LAMMPS Crash - RMD with Reax on Polymer System

Hi

I am attempting to run RMD with Reax in LAMMPS 64-bit 20140519 on a phenolic polymer system. I have successfully run an initial 5ps and would like to run a further 5ps after the removal of some key products. When I attempt to run this second 5ps on the new model the LAMMPS runs starts giving the initial starting values with the Pressure value giving the error -1.#IND before giving a windows notifications that lmp_serial.exe has stopped working.

I have tried decreasing the time step and this does help to a degree however I am still unable to successfully run the full model. Visualisation of the model does show a number of radical species from the initial 5ps output, could this be part of the problem?

Below is a copy of my input script and I would be more than happy to provide the rest of the required control, parameter and data files.

units real

atom_style charge

read_data 33pc_1000atm_5ps.LMP

pair_style reax/c lmp_control

pair_coeff * * ffield.reax.cho H C O

neighbour 2 bin

neigh_modify every 10 delay 0 check no

fix 1 all qeq/reax 1 0.0 10.0 1e-6 param.qeq

fix 2 all nvt temp 1250.0 1250.0 100.0

fix 3 all reax/c/bonds 1000 molfra_5ps.out

timestep 0.5

dump 1 all xyz 1000 dumpfile_5ps

run 10000

Please let me know if you have any ideas as to the problem or if you need more information.

Kind Regards,

Grace

By removing atoms you have likely created a system very far from equilibrium. You probably need to re-equilibrate the new system first (minimizations, low-temperature dynamics). 0.5 fs may still be too large as well.

-Ben

Hi,

24.09.2014, 17:45, "[email protected]" <[email protected]>:

Hi

I am attempting to run RMD with Reax in LAMMPS 64-bit 20140519 on a phenolic polymer system. I have successfully run an initial 5ps and would like to run a further 5ps after the removal of some key products. When I attempt to run this second 5ps on the new model the LAMMPS runs starts giving the initial starting values with the Pressure value giving the error -1.#IND before giving a windows notifications that lmp_serial.exe has stopped working.

I have tried decreasing the time step and this does help to a degree however I am still unable to successfully run the full model. Visualisation of the model does show a number of radical species from the initial 5ps output, could this be part of the problem?

Below is a copy of my input script and I would be more than happy to provide the rest of the required control, parameter and data files.

Why not make yourself more than happy? :slight_smile:

Oleg

Hi Ben, Oleg

Sorry for taking so long to respond I had a fairly major computer failure which has delayed my response. Thank you very much for your help. I have tried minimising the system but there is no change in final energy from the initial value and when I attempt a low temperature MD run LAMMPS once again crashes. When I start both the minimisation and MD runs the pressure value is again given as -1.#IND presumably this is affecting my ability to equilibrate the system. I can successfully run a large portion of the system with product removal but not the full model.

Unfortunately some commercial sensitivity issues have come up and I will no longer be able to send on the data and parameter files, Sorry!

Thank you again for your help.

Grace

Hi Ben, Oleg

Sorry for taking so long to respond I had a fairly major computer failure which has delayed my response. Thank you very much for your help. I have tried minimising the system but there is no change in final energy from the initial value and when I attempt a low temperature MD run LAMMPS once again crashes. When I start both the minimisation and MD runs the pressure value is again given as -1.#IND presumably this is affecting my ability to equilibrate the system. I can successfully run a large portion of the system with product removal but not the full model.

Unfortunately some commercial sensitivity issues have come up and I will no longer be able to send on the data and parameter files, Sorry!

Thank you again for your help.

Grace

In addition to visualizing the trajectories near when the system crashes as Oleg suggests, you can use a dissipative thermostat like fix langevin and a short coupling coefficient.

-Ben