AMBER Archive (2006)

Subject: AMBER: energy minimizations in amber8 vs. amber9

From: Seongeun Yang (seongeun_at_korea.ac.kr)
Date: Tue Nov 28 2006 - 01:43:35 CST

Hello all,
I found some inconsistencies between the minimization results with amber8 and amber9.
The system under study is a peptide solution in which a 21-aa peptide with end cappings is solvated by 3 Cl- ions and 5980 TIP3P waters.
The topology files were generated by 2 different force fields, ff99 and ff02pol.r1, both implemented in amber9.
Except for where the executable 'sander' came from, the other conditions were the same for all test minimizations.
All the tests were done on Itanium2.

With the following minimization options,
 &cntrl
   imin = 1, maxcyc = 1000,
   dx0 = 0.01, drms = 0.0001,
   ntpr = 10,
 &end
'sander in amber8' does minimization successfully, but 'sander in amber9' fails with 'LINMIN FAILURE'.

At the end of minimizations, the mdout looks like as follows in each case.
The same behavior was seen both with ff99 and ff02pol.r1.

----------------------------
 mdout by amber8/exe/sander
----------------------------
   NSTEP ENERGY RMS GMAX NAME NUMBER
   1000 -7.9923E+04 5.2892E-01 1.5817E+01 O 10830

 BOND = 11528.3911 ANGLE = 19.3555 DIHED = 181.1588
 VDWAALS = 19984.5427 EEL = -112446.2299 HBOND = 0.0000
 1-4 VDW = 44.6389 1-4 EEL = 765.3724 RESTRAINT = 0.0000

  Maximum number of minimization cycles reached.

----------------------------
 mdout by amber9/exe/sander
----------------------------
   NSTEP ENERGY RMS GMAX NAME NUMBER
     30 NaN NaN 0.0000E+00 HH31 1

 BOND = NaN ANGLE = NaN DIHED = NaN
 VDWAALS = -398.3763 EEL = NaN HBOND = 0.0000
 1-4 VDW = NaN 1-4 EEL = NaN RESTRAINT = 0.0000

     .... RESTARTED DUE TO LINMIN FAILURE ...

     ***** REPEATED LINMIN FAILURE *****

Please let me know why the two results are totally different.
Thanks for your reply in advance.

Seongeun Yang
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu