[Wien] 18.1 mixer (Mango)

Laurence Marks L-marks at northwestern.edu
Mon Jul 9 15:16:37 CEST 2018


As people start to use 18.1, I want to draw attention to a few points as
there are differences in the mixer.

   1. The treatment of the plane waves and density inside the spheres is
   different, more rigorous. There should be very little difference in final
   results (within mRyd or less in energy).
   2. The treatment of orbital potentials is better. It is still important
   to check that :MV has converged, as it remains possible to get a false
   fixed-point solution.
   3. Typically Mango uses much smaller GREED than previous versions. Since
   a more rigorous last-step process is used, this should not be an issue.
   (How much smaller surprised me.)
   4. The mixer now prints out the "Pseudo Charge", the plane wave density
   inside the muffin tins; there is a brief discussion in README_10.1. While,
   with exact arithmetic this does not matter, with a finite plane wave
   expansion and also in MSR1a it is *very* important in many cases. I
   strongly suspect that many cases of bad convergence are related to the
   Pseudo Charge, perhaps even ghost bands. (The Pseudo Charge issues have
   been present in all Wien releases, they are not new, but I don't think
   there importance was appreciated.)
   5. The mixer still has the older MSEC3 option. In some cases this is
   more stable. MSECa (MSEC3 + position optimization) works, but might get
   stuck in local minima or traps.
   6. As with all previous versions, no mixer can cope well with a badly
   posed physical (or chemical) problem.


-- 
Professor Laurence Marks
"Research is to see what everybody else has seen, and to think what nobody
else has thought", Albert Szent-Gyorgi
www.numis.northwestern.edu ; Corrosion in 4D: MURI4D.numis.northwestern.edu
Partner of the CFW 100% program for gender equity, www.cfw.org/100-percent
Co-Editor, Acta Cryst A
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20180709/ec36e112/attachment.html>


More information about the Wien mailing list