[CP2K-user] [CP2K:17580] Re: SGCPMD: number of steps per SCF run increase suddenly after initial 400 MD steps

[Marcella Iannuzzi]

Dear Onofrio, 

To me you system looks quite well behaved. 
The temperature is not low, maybe it helps reducing the time-step of 
activating a variable time-step, to avoid too large moves between two 
integration steps. 
Best
Marcella


On Tuesday, August 30, 2022 at 2:14:03 PM UTC+2 onofr... at gmail.com wrote:

> Dear CP2K community,
>
> I have done a second-generation Car-Parrinello (SGCP) MD simulation for a 
> system of 6 toluene molecules on a 4-layer Cu slab (6x6) at 900K, as 
> follows:
>
> 1. the initial system (including 3 adsorbed toluene + 3 toluene in gas 
> phase, see attached image) was first geometrically optimized
>
> 2. to equilibrate the geometrically-optimized system I used NVE + velocity 
> rescaling looking at the conservation of temperature per region (1°region: 
> Cu slab; 2°region: toluene molecules) by using the &THERMAL_REGION section 
> with DO_LANGEVIN_DEFAULT F and &DEFINE_REGION. See attached 
> “2_step___Equilibration” plot (4 ps of BOMD run)
>
> 3. 4 ps of sgcpMD run. I tuned GAMMA and NOISY_GAMMA per region monitoring 
> again the conservation of temperature per region (see attached plot 
> “3_step___sgcpMD” and text files).
>
>
> However, after about the first 400 sgcpMD steps the SCF runs started to 
> converge in more than 10 steps (between 11 and 16) from 3 steps, as can be 
> seen in the attached .out file. I have tried different settings in the &OT 
> section and I haven’t found any solution to keep the number of steps per 
> SCF run small. 
>
> Note: 
>
> - FULL_KINETIC as preconditioner seems to work better than FULL_ALL.
>
> - every time I restart the same sgcpMD run the first SCF runs converge in 
> few steps (typically 3), but then the number increase to more than 10 steps.
>
> How to keep the number of steps per SCF run small? This behavior could be 
> attributed to a bad initial geometry or exclusively to the &OT section 
> settings?
>
> Any help would be greatly appreciated.
>
> Thanks.
>
> Onofrio
>

-- 
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/2e9059b2-9149-479f-b1c2-0acc0f9bd047n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20220830/40ad62d5/attachment-0001.htm>