Re: System minimization: fail after 199 steps

From: Brian Bennion (brian_at_youkai.llnl.gov)
Date: Tue Nov 02 2004 - 13:09:32 CST

hello Charles,

The real problem is in the implementation of the new load balance
structure and I am clueless in this aspect.

However, negative signs are good when dealing with vdw and electrostatic
potential energies.
The fact that you have them now is a good thing.

I forgot to ask what type of machine/cluster you are using and the OS.
I have never seen the Rtask error before.

What is the size of your system? What ensemble are you using? I noticed
that no box dimensions are given in your input file. The bond and angle
energies appear to high for most large simulations.

Have you tried to minimize small segments of your system first, before
combining them altogether for general minimization?

Finally, did you compile this version of namd/charm or get a binary? The
core file might also be useful as you get a handle on the problem.

Regards
Brian
On Tue, 2 Nov 2004, Charles Danko wrote:

> Hi,
>
> Thanks to Brian for the help. I tried changing the interactions term
> to scaled1-4 with the default scaling factor or 1.0 but the
> minimization still died after 199 steps. The last couple steps are
> pasted below. The energy is substantially different; I am not sure
> that I understand the negative sign.
>
> If there are any more suggestions I would be grateful.
>

> Thanks again,
> Charles
>
> BRACKET: 2.50603e-07 0.090796 -114345 -267.772 826360
> NEW SEARCH DIRECTION
> INITIAL STEP: 6.4e-05
> GRADIENT TOLERANCE: 14979.9
> ENERGY: 195 14682.3820 22632.4249 7136.2497 282.3447
> -96578.5816 9615.8599 0.0000 0.0000 0.0000
> -42229.3205 0.0000 -42229.3205 -42229.3205 0.0000
>
> ENERGY: 196 14352.5204 22506.1721 7126.8372 277.9901
> -96822.7235 9098.0523 0.0000 0.0000 0.0000
> -43461.1514 0.0000 -43461.1514 -43461.1514 0.0000
>
> ENERGY: 197 14345.7336 22383.6335 7109.8393 271.1399
> -97309.3368 8429.8700 0.0000 0.0000 0.0000
> -44769.1205 0.0000 -44769.1205 -44769.1205 0.0000
>
> ENERGY: 198 16752.9576 22839.6644 7095.3013 264.8575
> -98275.7787 9208.5645 0.0000 0.0000 0.0000
> -42114.4333 0.0000 -42114.4333 -42114.4333 0.0000
>
> BRACKET: 0.000123859 2654.69 -4.84743e+07 -1.49018e+07 9.67884e+07
> ENERGY: 199 14522.5983 22380.1473 7106.8166 269.4865
> -97462.9771 8297.1243 0.0000 0.0000 0.0000
> -44886.8042 0.0000 -44886.8042 -44886.8042 0.0000
>
> LDB: LOAD: AVG 121.372 MAX 150.69 MSGS: TOTAL 171 MAXC 36 MAXP 4 None
> LDB: LOAD: AVG 121.372 MAX 145.646 MSGS: TOTAL 310 MAXC 88 MAXP 4 Alg7
> LDB: LOAD: AVG 121.372 MAX 123.798 MSGS: TOTAL 310 MAXC 88 MAXP 4 Alg7
> Rtasks fail:
> Rtask(s) 1 : exited with signal <11>
> Rtask(s) 4 2 3 7 5 6 8 10 9 : exited with signal <15>
> Rtask(s) 1 : coredump
>
>
>
>
> On Mon, 1 Nov 2004 15:56:32 -0800 (PST), Brian Bennion
> <brian_at_youkai.llnl.gov> wrote:
> > HI Charles,
> >
> > Is there a particular reason you use a 1-2 exclusion instead of a 1-4
> > exclusion?
> >
> > your vdw energies are not at all good and I think it is because your
> > including too many interactions with the 1-2 based exclusion.
> > Just for kicks, try using the 1-4 term and see how things go.
> >
> > Regards
> > Brian
> >
> > On Mon, 1 Nov 2004, Charles Danko wrote:
> >
> > > Hi,
> > >
> > > I am having a problem minimizing my system. The end of the
> > > minimization output file is as follows:
> > >
> > > ENERGY: 196 72896.2480 36970.3755 7529.1929 527.0135
> > > 302178.7322 230719.2954 0.0000 0.0000 0.0000
> > > 650820.8576 0.0000 650820.8576 650820.8576 0.0000
> > >
