[gmx-developers] Calling do_steep from within do_md integrator

[Andrew AbiMansour]

Hi Mark,

Thank you for the reply. I decided to just test my own simple SD
integrator. After calling do_force() I update the positions. Example:

int i = 0, dim = 3;
real sd_step_size = 0.01;

for(i = 0; i < mdatoms->nr; i++)
>                  for(d = 0; d < dim; d++) {
>                            state->x[i][d] += sd_step_size * f[i][d];


The code *works* , but the output doesnt make much sense. When I print the
force (f[i][d]), I get a lot of zeros. Am I doing this right? I'm afraid I
might not be accessing the forces (stored locally on each processor)
correctly. Any suggestions?

Thanks!


*Andrew Abi-Mansour*

On Sat, Feb 7, 2015 at 6:54 AM, Mark Abraham <mark.j.abraham at gmail.com>
wrote:

> Hi,
>
> Probably you are over- or under-managing periodic boundaries and/or domain
> decomposition. I would think you should refactor out of do_steep exactly
> the part you need, since it's not designed to be a callable utility
> routine. For example, you probably don't want it to ever call
> em_dd_partition_system()...
>
> Mark
>
> On Sat, Feb 7, 2015 at 7:15 AM, Andrew AbiMansour <anabiman at umail.iu.edu>
> wrote:
>
>> Hi,
>>
>> I would like to call the integrator function do_steep() from within
>> do_md(). I modified some constraints parameters in t_inputrec before
>> passing all the arguments supplied to do_md() by the user to do_steep().
>> This simple approach, however, does not work and GROMACS crashes
>> complaining about a charge group moving too far. The coordinates seem to
>> change by several nano meters over a single SD step. What is this
>> happening? and Do you have any suggestions on how to implement this right?
>>
>> Thanks
>>
>> *Andrew Abi-Mansour*
>>
>> --
>> Gromacs Developers mailing list
>>
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before
>> posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers
>> or send a mail to gmx-developers-request at gromacs.org.
>>
>
>
> --
> Gromacs Developers mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers
> or send a mail to gmx-developers-request at gromacs.org.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20150207/838e4d56/attachment.html>