[gmx-developers] Regression since 5b88d3c18 Remove state duplication in serial runs

[Vedran Miletić]

Hello Berk, others,

I observed a regression which results in mdrun NVT or production MD 
crashing under certain conditions and bisected it to 5b88d3c18 Remove 
state duplication in serial runs. The minimal example is that causes 
this is ACACACA peptide (easy to generate with e.g. PyMOL) running EM 
and MD with Justin's mdp files [1, 2] using the following sequence of 
commands:

gmx pdb2gmx -f acacaca.pdb -ignh
gmx editconf -f conf.gro -o conf_newbox.gro -c -d 1.0 -bt cubic
gmx solvate -cp conf_newbox.gro -cs spc216.gro -o conf_solv.gro -p topol.top
gmx grompp -f minim.mdp -c conf_solv.gro -p topol.top -o em.tpr
gmx mdrun -deffnm em -v
gmx grompp -f md.mdp -c em.gro -p topol.top -o md.tpr
gmx mdrun -deffnm md -v

Prior to 5b88d3c18 this will run fine, but 5b88d3c18 and later revisions 
result in

step 0
Step 1, time 0.002 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
rms 113.755112, max 983.434143 (between atoms 38 and 39)
bonds that rotated more than 30 degrees:
  atom 1 atom 2  angle  previous, current, constraint length
      22     24   48.4    0.1336   0.2057      0.1335
      24     25   50.6    0.1010   0.1576      0.1010
      24     26   87.9    0.1450   0.3411      0.1449
      26     27   85.5    0.1090   0.2146      0.1090
      26     28   76.3    0.1526   0.2979      0.1526
      26     32   80.4    0.1524   0.4376      0.1522
      28     29   50.7    0.1089   0.1687      0.1090
      28     30   54.4    0.1094   0.1528      0.1090
      28     31   52.0    0.1089   0.1709      0.1090
      32     33   82.8    0.1230   0.2838      0.1229
      32     34  149.8    0.1337   4.0620      0.1335
      34     35  104.7    0.1011   4.0236      0.1010
      36     37   87.6    0.1091   0.1962      0.1090
      36     38   94.4    0.1524   0.5549      0.1526
      36     43   91.1    0.1524   0.6512      0.1522
      38     39  122.4    0.1992 107.3033      0.1090
      38     40   91.5    0.1085   0.1488      0.1090
      38     41   73.0    0.1806   0.0914      0.1810
      43     44   33.7    0.1230   0.0828      0.1229
Segmentation fault

Since this particular revision did not change mdp settings, I presume 
MDP files from Justin's tutorial are fine.

Regards,
Vedran

[1] 
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/lysozyme/Files/minim.mdp
[2] 
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/lysozyme/Files/md.mdp

-- 
Vedran Miletić
vedran.miletic.net