Convergence problem

[vasp16888]

Dear vasp users:
I am calculating Tb0.25Dy0.75Fe2 system using vasp 5.2, and lda+u is considerd. In unit cell, there are 2 Tb atoms, 6 Dy atoms, and 16 Fe atoms, lattice constant: 7.377 Ang, cubic, (LAVES phase)
INCAR:
*********************
ISTART = 0 # job : 0-new 1-cont 2-samecut
ICHARG = 2
ISPIN = 2 # spin : 1-off 2-on
# MAGMOM = 12 12 -3 -3 -3 -3.....
(no change whether MAGMOM initialize or not)

ENCUT = 300 # cut-off energy
PREC = H # precision : medium, high low
NELM = 1000 # maximal ELM
EDIFF = 1E-04 # stopping-criterion for ELM
NELMDL = -5 # No. of inconsistent ELM
NELMIN = 4 # minimal ELM

AMIX = 0.05
BMIX = 0.0001
AMIX_MAG = 0.8
BMIX_MAG = 0.0001
LMAXMIX = 6

LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = 3 3
LDAUU = 6.0 6.0 #(for Tb and Dy)
LDAUJ = 0 0
LDAUPRINT = 1
************************
KPOINTS:
5*5*5

The result doesn't converge???
OSZICAR:
The first 20 scfs:

DAV: 1 0.482546610083E+04 0.48255E+04 -0.94517E+04 31812 0.374E+03

DAV: 2 0.896071378862E+03 -0.39294E+04 -0.35973E+04 24096 0.117E+03

DAV: 3 -0.141388105742E+03 -0.10375E+04 -0.96876E+03 20916 0.330E+02

DAV: 4 -0.322098524228E+03 -0.18071E+03 -0.16195E+03 23028 0.129E+02

DAV: 5 -0.359166932409E+03 -0.37068E+02 -0.34708E+02 31956 0.616E+01 0.147E+02
RMM: 6 -0.795145723598E+03 -0.43598E+03 -0.10270E+03 33864 0.837E+01 0.214E+02
RMM: 7 -0.245682740467E+03 0.54946E+03 -0.12015E+02 29035 0.457E+01 0.113E+02
RMM: 8 -0.222456194250E+03 0.23227E+02 -0.18020E+01 30128 0.184E+01 0.117E+02
RMM: 9 -0.211148049710E+03 0.11308E+02 -0.12087E+01 31610 0.109E+01 0.129E+02
RMM: 10 -0.219350236732E+03 -0.82022E+01 -0.16733E+01 31534 0.136E+01 0.984E+01
RMM: 11 -0.242680319537E+03 -0.23330E+02 -0.22745E+01 29604 0.179E+01 0.141E+02
RMM: 12 -0.253112894650E+03 -0.10433E+02 -0.76826E+00 35005 0.677E+00 0.142E+02
RMM: 13 -0.224466007771E+03 0.28647E+02 -0.29077E+00 33087 0.488E+00 0.135E+02
RMM: 14 -0.239731272783E+03 -0.15265E+02 -0.28747E+00 33024 0.455E+00 0.104E+02
RMM: 15 -0.231114318661E+03 0.86170E+01 -0.71226E+00 31789 0.809E+00 0.880E+01
RMM: 16 -0.229578839731E+03 0.15355E+01 -0.29535E+00 35041 0.356E+00 0.864E+01
RMM: 17 -0.232134162122E+03 -0.25553E+01 -0.20685E+00 34531 0.337E+00 0.882E+01
RMM: 18 -0.231076510199E+03 0.10577E+01 -0.81249E-01 33378 0.199E+00 0.876E+01
RMM: 19 -0.233891332926E+03 -0.28148E+01 -0.72756E-01 32609 0.224E+00 0.880E+01
RMM: 20 -0.238041005032E+03 -0.41497E+01 -0.34743E+00 30094 0.663E+00 0.856E+01

...................
...................
...................

The last 20 scfs:

RMM: 981 -0.227565296625E+03 -0.78763E-01 -0.81066E+01 14800 0.100E+01 0.515E-01
RMM: 982 -0.227382047255E+03 0.18325E+00 -0.79462E+01 14631 0.100E+01 0.539E-01
RMM: 983 -0.227519603836E+03 -0.13756E+00 -0.79840E+01 15063 0.992E+00 0.550E-01
RMM: 984 -0.227217014276E+03 0.30259E+00 -0.77424E+01 13306 0.992E+00 0.535E-01
RMM: 985 -0.227181659132E+03 0.35355E-01 -0.74614E+01 13291 0.961E+00 0.510E-01
RMM: 986 -0.227457325163E+03 -0.27567E+00 -0.76565E+01 13935 0.962E+00 0.554E-01
RMM: 987 -0.227475273095E+03 -0.17948E-01 -0.79457E+01 12808 0.990E+00 0.553E-01
RMM: 988 -0.227552064003E+03 -0.76791E-01 -0.80525E+01 12783 0.101E+01 0.574E-01
RMM: 989 -0.227420294258E+03 0.13177E+00 -0.79714E+01 12212 0.101E+01 0.555E-01
RMM: 990 -0.227236136827E+03 0.18416E+00 -0.77060E+01 12128 0.999E+00 0.542E-01
RMM: 991 -0.227209528659E+03 0.26608E-01 -0.75198E+01 13059 0.977E+00 0.483E-01
RMM: 992 -0.227564114070E+03 -0.35459E+00 -0.78390E+01 13284 0.973E+00 0.518E-01
RMM: 993 -0.227450575979E+03 0.11354E+00 -0.80025E+01 12916 0.100E+01 0.510E-01
RMM: 994 -0.227603270012E+03 -0.15269E+00 -0.80702E+01 11941 0.100E+01 0.559E-01
RMM: 995 -0.227497375984E+03 0.10589E+00 -0.81952E+01 12110 0.103E+01 0.509E-01
RMM: 996 -0.227121267694E+03 0.37611E+00 -0.76807E+01 13857 0.987E+00 0.469E-01
RMM: 997 -0.227578949588E+03 -0.45768E+00 -0.77799E+01 12989 0.955E+00 0.508E-01
RMM: 998 -0.227112417728E+03 0.46653E+00 -0.76691E+01 13081 0.991E+00 0.472E-01
RMM: 999 -0.227314075543E+03 -0.20166E+00 -0.74836E+01 12136 0.957E+00 0.496E-01
RMM: *** -0.227468767738E+03 -0.15469E+00 -0.78626E+01 12039 0.984E+00

1 F= -.22746877E+03 E0= -.22725282E+03 d E =-.431891E+00 mag= 72.1859

QUESTIONS:
Is anyone who know why it doesn't converge ?
Any help will be greatly appreciated

[Danny]

If you check
LDA+U-manual
you'll read:

Code: Select all

NB: LDAUL, LDAUU, and LDAUJ must be specified for all atomic species! 

So I guess in your case you are one value short each time.
If the order in which you define your ions is:
first all Tb, then all Dy and finally all Fe you should probably write :

Code: Select all

LDAUL = 3 3 -1 
LDAUU = 6.0 6.0 0.0 
LDAUJ = 0 0 0

Next, try to run with spin turned off (if this converges nicely(i.e. in 100 steps or so), continue from this preconverged case with spin turned on)
Also try a different convergence scheme (eg DAV)

Cheers
Danny

[vasp16888]

[quote="http://cms.mpi.univie.ac.at/vasp/vasp/node161.html"]LDA+U-manual[/url]
you'll read:

Code: Select all

NB: LDAUL, LDAUU, and LDAUJ must be specified for all atomic species! 

So I guess in your case you are one value short each time.
If the order in which you define your ions is:
first all Tb, then all Dy and finally all Fe you should probably write :

Code: Select all

LDAUL = 3 3 -1 
LDAUU = 6.0 6.0 0.0 
LDAUJ = 0 0 0

Next, try to run with spin turned off (if this converges nicely(i.e. in 100 steps or so), continue from this preconverged case with spin turned on)
Also try a different convergence scheme (eg DAV)

Cheers
Danny[/quote](4 cases)[/size]

(1)add LDAUL, LDAUU, LDAUJ to Fe atom
==========================

ISPIN = 2
ALGO = FAST
LDAUL = 3 3 -1
LDAUU = 6.0 6.0 0.0
LDAUJ = 0 0 0

scf is up tp around 200 steps, but still doesn't converge.
the last 30 scf steps are as follows:
RMM: 189 -0.214292951189E+03 -0.11938E+01 -0.20173E+02 33654 0.286E+01 0.219E+01
RMM: 190 -0.213367567070E+03 0.92538E+00 -0.19172E+02 33227 0.269E+01 0.220E+01
RMM: 191 -0.215438631670E+03 -0.20711E+01 -0.21152E+02 32080 0.266E+01 0.241E+01
RMM: 192 -0.213643634349E+03 0.17950E+01 -0.20803E+02 31772 0.272E+01 0.246E+01
RMM: 193 -0.213011525213E+03 0.63211E+00 -0.21463E+02 31726 0.267E+01 0.230E+01
RMM: 194 -0.211495095854E+03 0.15164E+01 -0.17801E+02 30227 0.256E+01 0.238E+01
RMM: 195 -0.212243282663E+03 -0.74819E+00 -0.17615E+02 30479 0.252E+01 0.222E+01
RMM: 196 -0.210193307121E+03 0.20500E+01 -0.13225E+02 31771 0.248E+01 0.224E+01
RMM: 197 -0.211493272955E+03 -0.13000E+01 -0.15556E+02 31480 0.245E+01 0.208E+01
RMM: 198 -0.210747272050E+03 0.74600E+00 -0.12574E+02 31764 0.230E+01 0.232E+01
RMM: 199 -0.211600635322E+03 -0.85336E+00 -0.16440E+02 30685 0.239E+01 0.222E+01
RMM: 200 -0.210421391409E+03 0.11792E+01 -0.13941E+02 31904 0.231E+01 0.211E+01
RMM: 201 -0.212446121678E+03 -0.20247E+01 -0.14828E+02 34325 0.243E+01 0.232E+01
RMM: 202 -0.211927651302E+03 0.51847E+00 -0.16778E+02 31919 0.245E+01 0.228E+01
RMM: 203 -0.213946952606E+03 -0.20193E+01 -0.18629E+02 31729 0.251E+01 0.257E+01
RMM: 204 -0.213417951290E+03 0.52900E+00 -0.20464E+02 30434 0.255E+01 0.258E+01
RMM: 205 -0.214082477987E+03 -0.66453E+00 -0.22062E+02 30452 0.256E+01 0.268E+01
RMM: 206 -0.212577742179E+03 0.15047E+01 -0.19423E+02 33441 0.267E+01 0.277E+01
RMM: 207 -0.211642729381E+03 0.93501E+00 -0.20852E+02 31875 0.265E+01 0.247E+01
RMM: 208 -0.213383734518E+03 -0.17410E+01 -0.18254E+02 31143 0.261E+01 0.278E+01
RMM: 209 -0.211908669457E+03 0.14751E+01 -0.18360E+02 31364 0.261E+01 0.238E+01
RMM: 210 -0.211206314221E+03 0.70236E+00 -0.15020E+02 29258 0.253E+01 0.243E+01
RMM: 211 -0.210808238735E+03 0.39808E+00 -0.16252E+02 33012 0.280E+01 0.213E+01
RMM: 212 -0.210879500494E+03 -0.71262E-01 -0.12016E+02 31705 0.261E+01 0.211E+01
RMM: 213 -0.211716073757E+03 -0.83657E+00 -0.13502E+02 32235 0.265E+01 0.250E+01
RMM: 214 -0.214319786231E+03 -0.26037E+01 -0.18643E+02 30755 0.274E+01 0.267E+01
RMM: 215 -0.213486477104E+03 0.83331E+00 -0.19414E+02 30411 0.281E+01 0.289E+01
RMM: 216 -0.214707351530E+03 -0.12209E+01 -0.19385E+02 31655 0.290E+01 0.341E+01
RMM: 217 -0.215037870701E+03 -0.33052E+00 -0.21245E+02 31333 0.285E+01 0.326E+01
RMM: 218 -0.214464670829E+03 0.57320E+00 -0.18240E+02 29743 0.279E+01 0.342E+01

