QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: Jin-Kyu Choi on May 22, 2014, 14:05
-
Dear all,
I have run a calculation of geometry optimization for a molecular junction of two probe system, in which a phenylene molecule is sandwiched between two gold electrodes.
(please find the input file attached, input.py)
The calculation has been performed for around 40 hours then I have obtained a result file (nc file).
When I clicked the nc result file in VNL window, I could have seen 16 geometries: gID000 -> gID015
(please see the image named "Result_list.gif")
But I found the geometry after optimization is the same with that before optimization.
(please see the image named "geometry.gif")
Here, I would like to ask two questions;
1. Why are there 16 geometries in the nc file from gID000 to gID015? and What are they?
2. Why are the geometries before and after optimizaton calculation the same? and Can I believe that the initial geometry before optimization was optimum?
I hope your help.
Thank you very much.
Sincerely,
-
1. indeed it is a bit weird.
your script looks fine and you save only two DeviceConfiguration objects in your Ph-Au_GO.nc file.
If I have to guess you ran the same job 8 times :)
in this case nlsave "appends" the configurations and does not overwrite the file.
maybe you run in parallel with the wrong setup and you are actually running 8 different jobs?
do you have a log file?
2. you can include a Stress analysis at the end of your optimization.
0.05 eV/Å**3 could be too high. try to decrease.
Also, you save a trajectory file Ph-Au_GO_trajectory.nc
what if you check this trajectory?
-
Dear Umberto Martinez,
I perform calculations using parallel-computing.
But, since I'm very poor at computing, I run the calculations under guide from a institute in Korea.
As you pointed, I may have run the same calculation jobs.
I have linked the log file. Please refer to it if it is necessary.
(http://bigmail.mail.daum.net/Mail-bin/bigfile_down?uid=4zBHnQlYkFZ2qzDJGrQPDxKdPKqR2K7X)
So... the calculation was run??
And the output was not save??
-
yes you ran 8 times the same job.
If you check you log file you'll see there is some messy parts and in particular if you check at the end you'll see many time reports.
this means that you did not parallelize your calculations.
which also means, if you manage to do so, it will take much less then 1 day and 16 hours.
please check (or tell your admin) to check our manual page here http://quantumwise.com/documents/tutorials/latest/ParallelGuide/index.html/
So, the job is done and it performed the (8) geometry optimization.
You should have 8 DeviceConfigurations objects corresponding to the initial state and 8 corresponding to the optimized geometry.
Check them and you should find the optimized geometry.
I would probably double check by running another geometry optimization starting from here.
-
Dear Umberto Martinez,
Thank you very much for your kind instruction.
I will discuss with the manager in the institute. I hope I could solve this issue.
Thank you again.