> > > ENERGY: 197 74621.1928 37566.9383 7549.9188 527.7283
> > > 301273.8956 231103.4061 0.0000 0.0000 0.0000
> > > 652643.0800 0.0000 652643.0800 652643.0800 0.0000
> > >
> > > BRACKET: 0.000110482 1822.22 -4.75282e+07 -1.29993e+07 6.80802e+07
> > > ENERGY: 198 73119.1726 37051.1602 7532.4822 526.9988
> > > 302022.8627 230476.5677 0.0000 0.0000 0.0000
> > > 650729.2442 0.0000 650729.2442 650729.2442 0.0000
> > >
> > > BRACKET: 7.36549e-05 1913.84 -1.29993e+07 -1.34062e+06 6.80802e+07
> > > ENERGY: 199 73237.2018 37093.2432 7534.1100 527.0121
> > > 301947.9529 230399.3378 0.0000 0.0000 0.0000
> > > 650738.8579 0.0000 650738.8579 650738.8579 0.0000
> > >
> > > LDB: LOAD: AVG 163.034 MAX 257.747 MSGS: TOTAL 240 MAXC 44 MAXP 4 None
> > > LDB: LOAD: AVG 163.034 MAX 195.64 MSGS: TOTAL 526 MAXC 145 MAXP 6 Alg7
> > > LDB: LOAD: AVG 163.034 MAX 166.29 MSGS: TOTAL 530 MAXC 145 MAXP 6 Alg7
> > > Rtasks fail:
> > > Rtask(s) 1 : exited with signal <11>
> > > Rtask(s) 2 5 4 3 9 6 8 7 10 : exited with signal <15>
> > > Rtask(s) 1 : coredump
> > > >
> > >
> > > I am using the script which is pasted below. So far, I have tried:
> > > decreasing the parameters minBabyStep and minTinyStep by 1, 2, and 3
> > > orders of magnitude; running minimization with the velocity quenching
> > > algorithm, and starting from a restart file. In all cases the system
> > > craps out after exactly 199 steps, including when the system is
> > > started from the restart file.
> > >
> > > My input files are rather large (~5MB pdb, ~10MB psf) but I can post
> > > if someone requests.
> > >
> > > I think that I am most likely missing something obvious.
> > >
> > > Thanks in advance for any help,
> > > Charles
> > >
> > > # initial config
> > > coordinates ../final.pdb
> > > structure ../final.psf
> > > temperature 0
> > >
> > > # output params
> > > outputtiming 1000
> > > outputname ./output/minimize
> > > binaryoutput no
> > >
> > > # for restart
> > > restartname ./minimize/system_minimize_restart
> > > restartfreq 100
> > > restartsave yes
> > > binaryrestart yes; # preserves more accuracy.
> > >
> > > # integrator params
> > > timestep 1.0
> > >
> > > # force field params
> > > paratypecharmm on
> > > parameters par_all27_prot_lipid.prm
> > > exclude 1-2
> > > switching on
> > > switchdist 8.0
> > > cutoff 12.0
> > > pairlistdist 13.5
> > > stepspercycle 20
> > >
> > > #velocityQuenching on
> > >
> > >
> > > #minimizethe yffstem
> > > minimization on; #this turns on fast minimization
> > > minTinyStep 1.0e-7;
> > > #minbabystep 1.0e-3; #If it doesnt work, this option is on for the next run.
> > > minimize 2000
> > >
> >
> > *****************************************************************
> > **Brian Bennion, Ph.D. **
> > **Computational and Systems Biology Division **
> > **Biology and Biotechnology Research Program **
> > **Lawrence Livermore National Laboratory **
> > **P.O. Box 808, L-448 bennion1_at_llnl.gov **
> > **7000 East Avenue phone: (925) 422-5722 **
> > **Livermore, CA 94550 fax: (925) 424-6605 **
> > *****************************************************************
> >
> >
>

*****************************************************************
**Brian Bennion, Ph.D. **
**Computational and Systems Biology Division **
**Biology and Biotechnology Research Program **
**Lawrence Livermore National Laboratory **
**P.O. Box 808, L-448 bennion1_at_llnl.gov **
**7000 East Avenue phone: (925) 422-5722 **
**Livermore, CA 94550 fax: (925) 424-6605 **
*****************************************************************

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:38:57 CST