The ABINIT Users Mailing List ( CLOSED )

[xd0533@163.com]

dear users;
   When I do YTaO4 geometry optimization,my result of acell is always in 
worse 
agreement with experimental value.It is about 10% error.This error is very 
large.In order to reduce the error,I try different kinds of method.Such as 
changing xc,pseudopotentials(FHI(GGA),TM,HGH(LDA,semi-core)) and so on.But 
there is no effect.The input file I have attached in 
http://www.abinit.org/wws/arc/forum/2007-06/msg00075.html I try to increase 
convergence criterion.But I think this is not sticking point. 
The structure of YTaO4 is monoclinic.Now I have an idea.I don't know whether 
it is reasonable.My idea is:first,I use optcell=0 to modify nuclear 
positions,second I use optcell=1 to optimize volume and acell,but optcell=1 
doesn't optimize rprim,so next step I use optcell=3 to optimize rprim.Because 
for second step,I have optimized volume,so in the third step I can fix 
volume.I don't know whether my idea is reasonable.I hope to get your help.
Thank everyone for answering my question.
Best Wish!
xiang
To:Masayoshi Mikami 
I have seen your article of calculation Y2O2S.Your result of acell is in good 
agreement with experimental value.This shows TM pseudopotential of Y is 
reasonable.And I saw article about KTaO3.He also uses TM pseudopotentials.So 
I believe TM pseudopotentials of YTaO4 is reasonable.But why is my result so 
bad?
What is your opinion about this problem,Mr Masayoshi Mikami.I hope your reply!
Thanks a lot!