[gmx-users] problem with git/4.6

[Justin A. Lemkul]

On 6/29/12 11:08 AM, Michael Brunsteiner wrote:
>
>
>> You should have it.  In CMakeLists.txt, PROJECT_VERSION should be set to
>> "4.6-dev" so you can check that.
>
> ithat what i looks like ... i now get:
>
> prompt> mdrun_d
> [...]
> :-)  VERSION 4.6-dev-20120629-9c6be1c  (-:
> [...]
>
> which gives me:
> g_bar_d -f mdv*.xvg -b 100
> [...]
> total   0.000 -  1.000,   DG -9.00 +/-  0.15
>
> with vanilla 4.5.5 i get:
> g_bar_d -f mdv*.xvg -b 100
> [...]
> total   0.000 -  1.000,   DG -9.00 +/-  0.15
>
> there's two points:
>
> 1) this value is fine, as, once combined with the coulomb part, it is very close to literature and
> exptl data. however, in both cases  get this warning ...  Second Law of Thermodynamics etc...
> and riduculously high s_B values at lambda=0.
>

This indeed sounds like a bug worth fixing since it is present in the 
release-4-6 branch.

> 2) the reason i saw different DeltaG values earlier was that i had included (by mistake)
> the dhdl file from a single additional window at lambda=0 that was produced with exactly the same
> input apart from the value of 0.3 (instead of 0.5) for sc_sigma ... as far as i understand things (not very
> well perhaps) the resulting free energy difference should NOT depend on this parameter ... so neither should
> the result of g_bar ... but maybe it does when using foreign lambdas as i did here ? if this is so
> it might be a good idea making g_bar read tpr files to give a warning in such a case ...
>

Whether or not sc_sigma has an effect depends on the topology.  If you change 
the value, you can change the result of the calculation.  g_bar has no need to 
read .tpr files, so I doubt it would be very useful to include them as an 
additional input.

> i guess for now i'll wait for a stable 4.6 - any ideas when this will be there?
>

No clue.  Maybe a few weeks, maybe not.  There are several very large changes 
being put in place.  But note that if you're getting the same error with the 
development version of the free energy code, you're likely to see the exact same 
thing in the 4.6 release.  The free energy code merge is considered done and 
ready for production.  If this is not the case, a bug report needs to be filed 
ASAP.  Please submit a report at redmine.gromacs.org with a minimal test case.

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================