Two problems about ' two-probes system optimization'

[fangyongxinxi]

for the attach files:
"mod" is the image of the calculation two-probes system
"ini.config Vs opt.config " is xyz.file of some elements  in the two-probes system
"opt.py" is exe file we made, and you can find the ini.configuration in this file.
"opt_process.txt" is got by "# atk opt.py > opt_process.txt "

Dear Sir,
I made a two-probes system as the tutorial said, then I made the "opt.py" to optimize this system, success, and got the "opt_process.txt ". In this process, I have two questions:
1, I find two atoms [1.( xyz +0.00000  +0.00000  +4.07800),2.(+1.44179  +7.20895 +29.89500) ]
    are constraint in the "opt-process.txt" , though I set no atoms constraint in "opt.py". why?
2, Question about the optimize result. Compare the atoms xyz.file in the before and after optimization, like the attach " ini.config Vs opt.config ", we find that the scattering region shrink to the center, which means the left surface atoms move to the right, and the right surface atoms move to the left. We tried the different bond distance in the single-atom line, and constraint some surface atoms, but this phenomenon is still happened.

Could you give me some advices on how to deal with these two questions? Thank you so much.

                                                                                                           FangYong   5.10.2010

[fangyongxinxi]

opt_process.txt

[Anders Blom]

The equivalent atoms are not allowed to move in a relaxation, and actually in general you are advised to keep at least the first surface layers closest to the electrodes fixed, to avoid weird results.

The resolution of your image with the coordinates is too low to read. I suggest you post the number as a text file, not a picture.

The effect you find about the atoms moving is probably a result of a large strain on the system. Note that the distance between the Au electrode surfaces isn't changed in the optimization; the atoms can only relax under the constraint of this fixed distance. So, in your system it might be too large. It's hard to optimize this without being able to compute the strain, a feature which will come soon in ATK 2010.xx. One option, for now, is to evaluate the total energy of the system for each value of the distance between the electrodes, and then change it in small steps to find the minimum of the total energy.