[QE-users] bfgs failed after 11 scf cycles and 10 bfgs steps, convergence not achieved

Pionteck, Mike Nico Mike.Pionteck at theo.physik.uni-giessen.de
Wed Mar 8 14:29:07 CET 2023


Dear users,

I am using QE version 7.1 and have an issue with pw.x. I want to perform calculations under uniaxial stress. Therefore, I first want to optimize my unit cell with vc-relax to a stress tensor of (almost) zero. In order to reduce the Pulay stress, I increased ecutwfc =85.0 and decreased forc_conv_thr = 1e-5. Then the calculation runs normally. However in the last bfgs step I get the following message

     Message from routine bfgs:
     history already reset at previous step: exiting
     Energy error            =      9.7E-09 Ry
     Gradient error          =      2.2E-05 Ry/Bohr
     Cell gradient error     =      7.2E-03 kbar

     bfgs failed after  11 scf cycles and  10 bfgs steps, convergence not achieved
     (criteria: energy <  1.0E-05 Ry, force <  1.0E-05 Ry/Bohr, cell <  5.0E-01 kbar)

although the energy and the forces seem to be converged. The message doesn’t appear for calculations with lower ecutwfc and  forc_conv_thr = 1e-5 or higher forc_conv_thr and ecutwfc =85.0. So, the issue seems to be the combination of  ecutwfc =85.0 and forc_conv_thr = 1e-5, but I would like to use these settings. Does anyone know how I can fix this problem?

Please find the input and output attached:

https://www.dropbox.com/scl/fo/23qm5su4cwaottsiegclq/h?dl=0&rlkey=bc1m6modrjhqvdjpagduxktda

—
Mike Pionteck
Justus-Liebig-Universität Gießen
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20230308/3c061be8/attachment.html>


More information about the users mailing list