I don't think it will converge, but I still let it running.
===========================


(2)turn off the spin
======================
ISPIN = 1
ALGO =FAST
LDAUL = 3 3 -1
LDAUU = 6.0 6.0 0.0
LDAUJ = 0 0 0

but it still doesn't converge, see the last 30 scf steps below:
RMM: 465 -0.107229608170E+03 -0.33204E+01 0.73043E+01 11060 0.994E+00 0.718E+00
RMM: 466 -0.104364219152E+03 0.28654E+01 0.12055E+02 12326 0.933E+00 0.676E+00
RMM: 467 -0.108658828098E+03 -0.42946E+01 0.55903E+01 12208 0.929E+00 0.628E+00
RMM: 468 -0.952245762255E+02 0.13434E+02 0.22107E+02 12283 0.953E+00 0.620E+00
RMM: 469 -0.109650523038E+03 -0.14426E+02 0.26086E+01 11274 0.979E+00 0.647E+00
RMM: 470 -0.104482764968E+03 0.51678E+01 0.13427E+02 12260 0.833E+00 0.697E+00
RMM: 471 -0.105738449941E+03 -0.12557E+01 0.88956E+01 13403 0.918E+00 0.901E+00
RMM: 472 -0.102840756918E+03 0.28977E+01 0.12543E+02 13110 0.893E+00 0.901E+00
RMM: 473 -0.104958664343E+03 -0.21179E+01 0.87039E+01 11929 0.915E+00 0.897E+00
RMM: 474 -0.983847698652E+02 0.65739E+01 0.16825E+02 10874 0.920E+00 0.888E+00
RMM: 475 -0.101114857232E+03 -0.27301E+01 0.11015E+02 10737 0.945E+00 0.910E+00
RMM: 476 -0.101925877343E+03 -0.81102E+00 0.10906E+02 11264 0.873E+00 0.866E+00
RMM: 477 -0.101311123477E+03 0.61475E+00 0.12057E+02 9790 0.848E+00 0.865E+00
RMM: 478 -0.946081214028E+02 0.67030E+01 0.19949E+02 11647 0.884E+00 0.893E+00
RMM: 479 -0.102430553687E+03 -0.78224E+01 0.81818E+01 12814 0.102E+01 0.919E+00
RMM: 480 -0.967479200749E+02 0.56826E+01 0.17578E+02 13110 0.921E+00 0.876E+00
RMM: 481 -0.976599066907E+02 -0.91199E+00 0.14657E+02 12779 0.950E+00 0.880E+00
RMM: 482 -0.987869928907E+02 -0.11271E+01 0.13034E+02 13433 0.932E+00 0.880E+00
RMM: 483 -0.972181780036E+02 0.15688E+01 0.14771E+02 12813 0.965E+00 0.894E+00
RMM: 484 -0.980266688249E+02 -0.80849E+00 0.13879E+02 11586 0.941E+00 0.895E+00
RMM: 485 -0.101705001275E+03 -0.36783E+01 0.94751E+01 10716 0.962E+00 0.908E+00
RMM: 486 -0.102696891972E+03 -0.99189E+00 0.11699E+02 13051 0.901E+00 0.870E+00
RMM: 487 -0.101430213991E+03 0.12667E+01 0.12366E+02 12071 0.894E+00 0.940E+00
RMM: 488 -0.104802349465E+03 -0.33721E+01 0.73157E+01 11912 0.851E+00 0.972E+00
RMM: 489 -0.103635049496E+03 0.11673E+01 0.10621E+02 10985 0.792E+00 0.917E+00
RMM: 490 -0.995260905881E+02 0.41090E+01 0.14328E+02 12116 0.825E+00 0.891E+00
RMM: 491 -0.101349231804E+03 -0.18231E+01 0.10191E+02 12261 0.916E+00 0.102E+01
RMM: 492 -0.100126524540E+03 0.12227E+01 0.12377E+02 13573 0.946E+00 0.106E+01
RMM: 493 -0.102471244131E+03 -0.23447E+01 0.94060E+01 12905 0.112E+01 0.112E+01
RMM: 494 -0.100933239161E+03 0.15380E+01 0.11883E+02 12710 0.977E+00 0.113E+01

no convergence phenomenon, eigher. (still running)
==========================

(3)change algorithm to DAV, turn on the spin
======================
ISPIN = 2
ALGO = NORMAL (DAV)
LDAUL = 3 3 -1
LDAUU = 6.0 6.0 0.0
LDAUJ = 0 0 0

see the last 30 scf steps, doesn't converge.

