Thanks for your really fast action, Ross.
I tried several hundred steps of md simulation with ntpr=1 and ntwx=1.
The system is N-methylacetamide (AM1) in cubic TIP3P water box.
There are differences between the numbers generated by the new implementaion
and those by manual calculations (single precision).
It seems probably that the differences come from different precisions in numbers, but I'm not sure.
If looking into just a few permanent dipole components,
the output from the new implementation is:
step 1: 2.897 0.608 3.776
.....
step 20: 1.565 1.464 4.246
.....
and the manual output is:
step 1: 2.903 0.604 3.781
.....
step 20: 1.564 1.463 4.240
.....
If you have any comments, please let me know.
I'll proceed with the new implementation if there is nothing important to be added in the patch.
Thanks a lot.
Seongeun
----- Original Message -----
From: "Ross Walker" <ross.rosswalker.co.uk>
To: <amber.scripps.edu>
Sent: Friday, November 02, 2007 2:12 AM
Subject: RE: AMBER: &dipoles output in QM/MM simulations
> Hi Seongeun,
>
> Amber 9 doesn't currently support the calculating of dipoles for QM atoms.
> To do this "correctly" would be difficult. However doing it with mulliken
> charges is not very difficult although of course you have to accept the
> approximations that this involves. From your email it seems that you would
> be happy with QM dipoles calculated using mulliken charges so please try
> applying the attached patch to your amber9 tree. Note this assumes your
> amber9 tree is fully patched already with the latest bug fixes.
>
> Save file in $AMBERHOME/src/sander/
>
> cd $AMBERHOME/src/sander/
> make clean
> patch -p0 <qmmm_mchg_dipoles.patch
> make sander
> mv sander ../../exe
>
> NOTE: This is something I hacked together in a few minutes. I do NOT have
> time at the moment to test it - other than to check it compiles...
>
> So "caveat emptor". You will need to test this yourself thoroughly to make
> sure it gives the correct answer. I would suggest setting up a small pure
> QM system in gas phase, say methane. Then run a few steps saving the
> coordinates on every step and printing the mulliken charges. This way you
> can calculate what the answer should be manually yourself and compare it to
> what the code gives. Then try again with QM/MM - say two methanes side by
> side, one QM one MM and again check what the code gives to what your manual
> calculations give.
>
> Note 2: I have absolutely no idea what will happen to the printed dipoles
> if you are using link atoms!
>
> or if it will work in parallel. It probably will but I didn't try it...
>
> Good luck,
>
> Ross
>
> /\
> \/
> |\oss Walker
>
> | HPC Consultant and Staff Scientist |
> | San Diego Supercomputer Center |
> | Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
> | http://www.rosswalker.co.uk | PGP Key available on request |
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Nov 04 2007 - 06:07:38 PST
