Does anyone have success in calculating a large surface cell? If you do, can you please share your input files?
I have trouble in obtaining the SCF convergence with a metal oxide (roughly 10 x10 x25 angstrom) surface even I am using a norm-conserving pseudo-potential.
I am kind of frustrated with ABINIT now. Those calculations are supposed to be a very routine and straightforward calculation with commercial software like VASP.
Large metal oxide surface cell SCF convergence
Moderator: bguster
Re: Large metal oxide surface cell SCF convergence
Dear Dyu,
The convergence might depend on the system you are running on. You said to run a metal oxide, do you really mean a metallic system or a transition-metal oxide?
How bad is your convergence? Does it succeed to converge up to a given precision and then it is stuck at this precision or is it completely bad from the beginning or does it converge and then starts to diverge, do you have an error/wraning message, etc?
The first input variables you can play with are diemix and diemac, did you try to modify them away from default values (reducing diemix for example)? You can also test to increase the nline flag (10 or 12 or more depending how it goes).
Something that might help for your case is to change the preconditioning, see iprcel input flag. I got some large and inhomogeneous systems (insulators with small gaps) going from not converging or converging within several hundred of steps to converging within 10 to 20 steps by tuning the iprcel flag.
The default values in Abinit might not be the best ones for your systems.
Best wishes,
Eric
The convergence might depend on the system you are running on. You said to run a metal oxide, do you really mean a metallic system or a transition-metal oxide?
How bad is your convergence? Does it succeed to converge up to a given precision and then it is stuck at this precision or is it completely bad from the beginning or does it converge and then starts to diverge, do you have an error/wraning message, etc?
The first input variables you can play with are diemix and diemac, did you try to modify them away from default values (reducing diemix for example)? You can also test to increase the nline flag (10 or 12 or more depending how it goes).
Something that might help for your case is to change the preconditioning, see iprcel input flag. I got some large and inhomogeneous systems (insulators with small gaps) going from not converging or converging within several hundred of steps to converging within 10 to 20 steps by tuning the iprcel flag.
The default values in Abinit might not be the best ones for your systems.
Best wishes,
Eric
Re: Large metal oxide surface cell SCF convergence
Eric,
Thanks for your reply. What I tried to study is CuFeS2 surface.
Below is my input file. The output file is also attached. The log file is too large to attach.
I tried your suggestions. But so far, none has worked.
Anyhow, I appreciate it your response.
