torch.eig

torch.eig(input, eigenvectors=False, *, out=None) -> (Tensor, Tensor)

Computes the eigenvalues and eigenvectors of a real square matrix.

Note

Since eigenvalues and eigenvectors might be complex, backward pass is supported only if eigenvalues and eigenvectors are all real valued.

When input is on CUDA, torch.eig() causes host-device synchronization.

Parameters

• input (Tensor) – the square matrix of shape $(n \times n)$ for which the eigenvalues and eigenvectors will be computed
• eigenvectors (bool) – True to compute both eigenvalues and eigenvectors; otherwise, only eigenvalues will be computed

Keyword Arguments

out (tuple, optional) – the output tensors

Returns

A namedtuple (eigenvalues, eigenvectors) containing

• eigenvalues (Tensor): Shape $(n \times 2)$ . Each row is an eigenvalue of input, where the first element is the real part and the second element is the imaginary part. The eigenvalues are not necessarily ordered.
• eigenvectors (Tensor): If eigenvectors=False, it’s an empty tensor. Otherwise, this tensor of shape $(n \times n)$ can be used to compute normalized (unit length) eigenvectors of corresponding eigenvalues as follows. If the corresponding eigenvalues[j] is a real number, column eigenvectors[:, j] is the eigenvector corresponding to eigenvalues[j]. If the corresponding eigenvalues[j] and eigenvalues[j + 1] form a complex conjugate pair, then the true eigenvectors can be computed as $\text{true eigenvector}[j] = eigenvectors[:, j] + i \times eigenvectors[:, j + 1]$ , $\text{true eigenvector}[j + 1] = eigenvectors[:, j] - i \times eigenvectors[:, j + 1]$ .

Return type

(Tensor, Tensor)

Example:

Trivial example with a diagonal matrix. By default, only eigenvalues are computed:

>>> a = torch.diag(torch.tensor([1, 2, 3], dtype=torch.double))
>>> e, v = torch.eig(a)
>>> e
tensor([[1., 0.],
        [2., 0.],
        [3., 0.]], dtype=torch.float64)
>>> v
tensor([], dtype=torch.float64)

Compute also the eigenvectors:

>>> e, v = torch.eig(a, eigenvectors=True)
>>> e
tensor([[1., 0.],
        [2., 0.],
        [3., 0.]], dtype=torch.float64)
>>> v
tensor([[1., 0., 0.],
        [0., 1., 0.],
        [0., 0., 1.]], dtype=torch.float64)