DAV: 184 -0.182947960892E+03 -0.14747E-01 -0.22966E-01 30948 0.542E+00 0.191E+01
DAV: 185 -0.182732425884E+03 0.21554E+00 -0.20454E-01 29340 0.413E+00 0.195E+01
DAV: 186 -0.186809305233E+03 -0.40769E+01 -0.11442E+01 28800 0.330E+01 0.285E+01
DAV: 187 -0.186973319724E+03 -0.16401E+00 -0.14766E+00 25716 0.379E+00 0.289E+01
DAV: 188 -0.186941849658E+03 0.31470E-01 -0.34366E-01 27408 0.369E+00 0.286E+01
DAV: 189 -0.186113465248E+03 0.82838E+00 -0.11287E+00 28356 0.795E+00 0.254E+01
DAV: 190 -0.186245003698E+03 -0.13154E+00 -0.90097E-01 28044 0.661E+00 0.267E+01
DAV: 191 -0.189623413102E+03 -0.33784E+01 -0.27393E+02 26316 0.201E+02 0.573E+01
DAV: 192 -0.203925345907E+03 -0.14302E+02 -0.73674E+01 29844 0.698E+01 0.617E+01
DAV: 193 -0.186740506939E+03 0.17185E+02 -0.50272E+01 31884 0.466E+01 0.270E+01
DAV: 194 -0.184185302761E+03 0.25552E+01 -0.57439E+01 27768 0.115E+02 0.343E+01
DAV: 195 -0.188503434471E+03 -0.43181E+01 -0.31267E+01 29208 0.608E+01 0.508E+01
DAV: 196 -0.185114896212E+03 0.33885E+01 -0.31934E+01 30744 0.416E+01 0.266E+01
DAV: 197 -0.185661839800E+03 -0.54694E+00 -0.69161E+00 30468 0.174E+01 0.274E+01
DAV: 198 -0.186610017627E+03 -0.94818E+00 -0.19713E+00 30480 0.954E+00 0.258E+01
DAV: 199 -0.186465120378E+03 0.14490E+00 -0.37517E+00 29568 0.245E+01 0.201E+01
DAV: 200 -0.186537068652E+03 -0.71948E-01 -0.87148E+00 30216 0.327E+01 0.170E+01
DAV: 201 -0.185369765560E+03 0.11673E+01 -0.95617E+00 31428 0.276E+01 0.154E+01
DAV: 202 -0.185972532471E+03 -0.60277E+00 -0.17984E+01 29136 0.203E+01 0.166E+01
DAV: 203 -0.186292967102E+03 -0.32043E+00 -0.37522E+00 31308 0.858E+00 0.165E+01
DAV: 204 -0.186278534015E+03 0.14433E-01 -0.13530E+00 26280 0.953E+00 0.138E+01
DAV: 205 -0.186605196927E+03 -0.32666E+00 -0.48339E-01 27312 0.464E+00 0.125E+01
DAV: 206 -0.185977579137E+03 0.62762E+00 -0.40797E+00 28224 0.203E+01 0.207E+01
DAV: 207 -0.187813737649E+03 -0.18362E+01 -0.16142E+01 29352 0.210E+01 0.198E+01
DAV: 208 -0.189101883042E+03 -0.12881E+01 -0.45933E+00 30912 0.187E+01 0.214E+01
DAV: 209 -0.188902513785E+03 0.19937E+00 -0.10020E+00 32112 0.104E+01 0.208E+01
DAV: 210 -0.189173157289E+03 -0.27064E+00 -0.34252E-01 28140 0.365E+00 0.208E+01
DAV: 211 -0.189009824037E+03 0.16333E+00 -0.20757E+00 24984 0.234E+01 0.164E+01
DAV: 212 -0.189229516091E+03 -0.21969E+00 -0.32002E-01 27840 0.545E+00 0.170E+01
DAV: 213 -0.189305556279E+03 -0.76040E-01 -0.14984E-01 28140 0.196E+00 0.177E+01

========================


(4)use DAV algorithm, turn off the spin
===========================
ISPIN = 1
ALGO = NORMAL (DAV)
LDAUL = 3 3 -1
LDAUU = 6.0 6.0 0.0
LDAUJ = 0 0 0

it converge after around 300-400 scf steps.