Best Regards,
DYU
####################################################
# Atoms Stucture created by CellToAbinit Converter !
####################################################
# Definition of the unit cell
#The length of the primitive vectors
# 1 Bohr=0.5291772108 Angstroms
# acell 10.497600 10.497600 25.000000 Angstroms
acell 19.837589 19.837589 47.243153
rprim
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000
#Definition of the atom types
# 10 kind of atoms
# ['H', 'C', 'O']
ntypat 3
znucl 29 26 16
#Definition of tha atoms
natom 64
typat 16*1 16*2 32*3
xred
0.0009752852724159 0.0009706455887686 0.3001052552236945
0.2510431823076404 0.2510627587651560 0.1042064982470400
0.2508868797775855 0.0009716751043642 0.1987684168066323
0.0009382144490147 0.2509448826398426 0.0000000000000000
0.5009752852724160 0.0009706455887686 0.3001052552236945
0.7510431823076408 0.2510627587651560 0.1042064982470400
0.7508868797775856 0.0009716751043642 0.1987684168066323
0.5009382144490147 0.2509448826398426 0.0000000000000000
0.0009752852724159 0.5009706455887686 0.3001052552236945
0.2510431823076404 0.7510627587651559 0.1042064982470400
0.2508868797775855 0.5009716751043642 0.1987684168066323
0.0009382144490147 0.7509448826398426 0.0000000000000000
0.5009752852724160 0.5009706455887686 0.3001052552236945
0.7510431823076408 0.7510627587651559 0.1042064982470400
0.7508868797775856 0.5009716751043642 0.1987684168066323
0.5009382144490147 0.7509448826398426 0.0000000000000000
0.0009380269960699 0.0009357052108406 0.0979545194752402
0.2509275772737494 0.2509063243995641 0.2963638269775226
0.2509573616826728 0.0009958466697171 0.0000039199999975
0.0009321972921387 0.2509131989197733 0.1966289128943849
0.5009380269960702 0.0009357052108406 0.0979545194752402
0.7509275772737495 0.2509063243995641 0.2963638269775226
0.7509573616826728 0.0009958466697171 0.0000039199999975
0.5009321972921389 0.2509131989197733 0.1966289128943849
0.0009380269960699 0.5009357052108405 0.0979545194752402
0.2509275772737494 0.7509063243995636 0.2963638269775226
0.2509573616826728 0.5009958466697171 0.0000039199999975
0.0009321972921387 0.7509131989197729 0.1966289128943849
0.5009380269960702 0.5009357052108405 0.0979545194752402
0.7509275772737495 0.7509063243995636 0.2963638269775226
0.7509573616826728 0.5009958466697171 0.0000039199999975
0.5009321972921389 0.7509131989197729 0.1966289128943849
0.1817200187612814 0.0884933525842759 0.3393552265550508
0.3201932735921075 0.4132977138357516 0.3393339708594195
0.1116742540284368 0.1211365491383456 0.1500537542312304
0.1214970073454634 0.3577896450660130 0.2479035614919708
0.3759627914951977 0.1206330018289918 0.0519043199999984
0.3803481412579053 0.1440096184562397 0.2479083626831271
0.1259647919524483 0.3812931527206231 0.0518892400000013
0.3902508329501212 0.3808357296738636 0.1501043590418803
0.6817199235014176 0.0884933525842759 0.3393552265550508
0.8201933688519789 0.4132977138357516 0.3393339708594195
0.6116741587685659 0.1211365491383456 0.1500537542312304
0.6214971026053346 0.3577896450660130 0.2479035614919708
0.8759627914951977 0.1206330018289918 0.0519043199999984
0.8803481412579052 0.1440096184562397 0.2479083626831271
0.6259647919524483 0.3812931527206231 0.0518892400000013
0.8902507376902499 0.3808357296738636 0.1501043590418803
0.1817200187612814 0.5884933525842763 0.3393552265550508
0.3201932735921075 0.9132977138357511 0.3393339708594195
0.1116742540284368 0.6211365491383459 0.1500537542312304
0.1214970073454634 0.8577895498061420 0.2479035614919708
0.3759627914951977 0.6206330018289918 0.0519043199999984
0.3803481412579053 0.6440096184562397 0.2479083626831271
0.1259647919524483 0.8812931527206231 0.0518892400000013
0.3902508329501212 0.8808358249337349 0.1501043590418803
