Re: [AMBER] Amber QM/MM

From: Ross Walker <>
Date: Thu, 12 Dec 2013 07:38:32 -0800

Hi Fengxue,

Note the internal diagonalizer in sander is only really efficient for up
to 60 QM atoms or so. For larger systems I suggest compiling using MKL -
You'll need to obtain and install MKL and then set the MKL_HOME
environment variable to point to it's location. Then recompile. Once that
is done set diag_routine=0 in the qmmm namelist and this will
automatically benchmark all the diagonalization options and choose the
most efficient one.

Good luck,


On 12/11/13 6:57 AM, "Gustavo Seabra" <> wrote:

>Hi Fengxue Li,
>What do you mean by "spends too much time"? QM calculations will normally
>take a lot longer than regular MM calculations, so the QM/MM calculation
>expected to take longer than regular MM.
>Also, if you are using sander for the QM part, it probably means you are
>already using a semi-empirical QM method - the fastest QM you can do.
>My only suggestions by now would be:
>1. Check to see if you are having convergence problems. This could make
>calculations even longer than usual for QM, and the results could be
>meaningless. (Sander will print warnings if this is the case.)
>2. If the time is still above what you can manage, would you consider
>reducing the QM zone?
>Gustavo Seabra <>
>Professor Adjunto, Departamento de Química Fundamental
>Coordenador, CENAPAD-PE
>Universidade Federal de Pernambuco
>Fone: +55-81-2126-7450
>On Wed, Dec 11, 2013 at 10:54 AM, Fengxue Li <> wrote:
>> Hi,all
>> I want to use the AMBER software suite to carry out QM/MM
>> system include 150 atoms in QM part ,When I used Sander order,it spends
>> much time ,I want to how to improve the computational efficiency?at the
>> same time ,I want to know how to add link atoms and the type of link
>> Can anyone help me?
>> Thank you advance!
>> _______________________________________________
>> AMBER mailing list
>AMBER mailing list

AMBER mailing list
Received on Thu Dec 12 2013 - 08:00:02 PST
Custom Search