Chemical bonds break

Dear LAMMPS developers,
A problem was encountered during MD. That chemical bonds of some water molecules broke puzzled me a lot. The attached file is the input script. Longing for your instructions. Thanks for your concern.
sincerely

in.ch4 (1.2 KB)

2011/5/14 tlx0509236.happy <[email protected]>:

Dear LAMMPS developers,
A problem was encountered during MD. That chemical bonds of some water
molecules broke puzzled me a lot. The attached file is the input script.
Longing for your instructions. Thanks for your concern.

how do you determine that bonds are broken?
it has to be an artifact of your visualization.

axel.

It seems nothing wrong with the visualization software. The accessory is one of my output files, from which the structure can be seen in VMD. Would you please have a look at the structure? Gratitude for your help.


[result10.xyz|attachment](upload://p3h44bRZaJgSKwtkRa3THmwSimC.xyz) (1.49 MB)

2011/5/15 tlx0509236.happy <[email protected]>:

It seems nothing wrong with the visualization software. The accessory is one
of my output files, from which the structure can be seen in VMD. Would you
please have a look at the structure? Gratitude for your help.

i didn't expect the visualization _software_ to be broken,
but how you _interpret_ what you see.

it has been explained here _many_ times, that an .xyz format
file is the lowest common denominator format and thus for many
reasons an inferior choice for visualizing MD data in VMD.

furthermore, it has also been repeatedly discussed, that LAMMPS
internally doesn't identify atoms with element, but simply with
numbers, so even more data is lost, which makes it impossible
for VMD to recover this information, unless you provide it through
other means.

indeed what is see when visualizing the results from your file,
all i see are bonds that are obviously broken due to PBC,
and atoms that are not properly visualized due to the way
how .xyz files are generated in LAMMPS, i.e. in lieu of any
alternate information, LAMMPS uses the atom type number
as element symbol, which leads VMD to believe that you
have a system consisting of hydrogen, helium and lithium atoms.

GIGO.

axel.


2011/5/15 tlx0509236.happy <[email protected]…127…>: >> It seems nothing wrong with the visualization software. The accessory is one >> of my output files, from which the structure can be seen in VMD. Would you >> please have a look at the structure? Gratitude for your help. > >i didn’t expect the visualization software to be broken, >but how you interpret what you see.> >it has been explained here many times, that an .xyz format >file is the lowest common denominator format and thus for many >reasons an inferior choice for visualizing MD data in VMD. > >furthermore, it has also been repeatedly discussed, that LAMMPS >internally doesn’t identify atoms with element, but simply with >numbers, so even more data is lost, which makes it impossible >for VMD to recover this information, unless you provide it through >other means.> >indeed what is see when visualizing the results from your file, >all i see are bonds that are obviously broken due to PBC,

Thanks for your patience, and I’m going to bother you again. I also calculated distances between O and H within the same water molecule, and thus to determine whether certain chemical bonds broke or not. Chemical bonds can be supposed to be broken, when the calculated value of O-H was obviously much larger than the equilibrium value 0.9572. The periodic boundary condition had been already taken into consideration, before O-H distances were determined. The attached files, b_902000 and b_904000, includes the infomation of abnormal O-H distances at timestep 902000 and 904000, respectively.

I found from the two files that some previously broken bonds were restored to its normal value around 0.9572 after 2000 timesteps, while some new O-H bonds broke, and that the broken bonds O-H distances seemed to be greater with time evolution.

So, what i concern about are the following: 1. Will the problem listed above seriously affected the MD ? 2. Does the inappropriate input script attached previously cause the above problem, and how can i avoid it ? Thank you.

I found

b_902000.txt (779 Bytes)

b_904000.txt (608 Bytes)

2011/5/15 tlx0509236.happy <[email protected]>:

Thanks for your patience, and I'm going to bother you again. I also
calculated distances between O and H within the same water molecule, and
thus to determine whether certain chemical bonds broke or not. Chemical
bonds can be supposed to be broken, when the calculated value of O-H was
obviously much larger than the equilibrium value 0.9572. The
periodic boundary condition had been already taken into consideration,
before O-H distances were determined. The attached files, b_902000 and
b_904000, includes the infomation of abnormal O-H distances at timestep
902000 and 904000, respectively.

I found from the two files that some previously broken bonds were restored
to its normal value around 0.9572 after 2000 timesteps, while some new O-H
bonds broke, and that the broken bonds O-H distances seemed to be greater
with time evolution.

this is nonsense. your bond distance calculation is wrong.
people have used tip4p water a lot with LAMMPS. why should
it through some magical event suddenly be broken (and it didn't
look broken to me) only for you?

So, what i concern about are the following: 1. Will the problem listed above
seriously affected the MD ? 2. Does the inappropriate input script

ad 1) this is a really stupid question.
either you find proper proof that something is incorrect
and then, of course, it will affect the MD. or you confirm
my claim, that the code is correct, but that you are making
mistakes when calculating the bonds. what would you write
in your report/paper? "i did not check the simulation, because
axel kohlmeyer said it was not needed"???

attached previously cause the above problem, and how can i avoid it ? Thank
you.

ad 2) the problems are all in your head. the only way to avoid them is
to stop trying to do MD without knowing what you are doing. PEBCAC!

pick a simpler test system and debug it and find you what is wrong
(with your thinking).

axel.

Thanks for your patience, and I'm going to bother you again. I also
calculated distances between O and H within the same water molecule, and
thus to determine whether certain chemical bonds broke or not. Chemical
bonds can be supposed to be broken, when the calculated value of O-H was
obviously much larger than the equilibrium value 0.9572. The

chemical bonds _cannot_ be broken if you use harmonic potentials or shake.

axel.

Are you calculating the bond distance correctly
if it crosses thru a periodic boundary?

Steve

2011/5/15 tlx0509236.happy <[email protected]>: