应《网络安全法》要求，自2017年10月1日起，未进行实名认证将不得使用互联网跟帖服务。为保障您的帐号能够正常使用，请尽快对帐号进行手机号验证，感谢您的理解与支持！

24小时热门版块排行榜    

CyRhmU.jpeg
小木虫论坛-学术科研互动平台 » 计算模拟区 » 第一性原理 » Siesta&Smeagol&Atk » 【求助】siesta计算中，如何进行BSSE修正？
101/1返回列表
查看: 3308  |  回复: 9
只看楼主@他人 存档 新回复提醒 (忽略) 收藏    在APP中查看

tailor

金虫 (小有名气)

[交流] 【求助】siesta计算中，如何进行BSSE修正？

即基组重叠误差，全称为Basis Set Superposition Error(BSSE)。对于双分子A和B的相互作用能应该表述为E_interaction = E_AB - E_A - E_B + E_BSSE。第三项即为修正项。对于弱相互作用，E_BSSE所占E_interaction的比例往往不小，甚至超过它，如果不进行校正，可能正负号都不对。
      这些也是从网上查到的，但是网上主要介绍的都是高斯算法中的BSSE修正，由于本人用siesta算，想麻烦各位高手多多赐教！！
回复此楼

» 收录本帖的淘帖专辑推荐

atk&transiesta

» 猜你喜欢

» 本主题相关商家推荐: (我也要在这里推广)

» 本主题相关价值贴推荐，对您同样有帮助:
1楼2010-09-06 21:06:56
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

ifmc1234

木虫 (著名写手)
★ ★ ★
tailor(金币+2):谢谢参与
qasd(金币+2):谢谢 2010-09-11 18:57:22
引用回帖:
Originally posted by tailor at 2010-09-06 21:06:56:
即基组重叠误差，全称为Basis Set Superposition Error(BSSE)。对于双分子A和B的相互作用能应该表述为E_interaction = E_AB - E_A - E_B + E_BSSE。第三项即为修正项。对于弱相互作用，E_BSSE所占E_interaction的 ...

不知道你算的是分子还是固体，建议不要用它来讨论。你可在userguide里搜一下，我没看有什么人用siesta讨论过BSSE,在userguide里的27页提到了BSSE。不过具体怎么做还不知道。说明书中只提到PAO.Basis里设置时，用Ghost原子可以处理BSSE。不过如果根据你上面的定义，我想可以这样E_interaction = E_AB - E_A - E_B + Sum_E_BSSE,其中Sum_E_BSSE为A,B单个的BSSE之和。也就是说用鬼原子来处理BSSE，然后根据上面求的
赞一下(2人) 回复此楼
2楼2010-09-06 21:29:23
 已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

zhangguangping

木虫 (著名写手)
★
tailor(金币+2):谢谢参与
引用回帖:
Originally posted by tailor at 2010-09-06 14:06:56:
即基组重叠误差，全称为Basis Set Superposition Error(BSSE)。对于双分子A和B的相互作用能应该表述为E_interaction = E_AB - E_A - E_B + E_BSSE。第三项即为修正项。对于弱相互作用，E_BSSE所占E_interaction的 ...

为什么要算这个？什么体系才学要考虑BSSE修正啊？
我不是很懂，但是对你的问题很感兴趣，想问问你。
赞一下(1人) 回复此楼
弘德明志博学笃行
3楼2010-09-06 21:30:41
 已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

guifan
★
tailor(金币+2):谢谢参与
4楼2010-09-06 21:41:19
 已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

tailor

金虫 (小有名气)

请教老外，得到的回复

★
zxzj05(金币+1):谢谢回帖交流 2010-09-25 22:34:07
引用回帖:
Originally posted by zhangguangping at 2010-09-06 21:30:41:

为什么要算这个？什么体系才学要考虑BSSE修正啊？
我不是很懂，但是对你的问题很感兴趣，想问问你。

What we calculated is the Basis Set Superposition Error (BSSE) that is an error occurring when you use a finite basis set to study, for example, a chemical reaction. In a tipical reaction you calculate the electronic structure of the reactants separately, let us say A and B, and then the electronic structure of the product, A+B. What happens in this case is that you are using different basis sets to represent the reactants in each step of the calculation. When you calculate the isolated reactant A, you use the basis set of A, and the electrons of A will be arranged spatially on these provided orbitals. However, when you put both systems together in the A+B, the electrons of A will have now a greater spatial freedom, since they can now occupy orbitals coming from B. The same happens with B. You have different basis sets to distribute the electrons of A in the two situations.
To correct this, you need to recalculate the energy of the isolated reactants (A and B) by considering the basis set of both reactants, i.e., the calculations of the isolated reactant A will now include the basis sets of B (without the system B itself). It needs to be done for each configuration that you take into account in the reaction of A+B. Now you will have the same basis set for the isolated reactants and the product.
You can look for more information searching for "counterpoise correction"
赞一下(3人) 回复此楼
5楼2010-09-10 19:36:33
 已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

zhangguangping

木虫 (著名写手)
★
zxzj05(金币+1):谢谢回帖交流 2010-09-25 22:34:19
引用回帖:
Originally posted by tailor at 2010-09-10 12:36:33:



What we calculated is the Basis Set Superposition Error (BSSE) that is an error occurring when you use a finite basis set to study, for example, a chemical reaction. In a tipical reaction you ...