DAV: 319 -0.121131021886E+03 -0.32972E-02 -0.35109E-03 9264 0.342E-01 0.100E+00
DAV: 320 -0.121123454447E+03 0.75674E-02 -0.13210E-02 8904 0.820E-01 0.102E+00
DAV: 321 -0.121102488083E+03 0.20966E-01 -0.38973E-02 11196 0.163E+00 0.534E-01
DAV: 322 -0.121100791788E+03 0.16963E-02 -0.10411E-02 9780 0.839E-01 0.615E-01
DAV: 323 -0.121101790740E+03 -0.99895E-03 -0.33440E-03 10500 0.359E-01 0.602E-01
DAV: 324 -0.121102275498E+03 -0.48476E-03 -0.18648E-03 10944 0.227E-01 0.571E-01
DAV: 325 -0.121088726353E+03 0.13549E-01 -0.81134E-03 10284 0.779E-01 0.550E-01
DAV: 326 -0.121094790009E+03 -0.60637E-02 -0.44720E-03 10968 0.650E-01 0.543E-01
DAV: 327 -0.121094934275E+03 -0.14427E-03 -0.11975E-03 8892 0.134E-01 0.544E-01
DAV: 328 -0.121090696954E+03 0.42373E-02 -0.52126E-04 6912 0.196E-01 0.546E-01
DAV: 329 -0.121065908646E+03 0.24788E-01 -0.18149E-02 9540 0.954E-01 0.585E-01
DAV: 330 -0.121058512319E+03 0.73963E-02 -0.12459E-02 11004 0.496E-01 0.598E-01
DAV: 331 -0.121043849412E+03 0.14663E-01 -0.24126E-02 11076 0.773E-01 0.616E-01
DAV: 332 -0.121062085067E+03 -0.18236E-01 -0.18586E-02 8928 0.103E+00 0.717E-01
DAV: 333 -0.121072141605E+03 -0.10057E-01 -0.41428E-03 8304 0.384E-01 0.720E-01
DAV: 334 -0.121080617419E+03 -0.84758E-02 -0.34708E-03 8604 0.314E-01 0.737E-01
DAV: 335 -0.121077893202E+03 0.27242E-02 -0.12830E-03 7476 0.208E-01 0.721E-01
DAV: 336 -0.121087267368E+03 -0.93742E-02 -0.18817E-02 9324 0.116E+00 0.534E-01
DAV: 337 -0.121107519340E+03 -0.20252E-01 -0.14834E-02 10020 0.946E-01 0.531E-01
DAV: 338 -0.121091487495E+03 0.16032E-01 -0.11839E-02 8388 0.712E-01 0.670E-01
DAV: 339 -0.121093289080E+03 -0.18016E-02 -0.20808E-02 9228 0.122E+00 0.665E-01
DAV: 340 -0.121108387294E+03 -0.15098E-01 -0.46584E-02 8388 0.149E+00 0.726E-01
DAV: 341 -0.121107918919E+03 0.46837E-03 -0.15987E-02 13548 0.600E-01 0.569E-01
DAV: 342 -0.121090460742E+03 0.17458E-01 -0.27643E-02 9444 0.113E+00 0.851E-01
DAV: 343 -0.121080365301E+03 0.10095E-01 -0.12648E-02 9132 0.828E-01 0.898E-01
DAV: 344 -0.121083347791E+03 -0.29825E-02 -0.20141E-02 9036 0.920E-01 0.933E-01
DAV: 345 -0.121078268663E+03 0.50791E-02 -0.65447E-03 8736 0.515E-01 0.877E-01
DAV: 346 -0.121100528072E+03 -0.22259E-01 -0.39372E-02 8964 0.129E+00 0.101E+00
DAV: 347 -0.121101359252E+03 -0.83118E-03 -0.32260E-03 12132 0.270E-01 0.101E+00
DAV: 348 -0.121101396130E+03 -0.36878E-04 -0.45068E-04 8088 0.841E-02

1 F= -.12110140E+03 E0= -.12047904E+03 d E =-.124472E+01

now I set:
ISTART = 1
ICHARG = 1
ISPIN = 2
to recalculate with the spin on.

=========================


QUESTIONS:
(1)Why there was such big change in energy when turn on or off the spin ? (turn on: -180 eV , turn off, -120 eV) Is this reasonable ?
(2)I think RMM is accurate than DAV, maybe slower than DAV, but why DAV with spin off converge, RMM with spin off dones't converge???
(3)Any further suggestions was appreciated.

Looking forward to everybody's suggestion
<span class='smallblacktext'>[ Edited Sat Apr 24 2010, 05:06AM ]</span>

[vasp16888]

Please help me out, many many thanks:)

[Danny]

there is something very wrong with those calculations...I have never seen such a bad convergence behavior. In general a SC finnishes within 60 steps, and even in badly behaved ones you will see convergence.

Maybe it's an idea to start your calculation from scratch, make a new INCAR and ONLY put in the parameters that you need, strip down everything:
1) USE a simple LDA functional
2) K-points 4x4x4 / 5x5x5 -> check if this influences
3) INCAR:
ISTART = 0 # job : 0-new 1-cont 2-samecut
ICHARG = 2
ISPIN = 1 # spin : 1-off 2-on

ENCUT = 300 # cut-off energy
PREC = High # precision : medium, high low
EDIFF = 1E-07 # stopping-criterion for ELM
IBRION = -1
NSW = 0
---------------------

4) If this works, add +U
5) then add spin(spin tends to make some convergence harder but not impossible

As long as you need more than 100 steps for a convergence something is wrong, and you will need to search the manual for ways to solve it I'm affraid.

Cheers
Danny

[vasp16888]

[quote author= 0 # job : 0-new 1-cont 2-samecut
ICHARG = 2
ISPIN = 1 # spin : 1-off 2-on

ENCUT = 300 # cut-off energy
PREC = High # precision : medium, high low
EDIFF = 1E-07 # stopping-criterion for ELM
IBRION = -1
NSW = 0
---------------------

4) If this works, add +U
5) then add spin(spin tends to make some convergence harder but not impossible

As long as you need more than 100 steps for a convergence something is wrong, and you will need to search the manual for ways to solve it I'm affraid.

