Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly  
We are NATO-free zone

Re^3: Error: Failure to locate stationary point. Too many steps taken..

Siddheshwar Chopra

Dear Luca,
Thank you for your fruitful suggestions. I have done the way you had suggested. I wish to ask you that the Gradient obtained after convergence is 0.0000954 (less than 0.0001). I get the message "Equilibrium geometry located". Is is assured that it is a stationary point? How do I confirm this? I don't want to run Hessian calculation to confirm this. Can I reduce the OPTTOL to a lower value and perform optimization again? Please suggest.


On Thu Dec 20 '12 8:25pm, Luca Maidich wrote
>Dear Siddheshwar,
>in my opinion your OPTTOL is ok and your NSTEP value probably is the reason of the failed convergence, you can do one of the following things (or both just to test)
>a) restart your job with the same input file but using a higher NSTEP, say for example 100;
>b) start a new job with the coordinates that have the lowest energy.

>Regarding point b) you can do this by using Gabedit (under Windows, MacOS X and Linux) or if you have Linux issuing the following command from a terminal window

>---begin code---
>grep 'FINAL ENERGY' output_file
>---end code---

>and looking in the output file to find the number and locate the geometry.

>I guess that DIIS=.T. refers to the $SCF group of the input file. I'm not too deep into the details of the code but I think that they are two different things: with METHOD=GDIIS in $STATPT you ask the software to use the "Geometry optimisation using Direct Inversion in the Iterative Subspace" in the search for the minimum, while with DIIS=.T. in $SCF you ask to use Pulay interpolation which is a convergence accelerator.

>If the geometry still not reaches convergence with these modifications in the output file, can you please attach both the input and output files? I suggest you to compress them in a zip archive.

>Kind regards

[ Previous ] [ Next ] [ Index ]           Mon Dec 24 '12 8:15pm
[ Reply ] [ Edit ] [ Delete ]           This message read 1059 times