意思就是说因为计算一个体系的时候前后，使用的基组不一样大造成了一定的误差，这个误差需要修正。如果第二次引入的那部分基函数和第一次的基函数都正交的话，也就是各自是一个子空间的话，这个修正就不用做了吧？
赞一下(1人) 回复此楼
弘德明志博学笃行
6楼2010-09-11 10:02:01
 已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

夕阳西下

金虫 (著名写手)
★ ★
tailor(金币+2):谢谢参与
qasd(金币+1):谢谢 2010-09-11 18:57:35
即使基矢正交，Ａ与Ｂ反应后的整个空间基矢也会发生改变。应该是原来两个空间的直积。如果Ａ和Ｂ是同种原子应该不用修正。
赞一下(2人) 回复此楼
7楼2010-09-11 12:08:57
 已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

cenwanglai

荣誉版主 (知名作家)

老和山猥琐派九段

优秀版主
★
tailor(金币+2):谢谢参与
这个在vasp中（体相材料或者表面结构）需要考虑BSSE吗？如何进行呢？
赞一下(1人) 回复此楼
8楼2010-09-25 16:50:46
 已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

skyscorpioa

新虫 (初入文坛)
★ ★ ★
tailor(金币+2):谢谢参与
zzy870720z(金币+2):谢谢指教 2010-11-20 08:14:28
引用回帖:
Originally posted by cenwanglai at 2010-09-25 16:50:46:
这个在vasp中（体相材料或者表面结构）需要考虑BSSE吗？如何进行呢？

VASP用的是plane wave，平面波不像siesta中的numerical atomic orbital 或者gaussian 中的Gaussian orbital 一样固定在原子上，只要晶格矢量和平面波阶段一样，不管原胞内原子数的变化，基函数都是一样的
赞一下(3人) 回复此楼
9楼2010-11-20 02:43:11
 已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

valenhou001

至尊木虫 (职业作家)
★
tailor(金币+2): 谢谢参与
这里有很详细的步骤：
https://www.mail-archive.com/siesta-l@uam.es/msg02916.html

拷贝如下：

Re: Re: [SIESTA-L] about basis-set superposition error (BSSE)

Abraham Hmiel Tue, 29 Mar 2011 21:51:07 -0700

"Should I set a ghost atom between the molecule and the nanotube ?"


No, this is what you must do:

You've already got the relaxed structure of the nanowire + adsorbed molecule
system, right? Let's call this system "A" and its total energy is E-A

You need 6 more calculations to complete the puzzle this is pretty much what
you should do:

1 calculation with the same k-point grid, mesh grid, and cell size as system
A, but only for the nanotube, fully relaxed to a similar tolerance as system
A. Find its total energy. Call this E-NT.

1 calculation with the same k-point grid, mesh grid, and cell size as system
A, but only for the adsorbed molecule, fully relaxed to a similar tolerance
as system A. Find its total energy. Call this E-AD

1 single-point calculation (0 CG steps) of the nanowire in the relaxed
geometry of system A (use the final .XV file or .xyz file or whatever you
want but remove the adsorbed molecule). Find its total energy. Call this
E-noghost-NT

1 single-point calculation (0 CG steps) of the adsorbed molecule in the
relaxed geometry of system A (use the final .XV file or .xyz file or
whatever you want but remove the adsorbed molecule). Find its total energy.
Call this E-noghost-AD

1 single-point calculation (0 CG steps) of the nanowire in the relaxed
geometry of system A, except replace any adsorbate chemical species with
ghost atoms. If you have species in the adsorbate that are present in the
nanowire, for example, a simulation of H2O on a hydrogen-passivated SiNW (or
C in methane on a CNT), then copy the H.psf file to a new file like
H_ghost.psf and then create a new chemical species H_ghost with atomic
number -1 and a different atomic species index. Replace any index of the H
in the adsorbate with the new index, and introduce a new basis set for
H_ghost that is identical to the one you used for H (except for the label
H_ghost). Find its total energy. Call this E-ghost-NT

1 single-point calculation (0 CG steps) of the adsorbed molecule in the
relaxed geometry of system A, except replace any nanowire chemical species
with ghost atoms and follow the procedure above if you have any species in
the adsorbate that are also present in the nanowire. Find its total energy.
Call this E-ghost-AD

The counterpoise correction is: (E-ghost-AD - E-noghost-AD + E-ghost-NT -
E-noghost-NT) call this E-CC. It should be a fraction of an eV, have a
negative sign and very sensitive to the adsorption site geometry. The BSSE
should _reduce_ the adsorption energy...

and the energy of adsorption is: E-NT + E-AD - E-A + E-CC

And that is how you do the counterpoise correction with SIESTA.

