[gmx-users] Re: QM/MM (Ramon Crehuet)

[ggroenh]

>
> Dear all,
> This is a follow-up of a previous question and I'll try to be as  
> precise
> as possible. I'd like to use QM/MM interface with MOPAC and Gaussian  
> 03.
> Here are my questions.
> 1. In Grubmueller's page the link to MOPAC7 is broken, but one can get
> mopac7.tar.gz. This file seems to be the modified version, because it
> already contains gmxmop.f.

Correct. Mopac can be obtained from http://sourceforge.net/projects/mopac7/ 
  I have changed the link.


> And it seems to be the SGI version (but they
> differ in one line). So should I replace the gmxmop.f in mopac7.tar.gz
> with the linux version?
> 2. I'd like to compile with ifort but Gerrit told me they got  
> unreliable
> results. In which sense are they unreliable? He recommended a f2c/gcc
> approach with 32-bit architecture, but mine is 64-bit. Can I use f77?
> When should I expect problems, when compiling gromacs or when  
> executing
> (harder to trace!)?

No, you could try to compile a static mdrun with libmopac and all  
other libraries built in on a 32-bit architecture, and copy the binary  
to you 63 bit machines. We have tried with icc, but seen that the  
results are not identical to the f2c/gcc and older f77 compilations. I  
am still in the process of sorting out this issue. Would making a  
binary available help?

>
> 3. Apart from setting --with-qmmm-gaussian when compiling gromacs, how
> should I proceed to compile with gaussian? Do I need to recompile
> gaussian?

No, you need a few modification to the source code, but can compile  
these seperately. There are now some instructions on the webpage on  
how to do that:

http://www.mpibpc.mpg.de/groups/grubmueller/start/people/ggroenh/qmmm.html#code

best,

Gerrit



> Tell gromacs where the gaussian sources/objects are? Gaussian
> is usually compiled with Portland compilers... are they compatible  
> with
> gromacs?
>
> Please, comment if you have some experience with *any* of these  
> points.
> I'll gladly share mine when I get something woriking.
>
> Cheers,
> Ramon
>

--
Gerrit Groenhof
MPI biophysical chemistry
Goettingen
Germany
http://www.mpibpc.mpg.de/groups/grubmueller/start/people/ggroenh/index.html