Density matrix is not changing but DIIS error ……The SCF is confused.解决方案

Density matrix is not changing but DIIS error= 1.29D-01 CofLast= 6.96D-04.
The SCF is confused.
Error termination via Lnk1e in /usr/local/g03/l502.exe

解决方案：

一、

The error about “Density matrix not changing…….The SCF is confused" is a condition where the HF is encountering problems and the density matrix, which is what is monitored to watch convergence. One thing which has been successful in settling this problem is to add SCF=NoVarAcc so that all of the small integrals are included from the first iteration.

密度矩阵不收敛，但能量收敛了。说明轨道还不够合理，换个方法猜测轨道。但真正的原因还是小积分项忽略。

二、

可以试试：这是g03 d01以上版本才会出现的，c02及以下不会出现。
1.   scf=(novaracc,nosymm,xqc)
2.   在scf里面采用 SD或 Fermi关键词。
3. 实在不行就从断点开始用SCF=qc优化几圈，然后改用一般SCF。qc很费时

三、

SCF(自洽场)不收敛一般是L502错误
修改坐标，使之合理；增加叠代次数 SCFCYC=N ；iop（5/13=1）忽略不收敛，继续往下做

四、最有效的方法是调整了一下初始结构，一切搞定！（自己亲自体会）

Other Examples

(1) 

I also had this problem (for a UHF calculation). Using the quadratic SCF (as
 stated below: scf=qc or scf=xqc) and taking of symmetry by using the NoSymm
 keyword solved the issue for me.
 Hope that helps.

(2)
Sent to CCL by: Eric Breynaert [eric.breynaert_._biw.kuleuven.be]
 Thanks for the suggestions. I'm giving an overview below.
 1/
  - SCF=qc will probably solve the problem, albeit at a cost
  - Change the SCF converger to either SD, Quadratic or Fermi
 2/
  - lower the symmetry of optimize with and optimize
    with the "nosymm" keyword

I solved the problem using a variation on the first suggestion. Normally the scf took less than 80 cycles to converge. So i used scf=(Maxconventionalcycles=100,xqc) whichresulted in a good compromise between using scf=qc and optimisation speed. In the case of the DIIS error the scf always took more than 100 cycles before the error, so by adding scf=(Maxconventionalcycles=100,xqc) the scf switched to qc after 100 cycles in the standard DIIS mode.
 Regards,
 Eric