Best,
Abraham Hmiel

Katherine Belz Groves Graduate Fellow in Nanoscience, Xue Group
The College of Nanoscale Science and Engineering at SUNY Albany
"Clouds are not spheres, mountains are not cones, coastlines are not
circles,
and bark is not smooth, nor does lightning travel in a straight line." -
Benoit Mandelbrot

On Wed, Mar 30, 2011 at 12:09 AM, yf liu <liuyf1...@gmail.com> wrote:

> Dear Herbert Fruchtl:
>      thank you very much for your reply.  I have read the manual carefully,
> but find little message about the ghost atoms. the new question is: How can
> I use the ghost atom to correct the binding energy etc. ?  For example, I
> want correct the calculation about the molecule adsorption on the nanotube.
> Should I set a ghost atom between the molecule and the nanotube ?
>
> looking forward your reply.
>
> 2011/3/20 Wei Hu <gyrw4...@mail.ustc.edu.cn>
>
>>    I am sorry, I am a freshman about the siesta. The input is listed in
>> the following.
>>    My calculations about the binding energy do not meet the experimental
>> results,so,I have to check the ghost atoms calculations.Now,The impact is
>> acceptable.
>>    Another problem is how to get the chemical potential or total energy of
>> free C or N atom. Does it need to calculate the the ghost atoms affected by
>> the supercell?
>>
>> SystemName      C62N_ghost
>> SystemLabel     C62N_ghost
>> NumberOfSpecies         3
>>
>> %block ChemicalSpeciesLabel
>>  1   6   C
>>  2   7   N
>>  3  -6   Cg
>> %endblock ChemicalSpeciesLabel
>>
>> %block PS.lmax
>> C 1
>> %endblock PS.lmax
>>
>> %include coord.fdf
>>
>> PAO.BasisSize DZP
>>
>> #SolutionMethod   dm_on
>>
>>
>> SolutionMethod   diagon
>>
>> MeshCutoff          200.0000000     Ry
>>
>> #MD.TypeOfRun  Broyden
>> #MD.TypeOfRun  CG
>> #MD.NumCGsteps 500
>>
>> WriteForces
>>
>> MD.MaxForceTol 0.04 eV/Ang
>> #DM.UseSaveDM T
>>
>> MaxSCFIterations 100
>> DM.MixingWeight  0.1
>> DM.NumberPulay   6
>> #DM.MixingWeight 0.25
>> #DM.NumberPulay 0
>>
>>
>> SpinPolarized .true.
>> #FixSpin .true.
>> #TotalSpin 2.0
>>
>> WriteMullikenPop  1
>>
>> NetCharge  -1.0
>>
>> %block kgrid_Monkhorst_Pack
>>   2  0  0  0.0
>>   0  2  0  0.0
>>   0  0  2  0.0
>> %endblock kgrid_Monkhorst_Pack
>>
>>
>>
>> > -----Original E-mail-----
>> > From: "Herbert Fruchtl" <herbert.fruc...@st-andrews.ac.uk>
>> > Sent Time: 2011-3-18 20:06:46
>> > To: siesta-l@uam.es
>> > Cc:
>> > Subject: Re: [SIESTA-L] about basis-set superposition error (BSSE)
>> >
>> >  From the energy part of the output alone we can't tell if the input was
>> > correct. You see a smallish difference and lower energy with the ghost,
>> which
>> > one would expect.
>> >
>> > Total energies are meaningless, and of course adding a few basis
>> functions won't
>> > change the total by a lot. You need to compare differences (like a
>> binding
>> > energy, as you indicated in your original post).
>> >
>> > In the example outputs below, the difference (in absolute energies) is
>> 0.09 eV.
>> > That's 2 kcal/mol or 9 kJ/mol. Depending on the interaction you are
>> looking at,
>> > this may or may not be negligible.
>> >
>> >    Herbert
>> >
>> > On 03/18/2011 02:00 AM, Wei Hu wrote:
>> > > Dear,the ghost atoms calculations seem to little effects on the total
>> energy from my results. Is there any wrong?
>> > >
>> > > My results about a nitrogen-vacancy center in bulk diamond C62N
>> (2*2*2):
>> > >
>> > > The results with ghost atoms:
>> > >
>> > > siesta: Program's energy decomposition (eV):
>> > > siesta: Eions   =     16680.576136
>> > > siesta: Ena     =      3653.202863
>> > > siesta: Ekin    =      7163.588024
