Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

Re^5: RPES scan initial coordinates

Fabian Weber

Hi everybody,

I'm using potential energy surface scans for cyclic compounds.
I had the same problem as Sei. I worked around that problem by
restarting the scan every 10 points with updated geometry.
Reading this thread I tried DLCs!
Interesting is, that I now need about 40 iterations per step,
where I needed about 20 iterations (average) without DLCs!
With some compounds Firefly now needs >100 iterations to find
the initial geometry (although I gave an optimized geometry
in my input-file).
My interpretation is, that the DLCs created by Firefly are not
as good for my problem as my z-matrix, which may of course be,
because I don't know much about DLCs... (I just used the
suggestions provided by Pavel without the additional options
for Sei's problem and the right IFZMAT-group for my problem)
Therefore I think it would be convenient to have such keyword to
reuse the optimized geometry from the previous scan point during
scans in conventional Z-matrix variables.

Best regards,

On Sat Dec 11 '10 5:24pm, Alex Granovsky wrote
>Dear Sei Higashiya,
>Dear Pavel Strashnov,

>Indeed, the behavior of relaxed scan procedure is different
>for scans in DLCs and for scans which use conventional Z-matrix
>variables. This is not a bug but rather by the code design.
>It should not be too difficult to modify code as well as to add
>a keyword or flag to allow reuse of optimized geometry from the
>previous scan point during scans in conventional Z-matrix variables;
>however, we consider this code as obsolete as scans in DLCs are
>clearly much more efficient and robust. On the other hand, it is
>always useful to know opinions of our users, so we'd be glad for
>any comments concerning these two types of relaxed scans.

>Best regards,
>Alex Granovsky
>On Fri Dec 10 '10 2:56am, Pavel Strashnov wrote
>>Hi Sei,

>>you can overcome the problem by using DLCs.

>>To do so simply add
>> $ZMAT dlc=.t. auto=.t. scan=.t. autofv=.t. IFZMAT(1)=3,10,3,2,4 $END
>>and delete vect1(24)=1 from $surf group.

>>Also you may want to add
>>       NONVDW(1)=2,4
>>       hbond=0
>>       ifdmod=2
>> to $ZMAT to improve your results.
>> Please find test job results attached.

>>On Thu Dec 9 '10 6:31am, sei higahsiya wrote
>>>Hi Pavel,

>>>Thank you for your interest. It is as followed.

>>> $SYSTEM TIMLIM=31536000 MEMORY=40000000 $END
>>> $GUESS GUESS=huckel $END
>>> $statpt opttol=1d-5 nstep=1000 IFREEZ(1)=24 $end
>>> $surf ndisp1=37 disp1=5 vect1(24)=1 orig1=0 $end
>>> $DATA
>>>C  1    1.52160
>>>C   2    1.53410  1 113.3733
>>>S   2    1.88637  1 112.6257  3 a1
>>>F   4    1.63249  2  94.5293  1 a2
>>>F   4    1.61622  2  90.6081  1 a3
>>>F   4    1.59735  2 176.9384  1 a4
>>>F   4    1.61180  2  94.9768  1  a5
>>>F   4    1.62172  2  90.2264  1 a6
>>>O   3    1.41096  2 115.1032  4  dx
>>>H   1    1.09237  2 107.2770  3 a7
>>>H   1    1.08607  11 107.9406  2 a8
>>>H   2    1.08902  3 107.4905  1 a9
>>>H   3    1.09146  10 111.8910  2 a10
>>>H   10    0.96211  3 108.7031  5 a11
>>>H   1    1.08691  11 108.0355  12 a12
>>>H   3    1.09283  14 107.3098  10 a13

>>> $END


>>>On Thu Dec 9 '10 5:02am, Pavel Strashnov wrote
>>>>your input/output should be very helpful.


>>>>On Thu Dec 9 '10 3:38am, sei higahsiya wrote
>>>>>Hi all,

>>>>>I am performing relaxed potential energy surface scan using Firefly 71G. The problem is when the step proceeds, the initial coordinates except the frozen one keep coming back to the original input values in the $data section rather than the relaxed coordinates in the previous step.

>>>>>Is there any way to use the relaxed coordinates in the previous step as an initial coordinate in the next step?

>>>>>Thanks in advance.

>>>>>Sei Higashiya

[ Previous ] [ Next ] [ Index ]           Mon Dec 13 '10 12:12pm
[ Reply ] [ Edit ] [ Delete ]           This message read 767 times