Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Nanoparticle monte carlo

  1. May 31, 2006 #1
    Hello guys,
    I am trying to do monte carlo modelling for a system of magnetic nanoparticles that my lab is studying (optinally in external magnetic fields). I played around with this last summer and wrote some simple monte carlo code in csharp (i wanted to learn it). Now ... its time for this to get more serious. I want something that will be cross platform. Is it reasonable to use Java for a task like this? Does anyone have any suggestions on where I could go for help on this issue?
  2. jcsd
  3. Jun 2, 2006 #2
    Though, this is not a gr8 help...

    go thru this paper recently published on the topic..

    http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=JAPIAU000099000010104306000001&idtype=cvips&gifs=Yes&type=ALERT [Broken]
    Last edited by a moderator: May 2, 2017
  4. Jun 4, 2006 #3


    User Avatar
    Science Advisor
    Gold Member

    Have seen some pretty efficient codes (not Monte Carlo but anyways) done with Java so at least to some point why not. I'm left thinking what all parameters are you going to sample in your analysis -> how heavy the solution will turn out (thinking whether standard MC would do it or might it be preferable to use something else like LHS, or what might consider best applicable for your problem)?
  5. Jun 7, 2006 #4
    having implemented a hybrid monte carlo package, im not so sure that you want to go with java - it really depends upon what you need to do. in my case i have a small periodic system, but there are MD steps and also a very intensive energy calculation that scales as ~9N^2 in addition to the MD. for me, MPI or a highly SMP machine is a must. i found C to be appropriate, but C++ would have worked just as well.

    what is the performance of java code these days?

    you may have trouble running a java code on a cray of MPI cluster - then again, i'm not a java coder so maybe it's not a big deal. do you need parallel computing? how intense are your calculations? how do they scale, and do you plan on ever running large system sizes? could someone wind up taking over your project at some time, and find themselves screwed by a java codebase in ways that weren't anticipated? these are things that i would consider.

    my $0.02
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook