Yes, you can change the contribution of each term (HF, exchange Slater, exchange non-local, correlation local, correlation non-local) in most QM software. In gaussian, see the discussion about User-Defined Models at http://www.gaussian.com/g_tech/g_ur/k_dft.htm

Hi there!

That depends highly on the program that you want to use.

The way to do it in TURBOMOLE is described in a HOWTO on this page:

http://www.cosmologic.de/turbomole/support-download/documentation-how-to.html

Cheers,

Arnim

Yes, you can change the contribution of each term (HF, exchange Slater, exchange non-local, correlation local, correlation non-local) in most QM software. In gaussian, see the discussion about User-Defined Models at http://www.gaussian.com/g_tech/g_ur/k_dft.htm

Yes, most programs will let you do this. However you are then doing a different functional to everyone else, so you will have to justify it well. Of course it would be difficult to claim such a calculation was either predictive or "ab initio", unless you had a very good underlying theory to justify your choice.

All the best,

Phil Hasnip

Ya! You can do it with the help of inernal overlay options (iops) in gaussian for eg:

iop(5/45=10000300,5/46=07000700,5/47=07001000)

Thank you Arnim Hellweg , Bartosz Trzaskowski and Philip James Hasnip  fro your valuable suggestions.

I work with G09 and I therefore referred to the following link

http://www.gaussian.com/g_tech/g_ur/k_dft.htm

I tried to understand the "User-Defined Models" and write IOp for B3LYP on my own.

I understand that the standard functional form of B3LYP is

EXC = 0.2*EX(HF) + 0.8*EX(LSDA) + 0.72*DEX(B88) + 0.81*EC(LYP) + 0.19*EC(VWN)

The general form given in the website is

P2EXHF + P1(P4EXSlater + P3ΔExnon-local) + P6EClocal + P5ΔECnon-local

Comparing the B3LYP functional form and the general form , we get

P1=1, P2=0.2, P3=0.72, P4=0.8, P5=0.81, P6=0.19

So I wrote the B3LYP functional form in terms of IOp and found my last term (i.e coefficient, P6) mismatching with what is given in the website.

What I arrived at.

IOp(3/76=1000002000) IOp(3/77=0720008000) IOp(3/78=0810009100)

What is on the website.

#P BLYP IOp(3/76=1000002000) IOp(3/77=0720008000) IOp(3/78=0810010000)

I have the following question, can you please explain.

1)  Why is P6 is given as 1 while in B3LYP functional form it is 0.19.

Kindly let me know if I am not clear.

Thanks,

Raghav

Yes, the coefficients for Ex you get are correct! Ec(LYP) includes both local and non-local terms whereas Ec(VWN) has no 'non-local' term, namely

Ec(LYP)  = 1.0*Ec_local(LYP) + 1.0*ΔEc_non-local(LYP)

Ec(VWN) = 1.0*Ec_local(VWN) + 0.0ΔEc_non-local(VWN)

Ec term in B3LYP is expressed as 0.81*Ec(LYP) + (1-0.81)*Ec(VWN)

then we get 1.0*Ec_local + 0.81*Ec_non-local

so the correct values are P5=0.81, P6=1.0.

cheers!

Thank you Yusuke Tamura for the clarification. That helps!!!