])	Compute LU decomposition of a matrix with partial pivoting.
lu_factor(a[, overwrite_a, check_finite])	Compute pivoted LU decomposition of a matrix.
lu_solve(lu_and_piv, b[, trans, ...])	Solve an equation system, a x = b, given the LU factorization of a
svd(a[, full_matrices, compute_uv, ...])	Singular Value Decomposition.
svdvals(a[, overwrite_a, check_finite])	Compute singular values of a matrix.
diagsvd(s, M, N)	Construct the sigma matrix in SVD from singular values and size M, N.
orth(A[, rcond])	Construct an orthonormal basis for the range of A using SVD
null_space(A[, rcond, overwrite_a, ...])	Construct an orthonormal basis for the null space of A using SVD
ldl(A[, lower, hermitian, overwrite_a, ...])	Computes the LDLt or Bunch-Kaufman factorization of a symmetric/ hermitian matrix.
cholesky(a[, lower, overwrite_a, check_finite])	Compute the Cholesky decomposition of a matrix.
cholesky_banded(ab[, overwrite_ab, lower, ...])	Cholesky decompose a banded Hermitian positive-definite matrix
cho_factor(a[, lower, overwrite_a, check_finite])	Compute the Cholesky decomposition of a matrix, to use in cho_solve
cho_solve(c_and_lower, b[, overwrite_b, ...])	Solve the linear equations A x = b, given the Cholesky factorization of A.
cho_solve_banded(cb_and_lower, b[, ...])	Solve the linear equations A x = b, given the Cholesky factorization of the banded Hermitian A.
polar(a[, side])	Compute the polar decomposition.
qr(a[, overwrite_a, lwork, mode, pivoting, ...])	Compute QR decomposition of a matrix.
qr_multiply(a, c[, mode, pivoting, ...])	Calculate the QR decomposition and multiply Q with a matrix.
qr_update(Q, R, u, v[, overwrite_qruv, ...])	Rank-k QR update
qr_delete(Q, R, k, int p=1[, which, ...])	QR downdate on row or column deletions
qr_insert(Q, R, u, k[, which, rcond, ...])	QR update on row or column insertions
rq(a[, overwrite_a, lwork, mode, check_finite])	Compute RQ decomposition of a matrix.
qz(A, B[, output, lwork, sort, overwrite_a, ...])	QZ decomposition for generalized eigenvalues of a pair of matrices.
ordqz(A, B[, sort, output, overwrite_a, ...])	QZ decomposition for a pair of matrices with reordering.
schur(a[, output, lwork, overwrite_a, sort, ...])	Compute Schur decomposition of a matrix.
rsf2csf(T, Z[, check_finite])	Convert real Schur form to complex Schur form.
hessenberg(a[, calc_q, overwrite_a, ...])	Compute Hessenberg form of a matrix.
cdf2rdf(w, v)	Complex diagonal form to real diagonal block form.
cossin(X[, p, q, separate, swap_sign, ...])	Compute the cosine-sine (CS) decomposition of an orthogonal/unitary matrix.

See also

scipy.linalg.interpolative – Interpolative matrix decompositions

Matrix Functions

expm(A)	Compute the matrix exponential of an array.
logm(A[, disp])	Compute matrix logarithm.
cosm(A)	Compute the matrix cosine.
sinm(A)	Compute the matrix sine.
tanm(A)	Compute the matrix tangent.
coshm(A)	Compute the hyperbolic matrix cosine.
sinhm(A)	Compute the hyperbolic matrix sine.
tanhm(A)	Compute the hyperbolic matrix tangent.
signm(A[, disp])	Matrix sign function.
sqrtm(A[, disp, blocksize])	Compute, if exists, the matrix square root.
funm(A, func[, disp])	Evaluate a matrix function specified by a callable.
expm_frechet(A, E[, method, compute_expm, ...])	Frechet derivative of the matrix exponential of A in the direction E.
expm_cond(A[, check_finite])	Relative condition number of the matrix exponential in the Frobenius norm.
fractional_matrix_power(A, t)	Compute the fractional power of a matrix.

Matrix Equation Solvers

solve_sylvester(a, b, q)	Computes a solution (X) to the Sylvester equation \(AX + XB = Q\).
solve_continuous_are(a, b, q, r[, e, s, ...])	Solves the continuous-time algebraic Riccati equation (CARE).
solve_discrete_are(a, b, q, r[, e, s, balanced])	Solves the discrete-time algebraic Riccati equation (DARE).
solve_continuous_lyapunov(a, q)	Solves the continuous Lyapunov equation \(AX + XA^H = Q\).
solve_discrete_lyapunov(a, q[, method])	Solves the discrete Lyapunov equation \(AXA^H - X + Q = 0\).

Sketches and Random Projections

clarkson_woodruff_transform(input_matrix, ...)

Applies a Clarkson-Woodruff Transform/sketch to the input matrix.

Special Matrices

block_diag(*arrs)	Create a block diagonal array from provided arrays.
circulant(c)	Construct a circulant matrix.
companion(a)	Create a companion matrix.
convolution_matrix(a, n[, mode])	Construct a convolution matrix.
dft(n[, scale])	Discrete Fourier transform matrix.
fiedler(a)	Returns a symmetric Fiedler matrix
fiedler_companion(a)	Returns a Fiedler companion matrix
hadamard(n[, dtype])	Construct an Hadamard matrix.
hankel(c[, r])	Construct a Hankel matrix.
helmert(n[, full])	Create an Helmert matrix of order n.
hilbert(n)	Create a Hilbert matrix of order n.
invhilbert(n[, exact])	Compute the inverse of the Hilbert matrix of order n.
leslie(f, s)	Create a Leslie matrix.
pascal(n[, kind, exact])	Returns the n x n Pascal matrix.
invpascal(n[, kind, exact])	Returns the inverse of the n x n Pascal matrix.
toeplitz(c[, r])	Construct a Toeplitz matrix.

Low-level routines

get_blas_funcs(names[, arrays, dtype, ilp64])	Return available BLAS function objects from names.
get_lapack_funcs(names[, arrays, dtype, ilp64])	Return available LAPACK function objects from names.
find_best_blas_type([arrays, dtype])	Find best-matching BLAS/LAPACK type.

See also

scipy.linalg.blas – Low-level BLAS functions

scipy.linalg.lapack – Low-level LAPACK functions

scipy.linalg.cython_blas – Low-level BLAS functions for Cython

scipy.linalg.cython_lapack – Low-level LAPACK functions for Cython