N2P2/Lammps interface for performing 4G-HDNNP MD simulations.

[ANCHENMA]

The question
I wonder if I can run a simulation in lammps interface with 4G-HDNNP obtained by RuNNer.

What did you already try?
By using RuNNer, I have finished the 4G-HDNNP training which included the long-range electrostatic interactions.The following data are obtained:
weights.008.data
weights.001.data
scaling.data
weightse.008.data
weightse.001.data
hardness.008.data
hardness.001.data
I have seen issue #122, #155 and #157. I wonder if I can run a simulation in lammps interface with 4G-HDNNP, Because I have tried by using the /examples/interface-LAMMPS/H2O_RPBE-D3 and I modifed some items in md.lmp , but there were some errors, such as
WARNING: Unknown keyword "symfunction H 2 H 0.000000 0.000000 12.000000 "
WARNING: Unknown keyword “global_hidden_layers_electrostatic 2”
WARNING: Unknown keyword “global_nodes_electrostatic 15 15”
WARNING: Unknown keyword “global_activation_electrostatic t t l”

If I comment out all the warning, the simulation will work, but how can I take out the charge of each atom in the configuration? I would be very grateful indeed for any help you could give me.

Background
H2O system.

[singraber]

I am sorry, but work on the 4G-HDNNP prediction with the LAMMPS interface is not yet finished. There will be a new n2p2 version released when the corresponding branches are merged and everything has been tested. This is not the case yet, there are still bugs and hence it is not recommended to do application work with the current preliminary code developments.

[ANCHENMA]

Noted with thanks.
Best wishes, ANCHEN