| Author | Message | ||||
     Jared Barber (jared_barber) Member Username: jared_barber Post Number: 27 Registered: 01-2007 |
Hello, I was trying to begin using the moving mesh for a particular problem. I have posted this problem. Whenever I run this problem, it quickly reaches the point where dt is too small. I tend to think that flexpde is having problems dealing with the fact that no variables, except the actual mesh variables, have any explicit time dependence. For instance, there should be no necessity for initial conditions. (Nonetheless, I have fed in a very good transfer file "guess" which gives the initial solution to my equations.) Unfortunately I haven't had the chance to study the samples on moving_meshes but none of them, just from my quick glances, seem to do what I'm trying to do. Either way, I was hoping for some help on what to do to make this work. Thanks for any help, Jared Actual code
"Initial conditions"
| ||||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 1051 Registered: 06-2003 |
Are you sure the initial velocity values are consistent with the boundary conditions on velocity? I set the initial timestep to 1e-6, and there appears to be a serious disagreement about velocities on the boundary. Also, the conjugate gradient iteration seems not to be converging. There is something that makes the matrix singular. | ||||
| Jared Barber (jared_barber) Member Username: jared_barber Post Number: 28 Registered: 01-2007 |
I'm fairly certain the initial values are consistent with the boundary conditions. I just checked that. I will check it again but for now can I just ask, why doesn't flexpde realize it doesn't need initial velocity values for this problem? That is, why doesn't it decouple the two systems and solve them separately? I'll try to send you plots/monitors to verify that my initial conditions are indeed consistent with the boundary conditions but I"ll have to get to that later. Thanks | ||||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 1052 Registered: 06-2003 |
There is an erratic error in the assignment of Velocity boundary conditions in versions 5.0.19 and earlier. Try version 5.0.20x3 from the www.pdesolutions.com/buglist5.html page |
