Question about harmonic constraint in fix colvars

Dear All,

I use the 1_Mar_2016 version on Unix cluster to run steered MD (pulling one I atom away from the surface) using fix colvars command. The I atom is only allowed to freely move along the pulling direction (z axis). The position of I atom is (25.125533, 19.449313, 73.174075). And the total system has been equilibrated already.

group I id 1676

velocity I set 0.0 0.0 NULL

variable x equal (xcm(I,x))
variable y equal (xcm(I,y))
variable z equal (xcm(I,z))

thermo 100
thermo_style custom step v_x v_y v_z
thermo_modify flush yes

fix nvt all nvt temp $T $T 100

fix smd all colvars I.colvars tstat nvt

fix f I setforce 0.0 0.0 NULL

run 50000

Firstly, I only add a harmonic constraint between I atom and a fix point (dummyatom). If I use a smaller forceconstant (20), the job runs but I atom fluctuates severely (z increases from 73 to 112). If I use a larger forceconstant (1000), the job is terminated with the error of atom missing. My question is what reason may cause these problems? The I.colvars file is:

# Steered MD simulation for PMF calculation

colvarsTrajFrequency 1000
colvarsRestartFrequency 1000

colvar {
name I

outputAppliedForce on

distance {
group1 {
atomNumbers 1676
}
group2 {
dummyAtom (25.125533, 19.449313,)
}
}
}

harmonic {
name PMF
colvars I
centers 20
forceConstant 20.0
# targetCenters 0.0
# targetNumSteps 3000

outputEnergy on
outputCenters on
# outputAccumulatedWork on
}

Best,

MC

Dear All,

I use the 1_Mar_2016 version on Unix cluster to run steered MD (pulling
one I atom away from the surface) using fix colvars command. The I atom is
only allowed to freely move along the pulling direction (z axis). The
position of I atom is (25.125533, 19.449313, 73.174075). And the total
system has been equilibrated already.

Firstly, I only add a harmonic constraint between I atom and a fix point
(dummyatom). If I use a smaller forceconstant (20), the job runs but I atom
fluctuates severely (z increases from 73 to 112). If I use a larger
forceconstant (1000), the job is terminated with the error of atom missing.
My question is what reason may cause these problems?

​if you have a soft spring, the atom​ can more around, if you have a stiff
spring, you probably need to reduce the length of tour time step, as steep
potentials require a better resolution in time integration.

axel.