Dear Amber Support Team,
I am currently running Metadynamics simulations using ABMD with wall restraints and have encountered a significant performance issue. My setup is as follows:
I'm using ABMD with FLOODING mode for a COM_DISTANCE collective variable:
codeCopy code
&abmd
mode = 'FLOODING'
timescale = 100.0
monitor_file = 'abmd.dat'
monitor_freq = 500
cv_file = 'cv.dat'
wt_temperature = 10000.0
wt_umbrella_file = 'wt_umbrella.nc'
/
The CV is defined as:
codeCopy code
&colvar
cv_type = 'COM_DISTANCE'
cv_ni = 5590
cv_i = …
cv_min = 20.0
cv_max = 110.0
resolution = 0.10
/
I've implemented wall restraints using a distance restraint file (disang.rst):
codeCopy code
&rst
iat = -1, -1,
igr1 = ...
igr2 = ...
r1 = 0.0,
r2 = 25.0,
r3 = 100.0,
r4 = 110.0,
rk2 = 500.0,
rk3 = 500.0,
&end
After implementing these wall restraints, I've observed a dramatic decrease in performance from 220 ns/day to 72 ns/day.
I'm wondering if there's a more efficient way to implement these boundaries, perhaps similar to PLUMED's uwall and lwall settings where the CVs can be directly used for the restraints?
Any suggestions for improving the performance while maintaining the wall restraints would be greatly appreciated.
Thank you for your time and assistance.
Best regards,
Lyndon LS Meng
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Feb 14 2025 - 03:00:02 PST
