[CP2K-user] [CP2K:17424] Geometry Optimization with Brent's Algorithm: CP2K and VASP

[Nicholas Winner]

I reduced the trust radius and the relaxation is dramatically more stable. 
In the manual it says the trust radius used to be 0.1 but was increased to 
0.25 for some reason. Not sure why, stability seems pretty good with 0.1

Thanks for helping.

Nick

On Monday, July 25, 2022 at 8:58:53 AM UTC-7 Matthias Krack wrote:

> Hi Nick
>
>  
>
> Such convergence problems are usually caused by noisy forces and/or a very 
> flat PES. Did you also try a smaller trust radius (e.g. 0.1) and/or 
> rational function optimization with BFGS?
>
>  
>
> Matthias
>
>  
>
> *From: *"cp... at googlegroups.com" <cp... at googlegroups.com> on behalf of 
> Nicholas Winner <nwi... at berkeley.edu>
> *Reply to: *"cp... at googlegroups.com" <cp... at googlegroups.com>
> *Date: *Sunday, 24 July 2022 at 21:29
> *To: *"cp... at googlegroups.com" <cp... at googlegroups.com>
> *Subject: *Re: [CP2K:17365] Geometry Optimization with Brent's Algorithm: 
> CP2K and VASP
>
>  
>
> Hi Matthias, normally I use BFGS.  In my case of an adsorbate on metal, 
> BFGS just hops around the minimum with no clear progress. CG (2pnt) does 
> slowly work toward the minimum which is why I switched to it, its just 
> slower.
>
>  
>
> I disabled USE_MODEL_HESSIAN and the convergence become much worse.
>
>  
>
> Still looking at some variations of the settings in "GOLD" line search to 
> match Filipe's recommendation, but to no avail. So, so far, 2 point line 
> search seems the best option, but there's something better hiding in the 
> settings I can feel.
>
> On Friday, July 22, 2022 at 3:44:39 AM UTC-7 Matthias Krack wrote:
>
> Hi Nick
>
>  
>
> According to my experience BFGS works best (fastest) in GEO_OPT and 
> CELL_OPT runs. I suggest to disable the default USE_MODEL_HESSIAN with 
> BFGS. For larger systems (>3000 atoms), I suggest LBFGS.
>
>  
>
> Matthias 
>
>  
>
> *From: *"cp... at googlegroups.com" <cp... at googlegroups.com> on behalf of 
> Nicholas Winner <nwi... at berkeley.edu>
> *Reply to: *"cp... at googlegroups.com" <cp... at googlegroups.com>
> *Date: *Friday, 22 July 2022 at 01:13
> *To: *"cp... at googlegroups.com" <cp... at googlegroups.com>
> *Subject: *[CP2K:17348] Geometry Optimization with Brent's Algorithm: 
> CP2K and VASP
>
>  
>
> Hi all,  
>
>  
>
> I'm working with some people that are using VASP while I am using CP2K. 
> While CP2K has really fast SCF procedure, it seems to be slower when it 
> comes to geometry optimization for some reason.
>
>  
>
> I've dug into VASP's manual and they claim that their CG method uses "a 
> modified version of Brent's algorithm." No further details are given. 
> Normally with CP2K, I use 2PNT line search with my CG minimizations, and I 
> think this is pretty standard for everyone else. I do know that CP2K 
> supports FIT and GOLD line search options which use Brent's algorithm 
> (actually I don't 100% know how FIT and Brent's Algorithm are related, but 
> gold uses it at least). However, GOLD line search is very slow in CP2K, as 
> we are warned in the manual.
>
>  
>
> So, my hunch, is that CP2K with some type of gold or fit line search would 
> be able to match this "modified Brent's algorithm" and provide faster 
> minimization than what I am currently getting. Are there some insights that 
> people can provide for choosing settings for gold line search that can give 
> decent results? To remind, the flags we have to play with are:
>
>
> BRACK_LIMIT {Real}
> *Limit in 1D bracketing during line search in Conjugate Gradients 
> Optimization.*
> Default value: 1.00000000E+002
>
>
> BRENT_MAX_ITER {Integer}
> *Maximum number of iterations in brent algorithm (used for the line search 
> in Conjugated Gradients Optimization)*
> Default value: 100
>
> BRENT_TOL {Real}
> *Tolerance requested during Brent line search in Conjugate Gradients 
> Optimization.*
> Default value: 1.00000000E-002
>
> INITIAL_STEP {Real}
> *Initial step size used, e.g. for bracketing or minimizers. Might need to 
> be reduced for systems with close contacts*
> Default value: 2.00000000E-001
>
>  
>
> Could it be that VASP uses a loose value for BRENT_TOL, or maybe a small 
> number of max steps? Furthermore, what is the tolerance in the algorithm? 
> Tolerance for the atomic position refinement, or forces?
>
>  
>
>  
>
> If anyone has any ideas, it would be much appreciated.
>
>  
>
> -Nick
>
> -- 
>
> 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+uns... at googlegroups.com.
> To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cp2k/c3f8726c-7587-4240-8e97-c2a97f3ebdafn%40googlegroups.com 
> <https://groups.google.com/d/msgid/cp2k/c3f8726c-7587-4240-8e97-c2a97f3ebdafn%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>
> -- 
> 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+uns... at googlegroups.com.
>
> To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cp2k/b4d40576-d050-4d64-944c-37d9aad53ae5n%40googlegroups.com 
> <https://groups.google.com/d/msgid/cp2k/b4d40576-d050-4d64-944c-37d9aad53ae5n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>
>

-- 
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/18c266b9-90ad-47c7-a245-b2580761ec03n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20220730/00e6aeba/attachment-0001.htm>