[gmx-users] g_nmeig convergence

[Rossen Apostolov]

Dear All,

I'm calculating the eigenvectors of a rather big system, 5000 atoms.
And here is what I get after a day from
..................................
$g_nmeig_d -f nm.mtx -s nm.tpr

Reading double precision matrix generated by Gromacs VERSION 3.3

Diagonalizing to find eigenvectors 1 through 5...
Calculation Ritz values and Lanczos vectors, max 100000 iterations...
Iteration 103797: 0 out of 5 Ritz values converged.
...................................

As I understand it, there's still no convergence for vector 1,
but why does it continue past the maximum 100,000 ?
How fast does it usually converge?

Before calculation of the Hessian the system was minimized to 1e-6,
if that matters.

Rossen