Cheers
Danny[/quote]DAV algorithm.[/size]
1. turn off the spin and lda+u
------------------------------------
DAV: 50 -0.185694451906E+03 -0.37281E-05 -0.44089E-05 13896 0.434E-02 0.184E-01
DAV: 51 -0.185694463822E+03 -0.11916E-04 -0.44626E-05 13752 0.607E-02 0.135E-01
DAV: 52 -0.185694467829E+03 -0.40077E-05 -0.17499E-05 13896 0.329E-02 0.589E-02
DAV: 53 -0.185694482254E+03 -0.14425E-04 -0.95543E-06 13164 0.228E-02 0.115E-01
DAV: 54 -0.185694486937E+03 -0.46834E-05 -0.69149E-06 13344 0.150E-02 0.595E-02
DAV: 55 -0.185694496399E+03 -0.94618E-05 -0.65204E-06 12276 0.554E-03 0.505E-02
DAV: 56 -0.185694507735E+03 -0.11336E-04 -0.13371E-05 10104 0.126E-02 0.632E-02
DAV: 57 -0.185694512222E+03 -0.44868E-05 -0.73196E-06 10260 0.124E-02 0.814E-02
DAV: 58 -0.185694512937E+03 -0.71579E-06 -0.32615E-06 11772 0.746E-03 0.854E-02
DAV: 59 -0.185694514151E+03 -0.12139E-05 -0.86088E-07 9744 0.306E-03 0.936E-02
DAV: 60 -0.185694513286E+03 0.86555E-06 -0.15542E-06 7884 0.748E-03 0.526E-02
DAV: 61 -0.185694513067E+03 0.21877E-06 -0.17363E-06 7770 0.971E-03 0.206E-02
DAV: 62 -0.185694513413E+03 -0.34578E-06 -0.92840E-07 9744 0.274E-03 0.182E-02
DAV: 63 -0.185694513461E+03 -0.48331E-07 -0.52267E-07 6978 0.305E-03

1 F= -.18569451E+03 E0= -.18497825E+03 d E =-.143253E+01
------------------------------------

2. then set ISTART=1, ICHARG=1 in INCAR,
and turn on the lda+u
AMIX = 0.05
BMIX = 0.0001
AMIX_MAG = 0.8
BMIX_MAG = 0.0001
LMAXMIX = 6

# LDA+U
LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = 3 3 -1
LDAUU = 6.0 6.0 0
LDAUJ = 0 0 0
LDAUPRINT = 1
LMAXMIX = 6 4

the result:
------------------------------------
DAV: 544 -0.160817685917E+03 -0.74998E-06 -0.13719E-05 10572 0.348E-02 0.141E-02
DAV: 545 -0.160817684533E+03 0.13842E-05 -0.82489E-06 13122 0.243E-02 0.122E-02
DAV: 546 -0.160817685685E+03 -0.11524E-05 -0.44696E-06 9138 0.169E-02 0.117E-02
DAV: 547 -0.160817685938E+03 -0.25319E-06 -0.41117E-06 7902 0.210E-02 0.122E-02
DAV: 548 -0.160817687220E+03 -0.12822E-05 -0.82482E-06 9150 0.222E-02 0.119E-02
DAV: 549 -0.160817686650E+03 0.57083E-06 -0.13919E-06 8580 0.127E-02 0.123E-02
DAV: 550 -0.160817681641E+03 0.50090E-05 -0.86816E-06 9090 0.410E-02 0.131E-02
DAV: 551 -0.160817682452E+03 -0.81173E-06 -0.34576E-06 11388 0.103E-02 0.121E-02
DAV: 552 -0.160817682247E+03 0.20576E-06 -0.11117E-06 11100 0.761E-03 0.114E-02
DAV: 553 -0.160817680275E+03 0.19716E-05 -0.16405E-06 9090 0.152E-02 0.117E-02
DAV: 554 -0.160817675114E+03 0.51612E-05 -0.64165E-06 11298 0.310E-02 0.947E-03
DAV: 555 -0.160817677053E+03 -0.19393E-05 -0.37191E-06 10158 0.195E-02 0.963E-03
DAV: 556 -0.160817671688E+03 0.53654E-05 -0.16388E-05 11700 0.283E-02 0.655E-03
DAV: 557 -0.160817671955E+03 -0.26708E-06 -0.59994E-06 11082 0.104E-02 0.752E-03
DAV: 558 -0.160817671223E+03 0.73209E-06 -0.95937E-07 10764 0.702E-03 0.685E-03
DAV: 559 -0.160817673559E+03 -0.23360E-05 -0.29024E-06 7548 0.168E-02 0.717E-03
DAV: 560 -0.160817673626E+03 -0.66935E-07 -0.66004E-07 8760 0.324E-03

