[CP2K-user] [CP2K:18258] Re: TS Dimer method gives unphysical results

[DMT]

Also tried the following settings:

&MOTION
    &GEO_OPT
        TYPE TRANSITION_STATE
        OPTIMIZER CG
        MAX_ITER 2000
        RMS_DR 0.10E-04                    !DEF 1.5E-3
        MAX_DR 0.30E-04                    !DEF 3.0E-3
        RMS_FORCE 0.10E-04                !DEF 3.0E-4
        MAX_FORCE 0.30E-04                !DEF 4.5E-4
        &CG
              MAX_STEEP_STEPS 0
              &LINE_SEARCH
                   TYPE 2PNT                                    !WORKS WITH 
CG
                   &2PNT
                     MAX_ALLOWED_STEP 0.20                !DEF 2.5E-1
                   &END 2PNT
              &END LINE_SEARCH
        &END CG
        &TRANSITION_STATE
            METHOD DIMER
            &DIMER
                DR 1.0E-2                            !DEF 1.0E-2
                ANGLE_TOLERANCE 2.0                    !DEFAULT 
8.72664626E-02
                INTERPOLATE_GRADIENT T                 !DEFAULT T
                &ROT_OPT
                    OPTIMIZER CG
                    &CG
                        MAX_STEEP_STEPS 0
                        &LINE_SEARCH
                            TYPE 2PNT                !WORKS WITH CG
                            &2PNT
                                MAX_ALLOWED_STEP 1.0E-1                !DEF 
2.5E-1
                            &END 2PNT
                        &END LINE_SEARCH
                    &END CG
                &END ROT_OPT
            &END DIMER
        &END TRANSITION_STATE
    &END GEO_OPT
&END MOTION

and nothing changed.
On Monday, December 26, 2022 at 1:15:50 AM UTC+2 DMT wrote:

> and files from the second reaction
>
> On Monday, December 26, 2022 at 1:14:34 AM UTC+2 DMT wrote:
>
>> Hello,
>>
>> I tried to use the Dimer method of optimizing TS and I get unphysical 
>> results.
>>
>> I used this google group to find different input files and read about the 
>> struggles to overcome and used the provided input settings of successful 
>> runs.
>>
>> In both cases (files attached include input, required .xyz and output 
>> .xyz from the optimization) Dimer drives the structures into having 
>> impossibly short bonds, until the SCF does not even want to converge (and 
>> it would be obviously impossible). It literally tries to merge some atoms 
>> in both cases. I use the &DIMER_VECTOR to input the XYZ displacements from 
>> a VIBRATIONAL_ANALYSIS job for the Reaction Coordinate that I am interested 
>> in, using the guessed near-TS geometries. Without the DIMER_VECTOR the 
>> method always chooses a wrong Reaction path. In both cases the TS geometry 
>> I start with is very simple to provide, in fact I have found both TS with 
>> other software, it is just very important for me to try simple systems in 
>> cp2k before I move to actual research.
>>
>> One of the reactions is substitution in which the N LP from NH3 attacks 
>> the C of CH3Cl and pushes the Cl away to form a quaternary ammonium cation 
>> with the Cl- around. The other reaction is monomolecular - a keto-O taking 
>> a H from a nearby amino group into an alcohol-imin as a Product.
>>
>> I have already tried higher values for EPS_DEFAULT (1.0E-13) and EPS_SCF 
>> (1.0E-7), and nothing changes.
>>
>> I feel like the Dimer simply doesn't know where to stop with increasing 
>> the Energy. Perhaps the choice of DFT (PBE) and basis (DZVP-MOLOPT-GTH (a 
>> good basis)) is bad for the particular PES ?
>>
>> I would like any advises at all. Do you have success using the Dimer 
>> method ? Can you provide input files from successful runs, other than the 
>> ones already in the group conversations ?
>>
>> Best Regards,
>> Dobromir
>>
>>
>>

-- 
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/53b199cd-9b76-419b-a5d2-6faeeb1f5ef6n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20221226/d87faf9d/attachment.htm>