|
|
【答案】应助回帖
★ ★ ★
gmy1990: 金币+3 2014-05-27 01:44:54
最近正看ccl上讨论了这个问题 copy到这里供参考
A:
I am trying to run a calculation on a heavily correlated cluster, which,
however, already fails at the SCF level. After playing along with damping
and level shifting, I believe that it now is converging (the energy
decreases smoothly). However the maximal value MAXIT=200 is probably not
enough since it does not actually reach convergence. I believe that further
iterations might be necessary. Is there a way to increase this? I remember
once reading that GAMESS must be recompiled with some value changed, but I
do not know anymore.....
B:
You may have to change this line from comp.src (line 1578 2013 version)
MAXIT = 200
to the umber of iteration you want as well as the line from inputa.src
as suggested by the other poster.
A:
thank you for your replies. I changed the upper MAXIT value in inputa.src and recompiled. Now everything works beautifully, with a typical (almost monotonous) energy convergence within about 400 cycles. The reason it takes so long is because I use a large damping to avoid oscillations caused by the many degenerate orbitals, and the results for all geometries look reasonable. Of course, post-HF calculations will follow. |
|
回复此楼