[gmx-users] glibc error with g_density (only for some cases)

[Mark Abraham]

Arun kumar wrote:
> Dear gmx-users,
>                        I want to discuss and understand about an
> error I am getting only at times.... I am simulating a DPPC double
> bilayer phase(got the coordinates from Dr Tieleman's website). I used
> genbox  -cs to get double bilayer. Then i started simulations
> normally.
> 
> When I use   g_density -b 500  I get the following error
> 
> *** glibc detected *** free(): invalid next size (normal):
> 0x0000000000713260 *** Aborted
> 
> It is giving a empty density.xvg file
> 
> I get same error if I use g_density -b 1000
> I checked the reproducibility of the error...... I am getting it
> always(I mean I am using g_density once again again,,,,, Not
> simulating the system once again again)
> 
> But when I use g_density -b 1500 or more I am getting densities 
> normally.....
> Actually  i am doing simulations at 4 different temperatures..... But
> I am getting the error only for three temperatures.... For one
> temperature Every thing is fine......
> Gromacs version I am using is : 3.3.1
> I checked the archives and in google.... I found a similar error in
> archives for make_ndx
> 
> After checking in google I came to know that its a common bug with
> glibc package of different versions.....  Now i want to know is there
> anything wrong in my simulations(possibilities) and also how I can rid
> of these errors.........

You stand a much better chance of getting a response if you make your 
requests easy to read. That means using conventional English 
punctuation, ie. not ending most sentences with an ellipsis. Your use of 
ellipses in the above text makes it read like it is a stream of 
consciousness that you haven't thought through clearly. That makes 
people not want to do your work - thinking clearly about your problem 
and describing it well - and thus never getting around to giving you any 
insight.

I have never used g_density, but I had look at the man page for 30 
seconds and I can think of two reasons why doing an analysis starting at 
point c would work when it doesn't work starting at a or b, when a<b<c. 
Since it only happens for one of your 4 near-identical simulations, this 
suggests one of my reasons isn't right... and so I think there is a 
problem with your output between b and c for the one that breaks. Go and 
look at the log file and see what went wrong.

I didn't apply much knowledge here, I just thought about the problem 
clearly. Perhaps this would have occurred to you if you'd thought about 
things this way before posting :-)

Mark