[...] = SBEig (...)
find a few eigenvalues of the symmetric, banded matrix
inverse power iteration is used for the standard and generalized
eigenvalue problem
[Lambda,{Ev,err}] = SBEig(A,V,tol) solve A*Ev = Ev*diag(Lambda)
standard eigenvalue problem
[Lambda,{Ev,err}] = SBEig(A,B,V,tol) solve A*Ev = B*Ev*diag(Lambda)
generalized eigenvalue problem
A is mxt, where t-1 is number of non-zero superdiagonals
B is mxs, where s-1 is number of non-zero superdiagonals
V is mxn, where n is the number of eigenvalues desired
contains the initial eigenvectors for the iteration
tol is the relative error, used as the stopping criterion
X is a column vector with the eigenvalues
EV is a matrix whose columns represent normalized eigenvectors
err is a vector with the aposteriori error estimates for the eigenvalues
