Octave Programming Tutorial/Linear algebra

Functions edit

  • d = det(A) computes the determinant of the matrix A.
  • lambda = eig(A) returns the eigenvalues of A in the vector lambda, and
  • [V, lambda] = eig(A) also returns the eigenvectors in V but lambda is now a matrix whose diagonals contain the eigenvalues. This relationship holds true (within round off errors) A = V*lambda*inv(V).
  • inv(A) computes the inverse of non-singular matrix A. Note that calculating the inverse is often 'not' necessary. See the next two operators as examples. Note that in theory A*inv(A) should return the identity matrix, but in practice, there may be some round off errors so the result may not be exact.
  • A / B computes X such that  . This is called right division and is done without forming the inverse of B.
  • A \ B computes X such that  . This is called left division and is done without forming the inverse of A.
  • norm(A, p) computes the p-norm of the matrix (or vector) A. The second argument is optional with default value  .
  • rank(A) computes the (numerical) rank of a matrix.
  • trace(A) computes the trace (sum of the diagonal elements) of A.
  • expm(A) computes the matrix exponential of a square matrix. This is defined as
 

Below are some more linear algebra functions. Use help to find out more about them.

  • balance (eigenvalue balancing),
  • cond (condition number),
  • dmult (computes diag(x) * A efficiently),
  • dot (dot product),
  • givens (Givens rotation),
  • kron (Kronecker product),
  • null (orthonormal basis of the null space),
  • orth (orthonormal basis of the range space),
  • pinv (pseudoinverse),
  • syl (solves the Sylvester equation).

Factorizations edit

  • R = chol(A) computes the Cholesky factorization of the symmetric positive definite matrix A, i.e. the upper triangular matrix R such that  .
  • [L, U] = lu(A) computes the LU decomposition of A, i.e. L is lower triangular, U upper triangular and  .
  • [Q, R] = qr(A) computes the QR decomposition of A, i.e. Q is orthogonal, R is upper triangular and  .

Below are some more available factorizations. Use help to find out more about them.

  • qz (generalized eigenvalue problem: QZ decomposition),
  • qzhess (Hessenberg-triangular decomposition),
  • schur (Schur decomposition),
  • svd (singular value decomposition),
  • housh (Householder reflections),
  • krylov (Orthogonal basis of block Krylov subspace).

Return to the Octave Programming Tutorial index