| Author | Message | |||
     pepper New member Username: pepper Post Number: 1 Registered: 09-2010 |
holle Can I plot the contour of "vector ((K*(-dx(h))),(K*(-dy(h))))" ?? Thank you | |||
| rgnelson Moderator Username: rgnelson Post Number: 1411 Registered: 06-2003 |
You can't plot a contour of a vector. A contour displays the levels of a scalar quantity. A vector is a directed line segment. You could plot a contour of the magnitude of a vector, or you could plot a contour of a component of a vector, or you could plot two contours of the components of a vector, or you could plot an arrow plot of the vectors. The last of these is simply: VECTOR(-k*dx(h), -k*dy(h)) or VECTOR(-k*grad(h)) The others are CONTOUR(magnitude(k*grad(h))) CONTOUR(-k*dx(h)) CONTOUR(-k*dy(h)) CONTOUR(-k*dx(h)),-k*dy(h)) etc. | |||
| pepper New member Username: pepper Post Number: 2 Registered: 09-2010 |
hello Can I plot the streamline of "div(k*grad(h))=0" or "dx(k*dx(h))+dy(k*dy(h))=0" Thank you | |||
| rgnelson Moderator Username: rgnelson Post Number: 1412 Registered: 06-2003 |
There is no built-in "Streamline" facility, but you can get the plot you want using an adjoint equation. See "fieldmap.pde" in the "Samples|Applications|Electricity" folder in version 6, or the "Samples|Steady_State|Electricity" folder in version 5. | |||
| pepper Junior Member Username: pepper Post Number: 3 Registered: 09-2010 |
May I know what is meaning(use) of "tangential(grad(v))" in field.pde. I don't seem to be able to find it in the manual. I have the folloeing equestions, this see the doc.file
| |||
| rgnelson Moderator Username: rgnelson Post Number: 1413 Registered: 06-2003 |
1) In our example Fieldmap.pde, we plot field lines by use of an adjoint equation. The contours of the adjoint solution are everywhere perpendicular to the contours of the solution. The Natural() bc specifies the normal component of the gradient. So the normal component of the adjoint gradient is (parallel to) the tangential component of the solution gradient. 2) The difference in the plots is simply in the choice of contour levels. Apparently the two-field contour plotter is selecting a range that includes zero, whereas the single field plot merely brackets the actual values. We will look into this. | |||
| 4maitland Member Username: 4maitland Post Number: 10 Registered: 05-2010 |
Adjoint Equation? are you referring to the conjugate-gradient method? | |||
| mgnelson Moderator Username: mgnelson Post Number: 209 Registered: 07-2007 |
The conjugate gradient method is an algorithm for solving a system of equations. The adjoint equation is a pde posed by the user in a script. You can find some basic information about these on Wikipedia. | |||
| pepper Member Username: pepper Post Number: 5 Registered: 09-2010 |
When I use "natural(q)=tangential(grad(v))" in my boundary condition, can I add timestep in my case?? I add timestep in my case is error.... just like doc.file
| |||
| mgnelson Moderator Username: mgnelson Post Number: 210 Registered: 07-2007 |
Your attachment has a graphic of the script text, so we cannot run it. Please attach the script file (*.pde) if you want us to look at it. I notice that you have added a TIME section to your script, but the problem is not time dependent. The set of equations are steady-state. In this scenario, I would expect FlexPDE to solve the steady-state problem over and over (unnecessarily) at each time step until reaching the end time you requested. I don't know why it would halt due to the time step falling below 1e-9 of the starting time step (this is the error report shown in your doc.) But again, we cannot try the script because it is a graphic, not text. | |||
| pepper Member Username: pepper Post Number: 6 Registered: 09-2010 |
this is my script file. In my case, the contour(q) is representative stream line ??
| |||
| mgnelson Moderator Username: mgnelson Post Number: 212 Registered: 07-2007 |
I cannot identify any reason why FlexPDE has cut the time step so drastically. As I said, I would expect FlexPDE to solve the steady state system at every time step. However, that does not change the fact that you do not have a time dependent problem. There is no reason to ask FlexPDE to solve this sytem in a time dependent fashion. Just remove the TIME section (and the time requests for plots) and the script runs and gives an answer very quickly. See attached.
|
field lines