[gmx-users] Error with g_bar

[Jong Wha Lee]

Dear gromax users,

 

I'm trying to perform free energy calculation with Gromacs. I firstly
followed Justin's tutorial
(http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/free
_energy/index.html), which worked well. 

 

However, when I tried to use smaller increments with λ values(the same with
http://dx.doi.org/10.1063/1.1587119), an error was seen with g_bar at the
next step with .xvg files. The error message was:

 

"Some dhdl files contain only one value (dH/dl), while others contain
multiple values (dH/dl and/or Delta H), will not proceed because of possible
inconsistencies."

 

There are bits that give successful g_bar execution, and some that do not.
For example, g_bar works for .xvg files from simulation with init_lambda
value of 0.67, 0.70, 0.73, but not with 0.79. Then I tried to modify some
lines in .xvg files. When I copied 0.73.xvg file and renamed it into
0.79.xvg, and changed the subtitle and legend lines in the .xvg file
accordingly, g_bar failed. On the other hand, When I renamed and modified
subtitle & legend so the file to become 0.50.xvg, g_bar worked! 

 

So what I have done was simply to change the 0.73 values below (from .xvg
files) into 0.45, 0.50, 0.72, 0.79, or 0.88. g_bar works with 0.45, 0.50, or
0.72, but not with 0.79 or 0.88.

 

@ subtitle "T = 298 (K), \xl\f{} = 0.73"

@ s0 legend "dH/d\xl\f{} \xl\f{} 0.73"

 

s1 legend and s2 legend do not affect whether above error shows up or not,
although warning shows up for insensible values.

 

I have checked the differences between the input files with gmxcheck, and
the only differences I could see was nkx, nky, and nkz values. These values
were 28 for some, and 25 for the others. Maybe this is the source of error?
g_bar runs well with .xvg files from simulations with identical nkx, nky,
nkz values. But, .xvg files do not contain the information on these values
so I cannot imagine how the differences in nkx,nky,nkz values affect how
g_bar runs.

Can anyone advise me on resolving this error?

 

Many thanks in advance,

 

Jong Wha