Dear LAMMPS Users and Developers,
I am trying to perform Hybrid MD + MC for polydisperse system. For this I firstly include a parameter file in which I defined the pair_coeff for each particle. Then in a loop I ran MD simulations for 100 steps and then 100 steps for MC swap, where for the MC swap I used random numbers in the type command which lies in the range of total number of atoms. But if the two random numbers are same then swap will accept with probability 1. So to keep these random numbers different is essential. I am trying different to apply this but I am not able to figure this out.
Also firstly equilibrating the system, I wanted to do production where I want the output according to a particular file. Earlier when I used only MD this can be done easily but not since I am also using MC swap I am encountering problem doing this as well.
I would be grateful if anyone can suggest something about this.
The sample script is a as follows :-
# 3d Lennard-Jones melt
package intel 0
units lj
atom_style atomic
dimension 3
boundary p p p
lattice sc 1.0
region cube block 0 5 0 5 0 5
create_box 125 cube
create_atoms 1 box
mass * 1.0
label loop
variable i loop 125
set atom ${i} type ${i}
next i
jump in.twoatoms loop
velocity all create 0.30 561564 mom yes rot yes
timestep 0.001
#global cutoff
pair_style lj/cut 1.25
include parameter.dat
neighbor 1.60 bin
neigh_modify every 1 delay 0 check yes
variable random1 equal random(1,125,1564848)
variable random2 equal round(v_random1)
variable random3 equal random(1,125,514447)
variable random4 equal round(v_random3)
variable t equal temp
label loopf
variable k loop 1
fix 1 all nvt/intel temp 0.30 0.30 0.1
run 100
unfix 1
label loopswap
variable j loop 100
#if "${random3} >= ${random4}" then "jump in.twoatoms reassign_random"
fix 2 all atom/swap 1 1 29494 0.30 types ${random2} ${random4}
#if "random2 == {random4}" then "fix 2 all atom/swap 1 1 29494 0.30 types ${random2} ${random4}"
run 1
unfix 2
#print "${random2} ${random4}" append info.dat
next j
jump in.twoatoms loopswap
next k
jump in.twoatoms loopf
reset_timestep 0
variable tmsd file sort.txt
variable tms equal next(tmsd)
dump 1 all custom 1 position.txt id type xu yu zu
dump_modify 1 sort id every v_tms #first yes
label loopf2
variable k loop 5
fix 1 all nvt/intel temp 0.30 0.30 0.1
#variable l equal "round((v_k - 1 ) * 100)"
#reset_timestep ${l}
#print "${tms} " file info.dat
run 100
unfix 1
label loopswap2
variable j loop 100
fix 2 all atom/swap 1 1 29494 0.30 types ${random2} ${random4}
run 1
unfix 2
#print "${random2} ${random4}" append info.dat
next j
jump in.twoatoms loopswap2
next k
jump in.twoatoms loopf2
undump 1
Here the “parameter.dat” file includes the parameters and “sort.txt” the timesteps I want my output at.
Best Regards
Puneet
2 Likes
First off, to avoid having the forum software incorrectly format your inputs, you need to quote it into triple backquotes “```” before and after the text block (like a python string block).
There are multiple issues with your random number generation. When defining an equal style variable you effectively are creating something that more resembles a function than a variable. So if you call a function like random() it will generate a new random number every time you evaluate the variable, so it cannot work the way you have set it up. You need to change the way you loop and the way how you define the variables. The following block of input will generate a pair of different random numbers within the given range.
label next_rand
variable type_a equal $(round(random(1,125,12343123)))
variable type_b equal $(round(random(1,125,43234256)))
if "${type_a} == ${type_b}" then "jump SELF next_rand"
Note that by using a $(…) expression the random number is not created when calling the equal style variables but rather whenever the input line is processed.
As for the dump. You are redefining the variable and the dump instance in every loop block. That will start it all over again all the time. Thus you need to define those outside the loop and you must not use the reset_timestep command inside your loop, either.
1 Like
Dear Alex,
Thank you for your help. I will change how the random numbers are defined. As for the dump command, I have tried many different types and also the method as suggested by you but I have still not figured it out. Since the particles positions are incremented only in MD, so I want my positions at the MD steps, but I have to use “run” command in “fix atom/swap” as well, which creates a problem.
I would be grateful if you could show in a block or script format how I can overcome the problem of dumping the positions at MD timesteps that are in a file named “sort.txt”.
Thanks and Regards
Puneet
Using fix atom/swap without also having time integration makes no sense.
Dear Alex,
What I mean to say is that the particle positions are not changed after the application of swap move only there radius changes.
Best Regards
Puneet
That is what I understood. It still does not make sense. You have to orchestrate your swaps differently, either through direct swapping with script commands or by using a modified version of the fix atom/swap that is implementing your swap algorithm. Due to your very unusual use of atom types, the logic implemented in the existing fix is not suitable and your use of the fix is some kind of “abuse” and thus not applicable.
Dear Alex,
Thanks again for your help. I will look into this.
Best Regards
Puneet
Dear Alex,
Can you suggest some correct way I can implement the method that I am trying to do in LAMMPS.
Best Regards
Puneet
Dear Alex,
I want to ask one more question.
The LAMMPS document says for the “run N” command that “A value of N = 0 is acceptable; only the thermodynamics of the system are computed and printed without taking a timestep”, so if I use
fix 2 all atom/swap 1 1 29494 300.0 types 1 2
run 0
unfix 2
will the swap attempt happen, since the atoms are not displaced at all by this command.
Best Regards
Puneet
This is a superfluous question. You can answer this question yourself by monitoring the coordinates.
But as I mentioned before, this kind of use is “abuse” of the fix command which is supposed to be used as part of the run and not hacked to be used like a standalone command and thus very problematic. Whether you do a run 1 or a run 0 does not make a difference in that regard.
Dear Alex,
First of all I want to say that what you are doing is really a great work. The time you put in for the research community to grow, is extraordinary.
Secondly thanks for the idea to check about the “run 0” command.
Thanks and Regards
Puneet
This is how research works! You start with a hypothesis, conduct a suitable experiment, observe carefully, and either confirm your hypothesis, reject it, or construct a more specific experiment to resolve inconclusive observations. It makes no difference whether this applies to physical experiments, computer simulations, or thought experiments.
1 Like
