[CP2K-user] [CP2K:17336] Re: Geometry doesn't converge

[Marcella Iannuzzi]

Dear Antun 

The problem is not Cholesky decompose. 
That error message appeared for a first run (see your log) where the 
EPS_SCF was set to
eps_scf:                                        5.00E-01
which is a very large value. The consequence is very wrong forces leading 
to unphysical structures and whatever possible error.

After that first run,  the simulation has been repeated (see log). Maybe 
starting again from the original structure or maybe form one of the 
unphysical structures that were generated in the previous run. I don't 
know. 
Anyway, there the eps_scf is set to a reasonable value
eps_scf:                                        5.00E-07
but the problem is that every optimisation of the electronic structure does 
not succeed, hence again wrong forces are provided and the dynamics is also 
wrong. 

The suggestion is: before starting MD, get a reasonable electronic 
structure for a reasonable (equilibrium?) initial structure. Check that the 
electronic structure model reproduces some known properties of the 
simulated system, testing the usual suspects: functional, cutoff, basis 
set, vdw corrections etc. 
Only if the electronic structure model has been assessed, the MD 
simulations can be trusted, this is for sure a necessary condition. 

Regards
marcella

On Wednesday, July 20, 2022 at 1:30:13 PM UTC+2 antunh... at gmail.com wrote:

> Dear all, 
> I'm trying to run some MD, but the geometry of my system doesn't converge. 
> I get the error:
> Cholesky decompose failed: the matrix is not positive definite or 
> ill-conditioned. 
> Does anyone know how I can fix this?
>
> Thanks in advance,
> Antun.
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/6827337d-8358-4242-93d5-e15b18cbb273n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20220720/f4146601/attachment-0001.htm>