>> > > siesta: Enl     =      -820.400928
>> > > siesta: DEna    =      -182.458272
>> > > siesta: DUscf   =        15.575306
>> > > siesta: DUext   =         0.000000
>> > > siesta: Exc     =     -3054.613637
>> > > siesta: eta*DQ  =         0.000000
>> > > siesta: Emadel  =         0.000000
>> > > siesta: Ekinion =         0.000000
>> > > siesta: Eharris =     -9905.679557
>> > > siesta: Etot    =     -9905.682781
>> > > siesta: FreeEng =     -9905.682783
>> > >
>> > > siesta: Final energy (eV):
>> > > siesta:       Kinetic =    7163.588024
>> > > siesta:       Hartree =     894.608032
>> > > siesta:    Ext. field =       0.000000
>> > > siesta:   Exch.-corr. =   -3054.613637
>> > > siesta:  Ion-electron =   -8393.758885
>> > > siesta:       Ion-ion =   -6515.506314
>> > > siesta:       Ekinion =       0.000000
>> > > siesta:         Total =   -9905.682781
>> > >
>> > > The results without ghost atoms:
>> > >
>> > > siesta: Program's energy decomposition (eV):
>> > > siesta: Eions   =     16680.576136
>> > > siesta: Ena     =      3653.202863
>> > > siesta: Ekin    =      7164.385562
>> > > siesta: Enl     =      -820.554764
>> > > siesta: DEna    =      -182.943454
>> > > siesta: DUscf   =        15.589130
>> > > siesta: DUext   =         0.000000
>> > > siesta: Exc     =     -3054.694636
>> > > siesta: eta*DQ  =         0.000000
>> > > siesta: Emadel  =         0.000000
>> > > siesta: Ekinion =         0.000000
>> > > siesta: Eharris =     -9905.587400
>> > > siesta: Etot    =     -9905.591435
>> > > siesta: FreeEng =     -9905.591437
>> > >
>> > > siesta: Final energy (eV):
>> > > siesta:       Kinetic =    7164.385562
>> > > siesta:       Hartree =     894.861106
>> > > siesta:    Ext. field =       0.000000
>> > > siesta:   Exch.-corr. =   -3054.694636
>> > > siesta:  Ion-electron =   -8394.637153
>> > > siesta:       Ion-ion =   -6515.506314
>> > > siesta:       Ekinion =       0.000000
>> > > siesta:         Total =   -9905.591435
>> > >
>> > >
>> > >
>> > >
>> > >> -----Original E-mail-----
>> > >> From: "Herbert Fruchtl"<herbert.fruc...@st-andrews.ac.uk>
>> > >> Sent Time: 2011-3-17 22:36:53
>> > >> To: siesta-l@uam.es
>> > >> Cc:
>> > >> Subject: Re: [SIESTA-L] about basis-set superposition error (BSSE)
>> > >>
>> > >> You can calculate counterpoise-corrected energies by specifying ghost
>> atoms.
>> > >> There is no way of doing counterpoise-corrected geometry
>> optimisations in SIESTA.
>> > >>
>> > >>    Herbert
>> > >>
>> > >> On 03/17/2011 01:57 PM, yf liu wrote:
>> > >>> Hello everybody,
>> > >>>
>> > >>> whether the basis-set superposition error (BSSE) is included in
>> siesta
>> > >>> calculation. when i calculate the molecule adsorption on carbon
>> nanotube, i
>> > >>> found the adsorption energy is some higher than the well know value.
>> how can i
>> > >>> add the BSSE in siesta calculation?
>> > >>>
>> > >>> yours
>> > >>> yufeng liu
>> > >>>
>> > >>
>> > >> --
>> > >> Herbert Fruchtl
>> > >> Senior Scientific Computing Officer
>> > >> School of Chemistry, School of Mathematics and Statistics
>> > >> University of St Andrews
>> > >> --
>> > >> The University of St Andrews is a charity registered in Scotland:
>> > >> No SC013532
>> > >
>> >
>> > --
>> > Herbert Fruchtl
>> > Senior Scientific Computing Officer
>> > School of Chemistry, School of Mathematics and Statistics
>> > University of St Andrews
>> > --
>> > The University of St Andrews is a charity registered in Scotland:
>> > No SC013532
>>
>>
>
赞一下 回复此楼
10楼2014-12-02 09:25:05
 已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖
相关版块跳转 我要订阅楼主 tailor 的主题更新
101/1返回列表
普通表情 龙 兔 虎 猫 高级回复(可上传附件)
信息提示
关闭
请填处理意见
关闭