I have been trying to obtain the lattice constant of LiF by straining the lattice (scale) starting with the example found in the distribution. The lattice constant derived from the Murnaghan fit is pretty close (1%-3%) to values reported in PRB (79, 085104, 2009), which uses Wien2k. However, energy values near the minimum energy deviate significantly from the expected trend. (See plot below.) Each solution on the energy curve is converged. I performed the same analysis (as closely as I can determine it's the same) with exciting, and the energy curve behaves exactly as expected (no discontinuity near the minimum.) In terms of output, the INFO.OUT files for both exciting and elk are essentially the same except the total number of orbitals is 18 for exciting and 10 for elk. Moreover, the values in the two files LINENGY.OUT from elk and exciting are significantly different. I have tried to determine whether exciting is using some default settings that are different than elk's default settings, but I have not found anything that helps explain my problem.
Thanks for any insight. By the way, I love how easy-to-use elk is. Great work.
On further study, I have found that the energy is correct for version 1.0.0, but breaks in the next release (1.0.16) and all subsequent releases. Also, results using LDA do not appear to exhibit this anomalous behavior.
Log in to post a comment.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.