24小时热门版块排行榜

返回列表

【奖励】本帖被评价3次，作者jjf_sxnu增加金币 3 个

当前主题已经存档。

jjf_sxnu

铜虫 (正式写手)

应助: 5 (幼儿园)
金币: 105.7
帖子: 324
在线: 18.7小时
虫号: 321353

[资源] 【共享资料】How to handle a SCF convergence fail (Link 502)

I find so many friends encounter the problem about a difficult SCF convergence. I find a completely resolution in a book, now paste it here to share with everybody.

1. Try a dfferent initial guess. Many programs have several dfferent initial
guess procedures, often based on semiempirical calculations.
2. For an open-shell system, try converging the closed-shell ion of the same
molecule and then use that as an initial guess for the open-shell calculation.
Adding electrons may give more reasonable virtual orbitals, but as
a general rule, cations are easier to converge than anions.
3. Another initial guess method is to first run the calculation with a small
basis set and then use that wave function as the initial guess for a larger
basis set calculation.
4. Try level shifting. This will usually work with the default parameters or
not at all.
5. If the SCF is approaching but not reaching the convergence criteria,
relax or ignore the convergence criteria. This is usually done for geometry
optimizations that do not converge at the initial geometry. Geometry
optimizations often converge better as they approach the equilibrium
geometry.
6. Some programs use reduced-accuracy integrals to speed the SCF. Using
full integral accuracy may be necessary for systems with di¨use functions,
long-range interactions, or low-energy excited states. Turning o¨
incremental Fock matrix formation may also be necessary for these
systems.
7. For DFT calculations, use a finer integration grid.
8. Try changing the geometry. First, slightly shorten a bond length. Then,
slightly extend a bond length and next shift the conformation a bit.
9. Consider trying a different basis set.
10. Consider doing the calculation at a di¨erent level of theory. This is not
always practical, but beyond this point the increased number of iterations
may make the computation time as long as that occurring with a
higher level of theory anyway.
11. Turn off the DIIS extrapolation. Give the calculation more iterations
along with this.
12. Give the calculation (with DIIS) more SCF iterations. This seldom
helps, but the next option often uses so many iterations that it is worth
a try.
13. Use a forced convergence method. Give the calculation an extra thousand
iterations or more along with this. The wave function obtained by
these methods should be tested to make sure it is a minimum and not
just a stationary point. This is called a stability test.
14. See if the software documentation suggests any other ways to change the
DIIS method. You may have to run hundreds of calculations to become
experienced enough with the method to know what works when and by
how much to adjust it.

ENJOY !!!

[ Last edited by jjf_sxnu on 2008-5-20 at 16:01 ]

回复此楼