Hi Peter, The real problem is that it's creating water molecules that are very tightly packed together in a very high energy configuration, and the simulation is blowing up. yes, this sorted it out. Phew :) I guess the issue is numerical round-off. When I find the time, I'll see if I get the same pr...

Hi Peter, I have run the Python script successfully and it reports 792450 atoms in its original setting (boxSize = 20). I tested both the CUDA platform and the OpenCL platform. Here's the complete output: Adding solvent Building system 792450 atoms Creating context Computing initial energy: -9582238...

Hi Peter, my PC has 12GB system memory and I am also running Ubuntu 12.04. I tried your script but have discovered that the Python installation does not work in my OpenMM 5.1. I'll have to sort that out first. I'm not normally using the Python scripts, I work with OpenMM C++ API directly. I'll get b...

Hi Peter, I've installed OpenMM5.1 from source and tested again, on Quadro 4000. The installation is Release with the optimisation on. This time, the water box simulation runs up to 45x45x45 water molecules (in OpenMM5.0.1 it reached 46x46x46). At 46x46x46 it fails in the middle of the simulation ru...

Hi Peter, I am running OpenMM5.0.1 at the moment and the tests with error printouts in my previous post were all done on a desktop PC (Dell Precision T5500) with 2 Quadro 4000 with 2GB memory. On Tesla, I only ran the simulation in various sizes to see where it gave up, I did not recompile and run i...

Hi Peter, I'm experiencing a similar issue. I can run simulations of up to 330,000 atoms on an NVidia Quadro 4000 with 2GB of graphics memory and only around 430,000 on Tesla M2075 with 6 GB of memory. It seems strange that the scaling is not better. I've run some tests on HelloWaterBox.cpp to estab...

I managed to resolve the problem above (segmentation fault) completely by including the "plugins" directory in the LD_LIBRARY_PATH. In my case, the segmentation fault mentioned by Socrates above was masking a different problem. It turned out that libOpenMMRPMDOpenCL.so and libOpenMMRPMDCUDA.so were ...

I've experienced the same problem on Ubuntu 12.04 LTS.

It seems to work fine if you preload the pthread library:
LD_PRELOAD=/lib/x86_64-linux-gnu/libpthread.so.0 gdb --args ./HelloArgon

Or, alternatively, run the example directly:
LD_PRELOAD=/lib/x86_64-linux-gnu/libpthread.so.0 ./HelloArgon

Spela

Hi Peter, thank you for your reply. I'm not using the barostat, so it looks like this could just do the job. > Another thing to note is that CustomHbondForce is implemented with a simple loop comparing every donor to every acceptor... Yes, the way I've set it up, the number of donors and the number ...

Hi Peter, I'd like to be able to count Hydrogen bonds between water molecules in a pure water simulation. I've seen the CustomHbondForce class which allows me to model Hydrogen bonding, but what I am interested in is not the modelling itself (I already have that done differently and have to keep it ...