forces in qdata.txt

[qasimpars]

Hi,

I would use qdata.txt output file for fitting the paramaters using ForceBalance but torsiondrive doesn't write the forces into qdata.txt. Why does it not write the forces to the output file (qdata.txt)? My input files are below:

Input.data:

memory 2 GB
molecule {
  0 1
  C  0.281   0.772   0.721
  O  -0.311   2.003   0.362
  H  -0.070   0.470   1.720
  H  1.371   0.871   0.723
  H  -0.011   0.011  -0.001
  H  0.021   2.231  -0.521
}
set {
  basis 6-31g_d_
}
set_num_threads(8)
gradient('b3lyp')

dihedrals.txt:

# dihedral definition by atom indices starting from 0
# i     j     k     l
  3     1     2     6

Thanks,

[yudongqiu]

The forces are not written in qdata.txt by TorsionDrive for several reasons:

1. TorsionDrive do not work with gradients directly. The only data TorsionDrive uses are structures and energies. It launches "constrained optimizations", and geomeTRIC or another geometry optimizer handles the gradients and generate new structures.
2. Writing out qdata.txt is a convenient feature after all the scan is finished, but even if TorsionDrive wants to use the gradients information in this step, geomeTRIC currently do not provide it.
3. We usually use the final structures provided by TorsionDrive and run a new batch of single point calculations with a higher level of theory. In this way you will get the final gradients there.

After all, we agree that not having gradients in the qdata.txt is not super convenient, and we will think about adding the "provide gradients" feature in geomeTRIC then let TorsionDrive use it.

[yudongqiu]

Closing this issue since this feature is implemented.