Dear Amber community,
I want to simulate a micelle system in explicit water using the TIP4P-EW
water model. After building and solvating the system in leap I wanted to
run energy minimization with the XMIN/L-BFGS method (following the
workflow of a publication). However, I observe that the virtual site of
the water model seems to be distorted and the resulting water box looks
wrong. The v-site drifts off, while the oxygen and hydrogens keep their
usual geometry. This happens after merely 100 steps of L-BFGS. Next, I
tried the OPC model, but the same thing happens.
I tried using sander.MPI on amber-22 and also on AmberTools25. After
minimization i convert the .crd to pdb using ambpdb with the flags -aatm
and -ep.
My minimization input:
&cntrl
nmropt=0,
imin=1, ! energy minimization mode
ntmin=3, ! XMIN minimizer (default LBFGS)
maxcyc=100, ! max minimization steps
ntb=1, ! periodic boundary, constant volume
ntp=0, ! no pressure scaling during minimization
cut=12.0, ! PME direct-space cutoff (Å)
ntc=1, ! NO SHAKE (must be off for minimization)
ntf=1, ! calculate all forces including H bonds
ntr=0, ! turn off positional restraints
/
Further I noticed that the usual steepest descent/conjugate gradient
minimization works fine with all water models. Of course I could just
stick to this, but I wanted to ask if anyone encountered similar
behavior with XMIN/L-BFGS and 4-point water models? Are there known
limitations of XMIN with extra-point waters? Is my minimization input
fine? Is it possibly a problem relating to SHAKE?
I hope to get a better understanding of the underlying problem.
Thank you very much,
Jefta Wucherpfennig
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jun 02 2026 - 01:00:02 PDT