0.6817199235014176 0.5884933525842763 0.3393552265550508
0.8201933688519789 0.9132977138357511 0.3393339708594195
0.6116741587685659 0.6211365491383459 0.1500537542312304
0.6214971026053346 0.8577895498061420 0.2479035614919708
0.8759627914951977 0.6206330018289918 0.0519043199999984
0.8803481412579052 0.6440096184562397 0.2479083626831271
0.6259647919524483 0.8812931527206231 0.0518892400000013
0.8902507376902499 0.8808358249337349 0.1501043590418803
#fix atoms
natfix 16
iatfix 4 8 12 16 19 23 27 31 37 39 45 47 53 55 61 63
#Basis definitions
ixc 11
ecut 20
ecutsm 0.5
# pawecutdg 30.
#geometry optimziation
ionmov 2 # Use the modified Broyden algorithm
ntime 100 # Maximum number of Broyden "timesteps"
tolmxf 5.0d-5 # Stopping criterion for the geometry optimization : when
# the residual forces are less than tolmxf, the Broyden
# # algorithm can stop
#Definition of occupation numbers
occopt 3
tsmear 0.01
#
#SCF cycle parameters
tolvrs 1.0d-10
nstep 200
#K-points and sym
kptopt 0 # Enter the k points manually
nkpt 1 # one k point
kpt 0 0 0
#dielectric
#iprcel 45
#pawsushat 1
#diemac 12.0
diemix 0.1
#nsppol 2
#spinmagntarget 0
charge 0
#pawovlp 15.0
iscf 12
usepotzero 2
#parallization parameters
autoparal 1
#I/O parameters
prtwf 1 prtden 0 prteig 0
getwfk -1
Thanks for your reply. What I tried to study is CuFeS2 surface.
Below is my input file. The output file is also attached. The log file is too large to attach.
I tried your suggestions. But so far, none has worked.
Anyhow, I appreciate it your response.
Best Regards,
DYU
####################################################
# Atoms Stucture created by CellToAbinit Converter !
####################################################
# Definition of the unit cell
#The length of the primitive vectors
# 1 Bohr=0.5291772108 Angstroms
# acell 10.497600 10.497600 25.000000 Angstroms
acell 19.837589 19.837589 47.243153
rprim
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000
#Definition of the atom types
# 10 kind of atoms
# ['H', 'C', 'O']
ntypat 3
znucl 29 26 16
#Definition of tha atoms
natom 64
typat 16*1 16*2 32*3
xred
0.0009752852724159 0.0009706455887686 0.3001052552236945
0.2510431823076404 0.2510627587651560 0.1042064982470400
0.2508868797775855 0.0009716751043642 0.1987684168066323
0.0009382144490147 0.2509448826398426 0.0000000000000000
0.5009752852724160 0.0009706455887686 0.3001052552236945
0.7510431823076408 0.2510627587651560 0.1042064982470400
0.7508868797775856 0.0009716751043642 0.1987684168066323
0.5009382144490147 0.2509448826398426 0.0000000000000000
0.0009752852724159 0.5009706455887686 0.3001052552236945
0.2510431823076404 0.7510627587651559 0.1042064982470400
0.2508868797775855 0.5009716751043642 0.1987684168066323
0.0009382144490147 0.7509448826398426 0.0000000000000000
0.5009752852724160 0.5009706455887686 0.3001052552236945
0.7510431823076408 0.7510627587651559 0.1042064982470400
0.7508868797775856 0.5009716751043642 0.1987684168066323
0.5009382144490147 0.7509448826398426 0.0000000000000000
0.0009380269960699 0.0009357052108406 0.0979545194752402
0.2509275772737494 0.2509063243995641 0.2963638269775226
0.2509573616826728 0.0009958466697171 0.0000039199999975
0.0009321972921387 0.2509131989197733 0.1966289128943849
0.5009380269960702 0.0009357052108406 0.0979545194752402
0.7509275772737495 0.2509063243995641 0.2963638269775226
0.7509573616826728 0.0009958466697171 0.0000039199999975
0.5009321972921389 0.2509131989197733 0.1966289128943849
0.0009380269960699 0.5009357052108405 0.0979545194752402
0.2509275772737494 0.7509063243995636 0.2963638269775226
0.2509573616826728 0.5009958466697171 0.0000039199999975
0.0009321972921387 0.7509131989197729 0.1966289128943849
0.5009380269960702 0.5009357052108405 0.0979545194752402
0.7509275772737495 0.7509063243995636 0.2963638269775226
