Hi Scott,
I think that the implementation of PIMD within LES is fine. There are
test cases for this in test/PIMD (all those with "part_*"). We can check
it more carefully at the meeting.
On the other hand, I do not know anything about LES/NEB.
Thanks,
Francesco
Scott Brozell wrote:
> Hi,
>
> Below is the nebmodule.f cvs log.
>
> Scott
>
> TEMPORARY PATCH to get sander.MPI to compile.
> See below for the compiler error.
> This needs to be investigated by the NEB and PIMD developers in
> conjunction with a thorough code inspection of LES/PIMD/NEB !!
> This inspection should consider all sanders and the results
> should be communicated to all developers !
> See the amber-developers list messages that contain neb in the subject
> for existing and potential issues.
> Here is another question:
> Why is there so much code duplication between modules pimd_force
> and nebmodule ?
>
> With configure -lam g95: mpif77 -c -O3 -fno-second-underscore -march=nocona -ffree-form -o force.o
> _force.f
> In file _force.f:2041
> call mpi_reduce(nebrms,etmp,1,MPI_DOUBLE_PRECISION, &
> 1
> Error: Symbol 'nebrms' at (1) is defined in multiple modules
>
>
> On Fri, 19 Oct 2007, Scott Brozell wrote:
>
>
>> On Fri, 19 Oct 2007, David A. Case wrote:
>>
>>
>>> On Thu, Oct 18, 2007, Carlos Simmerling wrote:
>>>
>>>
>>>> I've also sent out requests for info on nebrms, several times, with no reply.
>>>> I'm tempted to remove all of this code if nobody owns it or knows why
>>>> it is there.
>>>>
>>> It would clearly be very useful if we could get at least a temporary fix
>>> in *today*, so that the code can compile. Unfortunately, I am tied up and
>>> cannot volunteer.
>>>
>> Based on Dave Mathews reply, a simple patch exists.
>> I'll commit it, but we have to remember that this is only
>> a temporary fix.
>>
>> Scott
>>
>> On Thu, 18 Oct 2007, David Mathews wrote:
>>
>>
>>> I originally added nebrms to calcuate the RMS for energy
>>> minimization during neb calculations. There are reduced degrees of freedom
>>> because the atoms in the first and last image are frozen. This was a
>>> report item that was helpful for debugging with small systems. For large
>>> systems, it did not seem particularly important. That means that it can
>>> now be removed unless somebody objects. I can probably help with that
>>> cleanup next week. Hope this helps.
>>>
>>
>
>
>
Received on Sun Oct 21 2007 - 06:07:52 PDT