1 F= -.16081767E+03 E0= -.16054931E+03 d E =-.536725E+00
------------------------------------

3. last I turn on spin

ISPIN = 2
ISTART = 1
ICHARG = 1

the result:
------------------------------------
DAV: 200 -0.188198714724E+03 0.18753E+00 -0.26728E+00 20724 0.155E+01 0.719E+00
DAV: 201 -0.187827359911E+03 0.37135E+00 -0.25293E+00 28326 0.130E+01 0.616E+00
DAV: 202 -0.187906743389E+03 -0.79383E-01 -0.70826E-01 23646 0.284E+00 0.619E+00
DAV: 203 -0.187909291825E+03 -0.25484E-02 -0.47845E-02 34968 0.989E-01 0.626E+00
DAV: 204 -0.187910582727E+03 -0.12909E-02 -0.43057E-03 26976 0.581E-01 0.628E+00
DAV: 205 -0.187913075862E+03 -0.24931E-02 -0.16924E-02 26172 0.161E+00 0.611E+00
DAV: 206 -0.187956625542E+03 -0.43550E-01 -0.10089E+00 20982 0.112E+01 0.585E+00
DAV: 207 -0.188045446149E+03 -0.88821E-01 -0.10521E+00 26184 0.119E+01 0.641E+00
DAV: 208 -0.188072627147E+03 -0.27181E-01 -0.30363E-01 28998 0.224E+00 0.644E+00
DAV: 209 -0.188060555440E+03 0.12072E-01 -0.60693E-02 29550 0.163E+00 0.647E+00
DAV: 210 -0.188004154671E+03 0.56401E-01 -0.89031E-01 24420 0.945E+00 0.657E+00
DAV: 211 -0.188137951079E+03 -0.13380E+00 -0.32826E+00 24444 0.166E+01 0.704E+00
DAV: 212 -0.188269941030E+03 -0.13199E+00 -0.14608E+00 27888 0.118E+01 0.652E+00
DAV: 213 -0.188411780886E+03 -0.14184E+00 -0.75561E+00 25920 0.234E+01 0.650E+00
DAV: 214 -0.188561629741E+03 -0.14985E+00 -0.15685E+00 28104 0.592E+00 0.611E+00
DAV: 215 -0.188460007256E+03 0.10162E+00 -0.69528E-01 23844 0.108E+01 0.619E+00
DAV: 216 -0.188672400784E+03 -0.21239E+00 -0.20680E+00 22356 0.277E+01 0.702E+00
DAV: 217 -0.188650077988E+03 0.22323E-01 -0.36205E-01 27432 0.657E+00 0.690E+00
DAV: 218 -0.188617413767E+03 0.32664E-01 -0.73801E-01 24258 0.622E+00 0.708E+00
DAV: 219 -0.188593716461E+03 0.23697E-01 -0.25594E-01 30624 0.301E+00 0.726E+00
DAV: 220 -0.188548775231E+03 0.44941E-01 -0.90111E-01 24102 0.130E+01 0.690E+00
DAV: 221 -0.188321742548E+03 0.22703E+00 -0.36258E+00 26940 0.193E+01 0.609E+00

even around 200 scf steps, no convergence phenomenon. Let alone the 100 scf steps you suggested.
------------------------------------

If I change to use RMM algorithm, others are all the same as above, it didn't converge when turn on lda+u, let alone turn on the spin.

Honestly speaking, this Terfenol system has confused me for some time, I still didn't find a proper way to deal with it.

Still, any suggestion will be appreciated. I will adopt every suggestion in order to solve this problem.

Anyway, thanks, Danny
Looking forward to your reply.

[Danny]

hmm, is there any litterature of anyone actually succeeding in this material?

At this moment I have run out of suggestions...maybe one last really desperate one:
Are you sure VASP is working correctly? Do you have some benchmarks to check this? The reason I'm saying this is because you have two convergence problems here on the forum, and they seem equally bad...:s.

[Danny]

Some tests:
Test 1: diamond
INCAR

System = diamond

ISMEAR = -5

KPOINTS

4x4x4
0
Monkhorst
4 4 4
0 0 0

script

for a in 3.3 3.4 3.5 3.6 3.7
do
echo "a= $a"

cat >POSCAR <<!
cubic diamond
$a
0.5 0.5 0.0
0.0 0.5 0.5
0.5 0.0 0.5
2
direct
0.0 0.0 0.0
0.25 0.25 0.25
!

vasp


E=`tail -1 OSZICAR`
echo $a $E >>SUMMARY

done

Use the US C potential--> Results:

3.3 1 F= -.19639551E+02 E0= -.19639551E+02 d E =0.000000E+00
3.4 1 F= -.20138423E+02 E0= -.20138423E+02 d E =0.000000E+00
3.5 1 F= -.20331602E+02 E0= -.20331602E+02 d E =0.000000E+00
3.6 1 F= -.20277194E+02 E0= -.20277194E+02 d E =0.000000E+00
3.7 1 F= -.20028157E+02 E0= -.20028157E+02 d E =0.000000E+00

Test 2: diamond
INCAR

SYSTEM=Ni surface
ENMAX=400
ISMEAR=-5
ALGO=V
IDIPOL=3
LDIPOL=.TRUE.
LVTOT=.TRUE.

KPOINTS

4x4x4
0
Monkhorst
9 9 1
0 0 0

POSCAR

fcc Ni surface
3.53
0.70710678 0.0 0.0
-0.35355339 0.6123724 0.0
0.0 0.0 5.1961524
5 1 1
Selective Dynamics
direct
0.0 0.0 0.0 F F F
0.33333333 0.66666667 0.11111111 F F F
0.66666667 0.33333333 0.22222222 F F F
0.0 0.0 0.33333333 T T T
0.33333333 0.66666667 0.44444444 T T T
0.33333333 0.66666667 0.54029062 T T T
0.33333333 0.66666667 0.60298866 T T T

Use the PAW_GGA Ni potential--> Results:

very simple: the job should converge

The jobs take less than 1 minute each, and should be tested serial and parallel.
If you VASP fails at any of the above tests, you should consider redoing the compilation untill it succeeds in surviving those tests.

Danny

[vasp16888]

[quote="Danny"]hmm, is there any litterature of anyone actually succeeding in this material?

