When you calculate force in two different spin states, then you can get 梯度 for the different spin.

引用:

原帖由 ccidnetcom 于 2008-6-20 15:10 发表
给Gordon写信已有10天了吧,还没见回复.
DFT方法计算势能面梯度差我没试过,麻烦你能再深入介绍下吗?
十分感谢你的帮助!

BTW, can you share an ebook" NONADIABATIC TRANSITIONS Concepts, Basic Theories and Applications" with me or us?

文献我传到了网上,很高兴你也对这个方向感兴趣!
http://g.zhubajie.com/urllink.php?id=2588815rz88ok3sxzyluvhj
文献很不错,我看的不多,以后还希望你指点!

顺便还有问题,我用J.N.Harvey的Crossing2004程序包计算了两势能面的交叉点所对应的最低能量交叉点(MECP),如果我要计算MECP点的Landau-Zener跃迁几率,这时MECP点处的梯度差是否和G03下CP点附近的梯度差相同呢?
这是Crossing2004找到MECP后的部分输入文件:

Geometry at Step 18
  6     -1.8285865     -0.0366090      0.1594269
  1     -2.3166355     -0.5805176     -0.6551690
  1     -2.1388912      1.0102687      0.1009388
  1     -2.1913298     -0.4454553      1.1077786
  1      1.6058601      0.8704737     -0.4442845
75      0.3074879     -0.1753686      0.0126932

Energy of First State:     -118.7147209450
Energy of Second State:    -118.7147207560

Convergence Check (Actual Value, then Threshold, then Status):
Max Gradient El.:   0.000320 (0.000700)  YES
RMS Gradient El.:   0.000135 (0.000500)  YES
Max Change of X:    0.000704 (0.004000)  YES
RMS Change of X:    0.000291 (0.002500)  YES
Difference in E:    0.000000 (0.000050)  YES

Overall Effective Gradient:
  1     -0.00022589     -0.00000281      0.00009367
  2      0.00011155      0.00009810      0.00000019
  3      0.00009820     -0.00032041      0.00010019
  4      0.00008381      0.00027594     -0.00012887
  5     -0.00003501      0.00005792      0.00000540
  6     -0.00003265     -0.00010872     -0.00007057

Difference Gradient: (RMS * DE:   0.020900)
  1      0.02385443      0.04572778     -0.02059291
  2      0.01149468     -0.00117398     -0.00601337
  3     -0.00622857     -0.00707550      0.00145379
  4      0.01892810     -0.00331754      0.00854815
  5     -0.05129095      0.00119207      0.00356336
  6      0.00324231     -0.03535284      0.01304098

Parallel Gradient: (RMS:   0.000135)
  1     -0.00022526     -0.00000160      0.00009312
  2      0.00011185      0.00009807      0.00000003
  3      0.00009803     -0.00032060      0.00010022
  4      0.00008431      0.00027585     -0.00012864
  5     -0.00003637      0.00005795      0.00000549
  6     -0.00003256     -0.00010966     -0.00007023

The MECP Optimization has CONVERGED at that geometry !!!
Goodbye and fly with us again...

顺便还要说的是,Gordon在描述跃迁几率的时候,引用了37,38两篇文献,其中就有Nakamura的一篇(这篇我没看懂),一并上传来,希望能和你继续讨论.谢谢

http://g.zhubajie.com/urllink.php?id=25914152sxwmzwwmahn22s9
http://g.zhubajie.com/urllink.php?id=2591468tm3ia4mftw9fyoks

In fact, I am doing excited state MD and need to model the crossing by surface hopping. My molecule has same spin and symmetry. Therefore, I do not use SOC, but DC.

If I remember correctly.Harvey's MECP code calls G03 to calculate E and gradient at the different spin states. Then, gradient difference can be obtained in the MECP code. G03 does not do  gradient difference, nor the MECP code calculate gradient.

I am not a physical chemist. I also don't know equations. I heard there is another way to get transition probability, Zhu-Nakamura formula.

引用:

原帖由 ccidnetcom 于 2008-6-20 22:16 发表
文献我传到了网上,很高兴你也对这个方向感兴趣!
http://g.zhubajie.com/urllink.php?id=2588815rz88ok3sxzyluvhj
文献很不错,我看的不多,以后还希望你指点!

顺便还有问题,我用J.N.Harvey的Crossing2004程序包计 ...

关于S.Shaik猜测瞬时速率的问题,我发邮件问了后,这是他的相当简单的回复:
We use the kinetic theory of gases to evaluate relative velocities (this is a gas phase reaction).But of course getting probability values from CASSCF is perfect.I hope this answers your question.
等于没有说嘛