0.7509573616826728 0.5009958466697171 0.0000039199999975
0.5009321972921389 0.7509131989197729 0.1966289128943849
0.1817200187612814 0.0884933525842759 0.3393552265550508
0.3201932735921075 0.4132977138357516 0.3393339708594195
0.1116742540284368 0.1211365491383456 0.1500537542312304
0.1214970073454634 0.3577896450660130 0.2479035614919708
0.3759627914951977 0.1206330018289918 0.0519043199999984
0.3803481412579053 0.1440096184562397 0.2479083626831271
0.1259647919524483 0.3812931527206231 0.0518892400000013
0.3902508329501212 0.3808357296738636 0.1501043590418803
0.6817199235014176 0.0884933525842759 0.3393552265550508
0.8201933688519789 0.4132977138357516 0.3393339708594195
0.6116741587685659 0.1211365491383456 0.1500537542312304
0.6214971026053346 0.3577896450660130 0.2479035614919708
0.8759627914951977 0.1206330018289918 0.0519043199999984
0.8803481412579052 0.1440096184562397 0.2479083626831271
0.6259647919524483 0.3812931527206231 0.0518892400000013
0.8902507376902499 0.3808357296738636 0.1501043590418803
0.1817200187612814 0.5884933525842763 0.3393552265550508
0.3201932735921075 0.9132977138357511 0.3393339708594195
0.1116742540284368 0.6211365491383459 0.1500537542312304
0.1214970073454634 0.8577895498061420 0.2479035614919708
0.3759627914951977 0.6206330018289918 0.0519043199999984
0.3803481412579053 0.6440096184562397 0.2479083626831271
0.1259647919524483 0.8812931527206231 0.0518892400000013
0.3902508329501212 0.8808358249337349 0.1501043590418803
0.6817199235014176 0.5884933525842763 0.3393552265550508
0.8201933688519789 0.9132977138357511 0.3393339708594195
0.6116741587685659 0.6211365491383459 0.1500537542312304
0.6214971026053346 0.8577895498061420 0.2479035614919708
0.8759627914951977 0.6206330018289918 0.0519043199999984
0.8803481412579052 0.6440096184562397 0.2479083626831271
0.6259647919524483 0.8812931527206231 0.0518892400000013
0.8902507376902499 0.8808358249337349 0.1501043590418803
#fix atoms
natfix 16
iatfix 4 8 12 16 19 23 27 31 37 39 45 47 53 55 61 63
#Basis definitions
ixc 11
ecut 20
ecutsm 0.5
# pawecutdg 30.
#geometry optimziation
ionmov 2 # Use the modified Broyden algorithm
ntime 100 # Maximum number of Broyden "timesteps"
tolmxf 5.0d-5 # Stopping criterion for the geometry optimization : when
# the residual forces are less than tolmxf, the Broyden
# # algorithm can stop
#Definition of occupation numbers
occopt 3
tsmear 0.01
#
#SCF cycle parameters
tolvrs 1.0d-10
nstep 200
#K-points and sym
kptopt 0 # Enter the k points manually
nkpt 1 # one k point
kpt 0 0 0
#dielectric
#iprcel 45
#pawsushat 1
#diemac 12.0
diemix 0.1
#nsppol 2
#spinmagntarget 0
charge 0
#pawovlp 15.0
iscf 12
usepotzero 2
#parallization parameters
autoparal 1
#I/O parameters
prtwf 1 prtden 0 prteig 0
getwfk -1
- Attachments
-
a.out- output file
- (875.62 KiB) Downloaded 243 times
Re: Large metal oxide surface cell SCF convergence
Dear Dyu,
Did you try to converge more deeply the first scf step to have a better wave function? The converge is indeed very slow in your case but it is difficult to tell you more about it without running the calculation myself. The problem might also depends on the compilation flags you used to generate your executable, just to check it, which flags (or what did you put in your .ac file) did you use to compile abinit?
Best wishes,
Eric
Did you try to converge more deeply the first scf step to have a better wave function? The converge is indeed very slow in your case but it is difficult to tell you more about it without running the calculation myself. The problem might also depends on the compilation flags you used to generate your executable, just to check it, which flags (or what did you put in your .ac file) did you use to compile abinit?
Best wishes,
Eric