Discrepancy in CFF parameters between EMC versions

Dear:
I use the PCFF potential to create polymer molecules. Polymers contain Si, O, C, H. The molecules generated in the new version don’t include # Improper Coeffs and # AngleAngle Coeffs. The old version does not have this issue.

Dear user,

I am not aware of this problem and have no example for which this occurs. Regrettably, debugging will not be possible without you providing an explicit example.

This is my input and my output. Sorry, new users can not upload attachments.

ITEM	OPTIONS

replace		true
field		pcff
number		false
ntotal		5200
density		1

ITEM	END

# Groups

ITEM	GROUPS

SiCCO *[Si]()(C)(C)O*,1,SiCCO:2 #left and right stars are hidend,I don't know why the star on the left is hidden when displayed. I hope you understand.
SiCCC *[Si]()(C)(C)(C),1,SiCCO:2 #left star is hidend
CSiCCO C[Si]()(C)(C)O*,1,SiCCO:1#right star is hidend

ITEM	END

# Clusters

ITEM	CLUSTERS

polymer		alternate,1

ITEM	END

# Polymers

ITEM	POLYMERS

polymer
10		SiCCO,49,SiCCC,1,CSiCCO,1

ITEM	END

Difference:

New version output in params file: only the title

# Improper Coeffs
# AngleAngle Coeffs

Old version ouput in params file:

# AngleAngle Coeffs
improper_coeff	1 aa   -0.31570   -0.31570   -0.31570  107.66000  107.66000  107.66000  # c,hc,hc,hc

Dear user,

Thank you for your input. I actually went ahead and used insight2lammps.pl as provided by EMC in the ${EMC_ROOT}/scripts directory. This script converts InsightII input in the form of .car and .mdf – in combination with a ClassII force field such as PCFF – to produce a valid LAMMPS .data coordinate file, which includes the force field parameters.

I used the following setup.esh to create one toluene molecule with the PCFF force field, which serves a as a good proxy for your observed issue:

# Options section

ITEM	OPTIONS

replace		true
field		pcff
density		0.1
number		true
focus		true
insight		true
emc_execute	true

ITEM	END	# OPTIONS

# Shorthand section

ITEM	SHORTHAND

molecule	Cc1ccccc1,1

ITEM	END	# SHORTHAND

The insight true option creates setup.car and setup.mdf which is needed by insight2lammps.pl. The latter was called with

insight2lammps.pl -forcefield=pcff -base=tmp setup

to create tmp.data, which now serves as a difference source for obtaining ClassII force field constants and can be used to compare to setup.params. I noticed that not only the above mentioned line was missing, but also a second line:

# AngleAngle Coeffs

improper_coeff	1 aa    2.37940    3.01180    2.37940  111.00000  111.00000  107.66000  # c,cp,hc,hc
improper_coeff	2 aa   -0.31570   -0.31570   -0.31570  107.66000  107.66000  107.66000  # c,hc,hc,hc

EMC was adapted to include dummy entries for improper contributions:

# Improper Coeffs

improper_coeff	1    0.00000    0.00000  # c,cp,hc,hc
improper_coeff	2    0.00000    0.00000  # c,hc,hc,hc

I recompiled EMC for all targets and uploaded these corrected versions to https://sourceforge.net/projects/montecarlo/files/.