MathGroup Archive 2010

[Date Index] [Thread Index] [Author Index]

Search the Archive

NDSolve with rapidly time varying boundary condition

  • To: mathgroup at smc.vnet.net
  • Subject: [mg108818] NDSolve with rapidly time varying boundary condition
  • From: Clint <clint.zeringue at kirtland.af.mil>
  • Date: Thu, 1 Apr 2010 06:02:54 -0500 (EST)

Hello,

I have a triply coupled system of PDE. I have a boundary condition for one of the equations that vary's in time ie.
F[0,t]=A*sin(omega*t);

I am currently implementing the methodofLines using:

NDSolve[eqns,{z,0,L},{t,0,tfinal},Method->{"MethodOfLines","SpatialDiscretization"->{"TensorProductGrid","PrecisionGoal"->6,"AccuracyGoal"->4,"MaxStepSize"->0.01]

This works no problem for when omega is a modest number. The problem comes in as I try to scale omega to larger and larger values because I receive spatial discretization errors. I can fix the problem by changing "MaxStepSize" to smaller numbers ie 0.001 etc. However, as I do that my spatial grid is becoming quite large and the MaxStepSize I need to solve for a quite large value of omega is so small that I get a wierd message pop up that does not even seem to be a mathematica message. 

It is a windows popup that says "An unhandled win32 exception occurred in MathKernael.exe[1120]. Since I have visual studio installed on my machine it asks me if I wish to Dubug using the debugger in visual studio. 

I do not mind that the simulation time requires a long time to run with a finer spatial grid, but looking at my processors the CPU usage goes to zero after this message is generated and if I click no that i don't want to debug the mathematica kernal shuts down.

Can anyone out there give me some possible insight?

Thanks,


  • Prev by Date: scaling a plot via imagesize
  • Next by Date: Re: a little nasty problem?
  • Previous by thread: Re: scaling a plot via imagesize
  • Next by thread: Re: a little nasty problem?