At this moment I have run out of suggestions...maybe one last really desperate one:
Are you sure VASP is working correctly? Do you have some benchmarks to check this? The reason I'm saying this is because you have two convergence problems here on the forum, and they seem equally bad...:s.[/quote]


Yeah, about Fe12Ga4 system, there are literatures which has reported the calculation result by vasp, but no calculation details.
about Terfenol, so far as i know, there are only some result calculated by wien2k in the literature which I am calculating now, but wien2k is very slow, still need some time.
For strong correlated system, vasp has done a lot of successful system in the literature, I firmly believe vasp can deal with it.
Actually, I don't have many experience on doing strong correlated system by vasp, here, the most important thing for me is looking for some suggestions from vasp expert on dealing with strong correlated system by vasp.
Thanks:)

[vasp16888]

Some tests: [/size]
Test 1: diamond
INCAR

System = diamond

ISMEAR = -5

KPOINTS

4x4x4
0
Monkhorst
4 4 4
0 0 0

script

for a in 3.3 3.4 3.5 3.6 3.7
do
echo "a= $a"

cat >POSCAR <<!
cubic diamond
$a
0.5 0.5 0.0
0.0 0.5 0.5
0.5 0.0 0.5
2
direct
0.0 0.0 0.0
0.25 0.25 0.25
!

vasp


E=`tail -1 OSZICAR`
echo $a $E >>SUMMARY

done

Use the US C potential--> Results:

3.3 1 F= -.19639551E+02 E0= -.19639551E+02 d E =0.000000E+00
3.4 1 F= -.20138423E+02 E0= -.20138423E+02 d E =0.000000E+00
3.5 1 F= -.20331602E+02 E0= -.20331602E+02 d E =0.000000E+00
3.6 1 F= -.20277194E+02 E0= -.20277194E+02 d E =0.000000E+00
3.7 1 F= -.20028157E+02 E0= -.20028157E+02 d E =0.000000E+00

Test 2: diamond
INCAR

SYSTEM=Ni surface
ENMAX=400
ISMEAR=-5
ALGO=V
IDIPOL=3
LDIPOL=.TRUE.
LVTOT=.TRUE.

KPOINTS

4x4x4
0
Monkhorst
9 9 1
0 0 0

POSCAR

fcc Ni surface
3.53
0.70710678 0.0 0.0
-0.35355339 0.6123724 0.0
0.0 0.0 5.1961524
5 1 1
Selective Dynamics
direct
0.0 0.0 0.0 F F F
0.33333333 0.66666667 0.11111111 F F F
0.66666667 0.33333333 0.22222222 F F F
0.0 0.0 0.33333333 T T T
0.33333333 0.66666667 0.44444444 T T T
0.33333333 0.66666667 0.54029062 T T T
0.33333333 0.66666667 0.60298866 T T T

Use the PAW_GGA Ni potential--> Results:

very simple: the job should converge

The jobs take less than 1 minute each, and should be tested serial and parallel.
If you VASP fails at any of the above tests, you should consider redoing the compilation untill it succeeds in surviving those tests.

Danny

Diamond result:[/b](serial and parallel are the same)

3.7 1 F= -.20016897E+02 E0= -.20016897E+02 d E =0.000000E+00
3.6 1 F= -.20261989E+02 E0= -.20261989E+02 d E =0.000000E+00
3.5 1 F= -.20313363E+02 E0= -.20313363E+02 d E =0.000000E+00
3.4 1 F= -.20111723E+02 E0= -.20111723E+02 d E =0.000000E+00
3.3 1 F= -.19601621E+02 E0= -.19601621E+02 d E =0.000000E+00

fcc_Ni_surface
you wrote 5 1 1 in POSCAR which means there should be 3 different atoms in this systme, so please tell me the other two atoms. Thanks:)

[vasp16888]

Now, I am pretty sure that there is nothing wrong with the compilation of vasp, because I have successfully calculated a lot of systems.
Thanks anyway.

[Danny]

My bad, the second type is GGA C, and the third GGA O.

OK, that is good to know...so we can trust VASP to do what it is supposed to do...I was just worried, because you presented two problematic systems with possibly the same problem

[vasp16888]

[quote="Danny"]My bad, the second type is GGA C, and the third GGA O.

OK, that is good to know...so we can trust VASP to do what it is supposed to do...I was just worried, because you presented two problematic systems with possibly the same problem[/quote]

Hi Danny, about second system, it converges quickly in around 1 minutes, here are the result:

RMM: 31 -0.412748706956E+02 -0.13774E-03 -0.19911E-04 570 0.102E-01 0.858E-01
RMM: 32 -0.412744723371E+02 0.39836E-03 -0.30877E-04 614 0.126E-01 0.342E-01
RMM: 33 -0.412734602887E+02 0.10120E-02 -0.21943E-04 562 0.973E-02 0.937E-02
RMM: 34 -0.412739987900E+02 -0.53850E-03 -0.10667E-04 512 0.808E-02 0.180E-01
RMM: 35 -0.412734727670E+02 0.52602E-03 -0.56557E-05 506 0.568E-02 0.204E-01
RMM: 36 -0.412741219548E+02 -0.64919E-03 -0.40661E-05 505 0.490E-02 0.184E-01
RMM: 37 -0.412735979534E+02 0.52400E-03 -0.20957E-05 474 0.348E-02 0.102E-01
RMM: 38 -0.412735665362E+02 0.31417E-04 -0.10234E-05 439 0.249E-02

1 F= -.41273567E+02 E0= -.41273567E+02 d E =0.000000E+00

100 percent sure there is nothing wrong with the compilation.
So about the original question above, i am looking forward to your reply ? Thanks in advance:)

[Danny]

well, at this point I'm also running out of ideas...sorry

[alex]

Post the POSCAR, please.
And do a
grep TIT POTCAR

Thx

alex