Different energy values for constant pressure and volume optimizations for same crystal structure


I have optimized the crystal structure with conp flag and it got converged. Later I have fixed the cell parameters a, b, c and optimized the crystal structure with conv flag. I got different energy values, which supposed to be the same.
Once the system got converged with constant pressure optimization if we fix the cell parameters and reoptimize with a constant volume flag we must get the same energy. However different values have been obtained in my case. Can you explain the reason behind this?

Thank you

Obviously the energy should be the same, if the structure and the force field input is the same. Without seeing the specific input/restart files it’s hard to say anything about why this appears to be the case. If you can provide both files then the reason might be apparent.

I have shared the input and output files of gulp constant pressure and volume optimization files for metal-organic framework via hyperlink for google drive. (The site is not allowing to upload the files directly)

For constant pressure optimization, Final energy value is 461.69 ev
For constant volume optimization, Final energy value is 434.08 ev

Please check the following link for the files within the WEMCOI folder
[WEMCOI - Google Drive]


Thanks for providing the input/output files. I’ve now been able to investigate.
I think something must have gone wrong when you created the input for the constant volume run as you seem to have done this manually. I used GULP’s automatic restart file by adding:

dump every conp.res

and then editing this restart file to change conp to conv and I got the same energy back.
NB: You need to use the latest version of GULP6.0 since I have fixed a couple of issues with passing through the connectivity type and the precision of the uff1 parameters (though these weren’t a factor in your runs)

As a general note, using the “fix” keyword can also lead to different energies on restarting if the connectivity changes between the two initial geometries. However since you’ve supplied the connectivity here then this won’t be a problem for your run as fix makes no difference.

Hi Julian,

Would it be possible to share the input and output files you have run? I still observe different energies for constant pressure and volume optimization after using the dump file.


Hi Sridhar,
Just to make sure, did you install the version of GULP 6.0 dated 29th April? If not then the restart file wouldn’t work. If you have this version and it seems not to have worked then if you send an email to the gulpcode at curtin.edu.au then I’ll email the files to you.


Hi Julian

My apologies for the delay in the message. Thanks for the information, it is working in gulp 6.0. I would like to know what is the problem with gulp 5.2 and why it was showing different energy values for the same structure with constant pressure and constant volume optimization?

If you look in your original constant volume input file you will see that the bonding information contains things like;

connect 33 1 0 0 0 <80>¢9ëþ^?^@^@^@

which is obviously garbage! So the problem was that the bond type “half” wasn’t being written correctly to the restart file. This is what was fixed in the latest version.