Error in Geometry Optimization - Boundary Constraints

[TheGreatFox]

Hello,

I am encountering the following error in v2016.4 on Windows 10 64-bit:
** Back Engine Exception : No boundary constraints set.
** Location of Exception : fastfourier2dcalculator.cpp:532

What I am attempting to do:
- Geometric optimization of a Cu(111) 6x6 surface in order to prepare it for adsorption of DBP molecule

Process:
As a new user, I have been following  the tutorials (Geometry Optimization and Molecule-Surface systems) and then adjusting them to my particular project, but now I am encountering the above error and am stuck. I prepared the Cu(111) 6x6 surface by first creating the bulk Cu, cleaving it, and relaxing it. To save some time during test, I simply created two layers and fixed one of them, allowing the upper layer to relax. After a number of relaxation steps, the system returns the above error, and atkpython.exe stops working.

Any thoughts on how to fix this?

[Jess Wellendorff]

This should not happen. Could you attach the ATK Python script that produces this error? Thanks!

[TheGreatFox]

Thanks for the response. Attached in this message are Cu(111).py and the relevant output file of the last run that I tried. In this case, I attempted to constrain just in the X, Y directions to see if maybe that might reveal more information about the bug, but the result was the same as runs without constraints. I tried this a number of different ways just to see if it was a fluke, but the same error kept popping up. Initially I thought it might be a bug when I had no lattice constraints, but since I was not applying any pressure, there would be no need. Maybe something I misconfigured/unchecked incorrectly?

[Anders Blom]

This might be caused by atoms too close to the cell edges, when using the FFT2D Poisson solver. If you center the slab in the middle of the cell, it might work better, but FFT will work fine too and is faster.

Edited, I read the log file a bit wrong

[TheGreatFox]

Thanks for the response. Just rechecking at the moment with both FFT and then a recentered slab. I don't know why I didn't check just FFT earlier, but it already is showing a faster convergence. I will let you know the final results today and then hopefully this thread can be closed.