Why do semi-positive definite matrices have negative eigenvalues when I use the function torch.eig()?

Question

Why do semi-positive definite matrices have negative eigenvalues when I use the function torch.eig()?How can I resolve this problem? enter code here