Re: Can extending MIN solve the problem of "Constraint failure in RATTLE algorithm"

From: dimka (
Date: Sat Dec 24 2005 - 19:51:31 CST

Hi Shirley,

The atoms you're talking about may be "tangled". If that is the case, you
can use VMD's move atom/residue option to help bring the atom in the right

When you plot the energies from NAMD log you'll see that this value
converges or reaches a plateau at some point. From experience, it takes
about 300-2000 steps to reach that point. Increasing the number of steps
won't do much.


On 12/24/05, Shirley Li <> wrote:
> Dear NAMD experts,
> After finishing mimization, I tried running heating the system, but got
> the following message from heating:
> ERROR: Constraint failure in RATTLE algorithm for atom 29326!
> ERROR: Constraint failure; simulation has become unstable.
> ERROR: Exiting prematurely
> I checked the minimized structure, the atom 29326 and the corresponding
> residue are too far from its two neighboring residues, just as what the
> related messages in NAMD archive suggested. The number of steps for the
> minimization was 26000 for fix-backbone MIN + 6000 for restraint-free MIN.
> I'm thinking of increasing them to 56000 + 26000 respectively. I have some
> questions here:
> 1) Is it possible to solve this problem by extending the minimization?
> 2) Are 56000 +26000 steps! too long?
> 3) Is there any other better way for this problem?
> Your help will be greatly appreciated.
> Shirley
> ------------------------------
> Yahoo! DSL<**>Something to write home about. Just $16.99/mo. or less

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:41:28 CST