Frequently Asked Questions¶
There are no reactions¶
Increase the temperature
NanoReactor: Increase
DiffusionTime
, setMinVolumeFraction
to a smaller value, setInitialRadius
to a smaller valueLatticeDeformation: Set
MinVolumeFraction
to a smaller value, decreasePeriod
Set
MolecularDynamics UseDeuterium=No
Ensure that you use a reactive potential (e.g. ReaxFF, DFTB, MLPotential)
Note: The best value for MinVolumeFraction
depends on the initial density
(MolecularDynamis%BuildSystem%Density
). If you decrease the initial
density, you may need to also decrease the MinVolumeFraction
.
Tip
Run the simulation with the fast UFF force field to get a feeling for if the initial settings are reasonable. With UFF no reactions will happen but you can still see how the density fluctuates in the MD simulation.
There are too many reactions¶
Decrease the temperature
NanoReactor: Decrease
DiffusionTime
, setMinVolumeFraction
to a larger value, setInitialRadius
to a larger valueLatticeDeformation: Set
MinVolumeFraction
to a larger value, increasePeriod
Set
MolecularDynamics UseDeuterium=Yes
Note: The best value for MinVolumeFraction
depends on the initial density
(MolecularDynamis%BuildSystem%Density
). If you decrease the initial
density, you may need to also decrease the MinVolumeFraction
.
The MD simulations are too slow¶
Decrease the number of atoms
Decrease the number of NanoReactor or LatticeDeformation cycles
Increase the MD time step
How should I set the density and compression factor?¶
Nanoreactor: The density should be approximately the normal liquid density of your system, with a compression factor of about 0.5-0.7
Lattice deformation: The density should be about half the normal liquid density of your system, with a compression factor of about 0.15-0.30
The simulation explodes¶
Follow the steps for There are too many reactions.
Decrease the MD time step
How do I use computing resources efficiently?¶
The steps are parallelized as follows:
MolecularDynamics runs as many jobs in parallel as possible, respecting the allocation (
NSCM
)NetworkExtraction runs
NumSimulations
ChemTraYzer2 jobs sequentially in serial; thenNSCM
geometry optimizations and single points are run in parallel.ProductRanking runs in serial
Thus, most of the computational steps except ProductRanking are run efficiently
in parallel. You may thus choose to set ProductRanking Enabled=False
if you
have a large node allocated, and then restart from the previous results using
ProductRanking Enabled=True
on a smaller allocation.