1. General topics
  3. Other
  5. Failed to run simulation

normal Failed to run simulation

  • lihie
  • lihie's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
12 years 4 months ago #875 by lihie
Failed to run simulation was created by lihie
Hello all,

I'm new in CG simulation, after running successfully em, the simulation crashes in pr. The system is a protein in a water box. The topology files were created using the martinize.py script. The system was properly minimized but now I'm, receiving the following error while running it on 8 nodes: "Domain decomposition has not been implemented for box vectors that have non-zero components in directions that do not use domain decomposition: ncells = 1 2 4, box vector[2] = -nan -nan -nan"
And the following error while running it on 1 node: "Number of grid cells is zero. Probably the system and box collapsed."

I tried to run a new simulation with a different protein, and it all went well, so I guess I'm doing it ok.

If someone can help me I'll be more then thankful,

Lihie

Please Log in or Create an account to join the conversation.

More
12 years 4 months ago #876 by xavier
Replied by xavier on topic Failed to run simulation
This indicates that your gromacs is having troubles cutting your system into pieces that fit the 8 CPUS. This is a problem with dynamic load balancing ... you could try to use the -pd (for particle decomposition option) or use the -rdd 1.5 option. Note that the latter option might not solve the problem ... as a protein in water is a small system and it might not find a good decomposition of the system.

lihie wrote: Hello all,

I'm new in CG simulation, after running successfully em, the simulation crashes in pr. The system is a protein in a water box. The topology files were created using the martinize.py script. The system was properly minimized but now I'm, receiving the following error while running it on 8 nodes: "Domain decomposition has not been implemented for box vectors that have non-zero components in directions that do not use domain decomposition: ncells = 1 2 4, box vector[2] = -nan -nan -nan"
And the following error while running it on 1 node: "Number of grid cells is zero. Probably the system and box collapsed."

I tried to run a new simulation with a different protein, and it all went well, so I guess I'm doing it ok.

If someone can help me I'll be more then thankful,

Lihie

Please Log in or Create an account to join the conversation.

More
12 years 3 months ago #883 by Clement
Replied by Clement on topic Failed to run simulation
"Number of grid cells is zero. Probably the system and box collapsed." means you're trying to simulate a box with dimension 0. EM does *not* change the dimensions of your box if initially set to 0 (it changes it to -NaN). Check the last line of the GRO file or the header of the PDB you're using...

Run a:
editconf -f <your GRO|PDB> -c -d 0 -o <new GRO|PDB>
It will add dimensions to the box by looking at the coordinates of the particles in your system. Then regenerate your TPR using this new file; this would fix the problem.

Please Log in or Create an account to join the conversation.

  • lihie
  • lihie's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
12 years 3 months ago #888 by lihie
Replied by lihie on topic Failed to run simulation
Hello Clement,

Thank you for replying. The box dimension appears in the gro file so this is not the problem. I read in the gromacs manual on grid and the way it calculates the atoms in the system and with xavier suggestion I tried to build different sizes of the box, and finally it worked.

Thank's again,

Lihie

Please Log in or Create an account to join the conversation.

  1. General topics
  3. Other
  5. Failed to run simulation
Time to create page: 0.093 seconds