刚刚在J. Chem. Theory Comput., ASAP Article中发布的论文：

Stereoelectronic Substituent Effects in Saturated Main Group Molecules: Severe Problems of Current Kohn-Sham Density Functional Theory

作者：S. Grimme, M. Steinmetz, and M. Korth
说：

...It is found that the unacceptably large errors observed previously for hydrocarbons persist also for the here considered more polar systems...

这对DFT在有机分子计算中的应用有何影响？

论文摘要如下：

The Hartree-Fock method, two common density functionals (PBE and B3LYP), and two new functionals (B97-D and B2PLYP) together with very large AO basis sets are used to compute the isomerization energies for substituted (R=H, F, Cl) branched to linear alkanes and silanes. The results of accurate SCS-MP2 computations are taken as reference. These reactions are an important test of how nonlocal electron correlation effects on medium-range lengths scales in saturated molecules are treated by approximate quantum chemical methods. It is found that the unacceptably large errors observed previously for hydrocarbons persist also for the here considered more polar systems. Although the B97-D and B2PLYP functionals provide improved energetics, the problem is not fully solved, and thus these systems are suggested as mandatory benchmarks for future density functionals.

有点夸大了。现有的DFT处理这种medium-range或者是long-range的弱相互作用处理得不好是众所周知的，作者只不过是找了一个含这种若相互作用的体系用几个新的泛函尝试了一下然后又和MP2的结果进行了一下对比而已，感觉没什么新东西，umn的truhler有一系列这方面比较新泛函的文章

I attended a talk from the author. I am suprised that the error is quite large when for isomer. I think you promote his method: includion of MP2 energy to DFT.

Truhlar developed or tune many functionals for many situations, but you will be confued to select the best functional for studying large systems containg diff. bonding. Every one hope to have one universal functional, but many.

引用:

原帖由 alpha 于 2006-12-5 09:39 发表
有点夸大了。现有的DFT处理这种medium-range或者是long-range的弱相互作用处理得不好是众所周知的，作者只不过是找了一个含这种若相互作用的体系用几个新的泛函尝试了一下然后又和MP2的结果进行了一下对比而已，感 ...

我比较同意alpha 的看法，现有的DFT在处理分子间弱相互作用上面确实存在着比较突出的弱点，而这也是现在做关于DFT泛函的科研组所努力解决的主要问题。UMN的Truhlar在这方面有着很重要的贡献，去年ACS的hot articles就有两篇他的DFT研究文章。另外我知道DUKE的 Weitao Yang也做的不错，算是我知道里头做DFT最好的中国学者了。

引用:

原帖由 choscar 于 2006-12-16 17:11 发表
I attended a talk from the author. I am suprised that the error is quite large when for isomer. I think you promote his method: includion of MP2 energy to DFT.

Truhlar developed or tune many fun ...

http://www.quantumchemistry.net/ ... mm/200610/1159.html
“第62届ACS西南区分会会议感受
虽然只是一个分会，不过也有几百人参加。头一天下午有一个DFT理论的进展和应用的论坛，在那里终于见到了闻名已久的JP Perdew和W Yang。从他们的报告看来，Perdew还在继续开发新的泛函，不过现在的重点应该是集中在尽量用一种泛函能够精确计算long distance的结构，比如说模拟分子的离解过程。我曾提问他对MCSCF和DFT结合的前景怎么看，他的回答也很有意思：“Everything is possible。”，但是他本人更喜欢用一个泛函去得到更精确的结果。这也可能是为什么他开发的泛函都是pure density functional，而不是混合泛函的原因吧。Weitao Yang现在正在做一个potential functional theory，目的是从根本上解决DFT对长距离结构不能精确描述的问题，所需的计算量根据potential functional的不同，计算量应该小于DFT的5倍。没有听太懂理论的细节，不过感觉还是很不错的。其他的还有不少，Rice大学由于是东道主的关系，也报告了很多他们在开发DFT方面的进展，东西太多，已经记不清细节了。可惜DFT的论坛只有一下午，其他时间也没有专门的理论化学论坛，都分散在不同的主题里，就只好看着会议的秩序册，在各个报告厅钻来钻去，找自己感兴趣的题目了。”

I think this is due to non-balanced treatment of Fermi and Coulomb correlation in the most presently available DFT methods(GGA, meta-GGA, hybrid). Non-local correlation sometimes is extremely crucial to describe the electronic structure, even in qualitative way. Now we are studying the spin state energetics employing different DFT methods. We found B2-PLYP is the only one DF which is able to pridict the correct spin-multiplicities of the ground states for a range of iron complexes.

To ebwilson, how do you get the correct spin of the ground state?  Experimentally or other calculations?
BTW, it's quite big a name that you are using.

According to the experimental findings.

JCTC这个杂志办了有一年了吧，目的大概是想和JCC抗衡，感觉不算好，有价值的文章不多。

引用:

原帖由 curioz 于 2007-3-26 04:29 发表
JCTC这个杂志办了有一年了吧，目的大概是想和JCC抗衡，感觉不算好，有价值的文章不多。

agree！