Problem with charged system

Dear Developers,

Hope this is the right place i am posting on.

I just started using Cassandra and trying to setup a simple CO2(TraPPE model) NPT simulation. However, when i run the input file with "Charge_Style" set to Ewald, i get an error saying "segmentation fault (core dumped)" . Surprisingly , when i set the Charge_Style to NONE, the code ran successfully. I even tried to run the SPC water model in the example directory, i got the same error(also solved with charge style of NONE).
Because i am running cassandra on multiple cores (following the instructions in the README file to compile it), i suspect the problem is with dividing the workload when the Ewald sum is on.

Wondering if the source of error could be attributed to the incorrect compilation?

I would appreciate if you ould help me on this.
Here is the required files.

Regards,
Javad

Dear Javad,

Could you let us know the type of compiler you used? Do you get the same error in the debug mode too - compiling with Makefile?

Also, if you can send us the input file for water, we will try to replicate the issue you are having.

Jindal

Dear Jindal,

I have not tried the debug mode,and i used the gfortran compiler on a linux machine.
It works fine for constant volume simulations of uncharged systems, but it crashes for cases in which a computationally expensive task such as calculation of Ewald sum or volume attempt is being performed (so, the problem is connected with parallelization scheme).

By the way, the input files for SPC water was the original files in the example directory of the documentation. The attached files are the inputs i used for running TraPEE CO2 and you might want to run it on your machine.

Javad

!**********************************************************************************
!Molecular connectivity file for CO2.pdb
!**********************************************************************************

# Atom_Info
3
1 C1_s1_s1 C 12.0107 0.7 LJ 7 2.8
2 O2_s1_s1 O 15.9994 -0.35 LJ 79 3.05
3 O3_s1_s1 O 15.9994 -0.35 LJ 79 3.05

# Bond_Info
2
1 1 3 fixed 1.16
2 1 2 fixed 1.16

# Angle_Info
1
1 3 1 2 fixed 180

# Dihedral_Info
0

# Fragment_Info
1
1 3 1 3 2

# Improper_Info
0

# Fragment_Connectivity
0


END

---

!*******************************************************************************
! This is an input file for CO2 .
!*******************************************************************************
# Run_Name
TraPPE_CO2

# Sim_Type
NPT_MC

# Nbr_Species
1

# VDW_Style
LJ cut_tail 12.0

# Charge_Style
NONE

# Intra_Scaling
0.0 0.0 0.0 1.0
0.0 0.0 0.0 1.0

# Mixing_Rule
LB

# Seed_Info
80411904 80411904

# Rcutoff_Low
1.0

# Pair_Energy
TRUE

# Molecule_Files
CO2.mcf 100

# Box_Info
1
CUBIC
30.00 30.00 30.00

# Temperature_Info
336.0

# Pressure_Info
100.0

# Move_Probability_Info

# Prob_Translation
0.55
1.0

# Prob_Rotation
0.2
120.0

# Prob_Regrowth
0.245
1.0

# Prob_Volume
0.005
200.0

# Done_Probability_Info

# Start_Type
make_config
100

# Run_Type
Equilibration 100 10

# Average_Info ( 0 == yes, 1 = No)
1

# Frequency_Info
freq_type none
Nthermofreq 1000
Ncoordfreq 10000
MCsteps 1000000
# Done_Frequency_Info

# Property_Info 1 0
Energy_Total
Nmols
Density
Volume
Pressure

# Fragment_Files
species1/frag1/frag1.dat 1

# Bond_Prob_Cutoff
1.0E-10

END

---

Javad,

The water SPC simulation input file in the Examples directory runs without any error for me. So we need to collect more information to locate the part of the code that's throwing the error. As I suggested, please run it in the debug mode and send us the detailed error message, if any, you see.

Jindal

Jindal,
You were right, i think the problem was with my PC. I managed to run Cassandra on a different machine , it works fine for SPC model.

I was going to use Cassandra for simulation of drug-like molecules, usually contain multiple rings , in an Osmotic Ensemble,
but i noticed that Cassandra does not support double rings at the moment.

Also, the web page claims that Cassandra is able to simulate Osmotic Ensemble(OS) , and the tutorial also mentions fugacity in a table listing cassandra input units, but i do not see any input key for fugacity to simulate OS.

Wondering if by OS , you mean two box-two pressure simulation in NPT variant of Gibbs Ensemble.
Because Cassandra is able to assign different pressure to each box , this way the pressure of one of the boxes can be set on the equivalent fugacity to simulate Osmotic Ensemble.

Javad