[Wien] interstitial magnetic moment in AFM calculations

[Yin, Weiguo]

Dear WIEN developers and users:

[Questions] 
Why wasn't the interstitial magnetic moment (MMINT) constrained to zero in runafm_lapw calculations?
Is this a bug in runafm_lapw or something else?

[Symptoms]
- I am running wien2k_14.2 on a linux cluster with ifort_12.1.3 compiler plus MKL.
- All my calculations converged to -cc 0.0001 and -ec 0.00005. 
- The purpose of my calculations is to test runafm_lapw.
- During init_lapw, afminput was run successfully and a correct case.inclmcopy was generated.

Test 1:
- After runafm_lapw, all atomic magnetic moments meet the AFM requirement, e.g., 
        MMI001=3, MMI002=-3, MMI003=0...   So far, it seems perfect. 
- The problem is the existence of substantial interstitial magnetic moment:
       MMINT=-1,        ENE= -43408.62068 
- runsp_lapw *subsequently* converged to a metastable ferrimagnetic state:
       MMI001=1.9, MMI002=-2.3, MMINT=0.1,   ENE= -43407.99546

Test 2:
- runsp_lapw from A to Z (without runafm_lapw) converged to a nearly true AFM state with
       MMI001=1.43896, MMI002=-1.43882, MMINT=0.00005,    ENE=-43408.34858
  which is about 30meV per magnetic atom lower in energy than the nonmagnetic state (ENE=-43408.33109).

Test 3:
- Then, runafm_lapw starting from the nearly true AFM state obtained in Test 2, the results of Test 1 were reproduced.

Regards,
Weiguo Yin
Condensed Matter Theory Group
Condensed Matter Physics and Materials Science Department
Brookhaven National Laboratory
Upton, New York 11973-5000
USA